Open MHA Developer Manual

User Manual: Pdf

Open the PDF directly: View PDF .
Page Count: 328 [warning: Documents this large are best viewed by clicking the View PDF Link!]

Overview
- Structure
- Platform Services and Conventions
Module Documentation
Namespace Documentation
Class Documentation
File Documentation
Index

The Open Master Hearing Aid

(openMHA)

4.5.5

Plugin Developers’ Manual

The Open Master Hearing Aid (openMHA) – Plugin Developers’ Manual

HörTech gGmbH

Marie-Curie-Str. 2

D–26129 Oldenburg

iii

LICENSE AGREEMENT This ﬁle is part of the HörTech Open Master Hearing Aid (openMHA)

gGmbH.

openMHA is free software: you can redistribute it and/or modify it under the terms of the GNU

Affero General Public License as published by the Free Software Foundation, version 3 of the

License.

openMHA is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY;

without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR

PURPOSE. See the GNU Affero General Public License, version 3 for more details.

You should have received a copy of the GNU Affero General Public License, version 3 along

with openMHA. If not, see <http://www.gnu.org/licenses/>.

Contents

1 Overview 1

1.1 Structure ........................................ 1

1.2 Platform Services and Conventions ......................... 2

2 Module Documentation 4

2.1 Concept of Variables and Data Exchange in the openMHA ............ 4

2.2 The openMHA Plugins (programming interface) .................. 6

2.3 Writing openMHA Plugins. A step-by-step tutorial ................. 10

2.4 The MHA Framework interface ............................ 26

2.5 Communication between algorithms ......................... 27

2.6 Error handling in the openMHA ........................... 31

2.7 The openMHA conﬁguration language ....................... 33

2.8 The openMHA Toolbox library ............................ 34

2.9 Vector and matrix processing toolbox ........................ 36

2.10 Complex arithmetics in the openMHA ........................ 53

2.11 Fast Fourier Transform functions ........................... 58

3 Namespace Documentation 64

3.1 AuditoryProﬁle Namespace Reference ....................... 64

3.2 DynComp Namespace Reference .......................... 64

3.3 MHA_AC Namespace Reference .......................... 66

3.4 MHA_TCP Namespace Reference ......................... 66

3.5 MHAEvents Namespace Reference ......................... 67

3.6 MHAFilter Namespace Reference .......................... 68

3.7 MHAIOJack Namespace Reference ......................... 71

3.8 MHAJack Namespace Reference .......................... 71

3.9 MHAMultiSrc Namespace Reference ........................ 73

3.10 MHAOvlFilter Namespace Reference ........................ 73

3.11 MHAOvlFilter::FreqScaleFun Namespace Reference ............... 74

3.12 MHAOvlFilter::ShapeFun Namespace Reference ................. 75

3.13 MHAParser Namespace Reference ......................... 76

3.14 MHAParser::StrCnv Namespace Reference .................... 79

3.15 MHAPlugin Namespace Reference ......................... 80

3.16 MHASignal Namespace Reference ......................... 81

3.17 MHATableLookup Namespace Reference ..................... 86

3.18 MHAWindow Namespace Reference ........................ 87

4 Class Documentation 88

4.1 algo_comm_t Struct Reference ........................... 88

4.2 AuditoryProﬁle::fmap_t Class Reference ...................... 93

4.3 AuditoryProﬁle::parser_t Class Reference ..................... 94

4.4 AuditoryProﬁle::proﬁle_t Class Reference ...................... 94

4.5 AuditoryProﬁle::proﬁle_t::ear_t Class Reference .................. 95

4.6 comm_var_t Struct Reference ............................ 95

4.7 DynComp::dc_afterburn_rt_t Class Reference ................... 96

4.8 DynComp::dc_afterburn_t Class Reference ..................... 97

4.9 DynComp::dc_afterburn_vars_t Class Reference ................. 97

4.10 DynComp::gaintable_t Class Reference ....................... 98

4.11 expression_t Class Reference ............................100

4.12 io_ﬁle_t Class Reference ...............................100

CONTENTS v

4.13 io_lib_t Class Reference ...............................101

4.14 io_parser_t Class Reference .............................102

4.15 io_tcp_fwcb_t Class Reference ...........................103

4.16 io_tcp_parser_t Class Reference ..........................105

4.17 io_tcp_sound_t Class Reference ...........................110

4.18 io_tcp_sound_t::ﬂoat_union Union Reference ...................113

4.19 io_tcp_t Class Reference ...............................113

4.20 MHA_AC::ac2matrix_t Class Reference .......................114

4.21 MHA_AC::acspace2matrix_t Class Reference ...................116

4.22 MHA_AC::double_t Class Reference ........................118

4.23 MHA_AC::ﬂoat_t Class Reference ..........................118

4.24 MHA_AC::int_t Class Reference ...........................119

4.25 MHA_AC::spectrum_t Class Reference .......................119

4.26 MHA_AC::waveform_t Class Reference .......................120

4.27 mha_audio_descriptor_t Struct Reference .....................122

4.28 mha_audio_t Struct Reference ............................122

4.29 mha_channel_info_t Struct Reference ........................123

4.30 mha_complex_t Struct Reference ..........................123

4.31 mha_dblbuf_t<FIFO >Class Template Reference ................124

4.32 mha_direction_t Struct Reference ..........................127

4.33 mha_drifter_ﬁfo_t<T>Class Template Reference ................127

4.34 MHA_Error Class Reference .............................132

4.35 mha_ﬁfo_lw_t<T>Class Template Reference ..................132

4.36 mha_ﬁfo_t<T>Class Template Reference ....................135

4.37 mha_ﬁfo_thread_guard_t Class Reference .....................137

4.38 mha_ﬁfo_thread_platform_t Class Reference ....................137

4.39 mha_rt_ﬁfo_element_t<T>Class Template Reference .............139

4.40 mha_rt_ﬁfo_t<T>Class Template Reference ...................140

4.41 mha_spec_t Struct Reference ............................141

4.42 MHA_TCP::Async_Notify Class Reference .....................143

4.43 MHA_TCP::Client Class Reference .........................143

4.44 MHA_TCP::Connection Class Reference ......................144

4.45 MHA_TCP::Event_Watcher Class Reference ....................149

4.46 MHA_TCP::Sockread_Event Class Reference ...................150

4.47 MHA_TCP::Thread Class Reference ........................150

4.48 MHA_TCP::Timeout_Watcher Class Reference ..................152

4.49 MHA_TCP::Wakeup_Event Class Reference ....................153

4.50 mha_wave_t Struct Reference ............................154

4.51 mhaconﬁg_t Struct Reference ............................155

4.52 MHAEvents::emitter_t Class Reference .......................156

4.53 MHAEvents::patchbay_t<receiver_t >Class Template Reference ........156

4.54 MHAFilter::adapt_ﬁlter_t Class Reference .....................157

4.55 MHAFilter::blockprocessing_polyphase_resampling_t Class Reference . . . . . 157

4.56 MHAFilter::complex_bandpass_t Class Reference .................159

4.57 MHAFilter::diff_t Class Reference ..........................160

4.58 MHAFilter::fftﬁlter_t Class Reference ........................160

4.59 MHAFilter::fftﬁlterbank_t Class Reference .....................163

4.60 MHAFilter::ﬁlter_t Class Reference .........................165

4.61 MHAFilter::gamma_ﬂt_t Class Reference ......................167

4.62 MHAFilter::iir_ﬁlter_t Class Reference ........................168

4.63 MHAFilter::iir_ord1_real_t Class Reference .....................170

vi CONTENTS

4.64 MHAFilter::o1_ar_ﬁlter_t Class Reference .....................171

4.65 MHAFilter::o1ﬂt_lowpass_t Class Reference ....................173

4.66 MHAFilter::o1ﬂt_maxtrack_t Class Reference ...................174

4.67 MHAFilter::o1ﬂt_mintrack_t Class Reference ....................176

4.68 MHAFilter::partitioned_convolution_t Class Reference ...............177

4.69 MHAFilter::partitioned_convolution_t::index_t Struct Reference ..........179

4.70 MHAFilter::polyphase_resampling_t Class Reference ...............180

4.71 MHAFilter::resampling_ﬁlter_t Class Reference ..................182

4.72 MHAFilter::smoothspec_t Class Reference .....................183

4.73 MHAFilter::transfer_function_t Struct Reference ..................185

4.74 MHAFilter::transfer_matrix_t Struct Reference ...................187

4.75 MHAIOJack::io_jack_t Class Reference .......................188

4.76 MHAJack::client_avg_t Class Reference ......................189

4.77 MHAJack::client_noncont_t Class Reference ....................190

4.78 MHAJack::client_t Class Reference .........................191

4.79 MHAJack::port_t Class Reference ..........................193

4.80 MHAMultiSrc::base_t Class Reference .......................195

4.81 MHAOvlFilter::fftfb_t Class Reference ........................196

4.82 MHAOvlFilter::fftfb_vars_t Class Reference .....................197

4.83 MHAOvlFilter::fspacing_t Class Reference .....................198

4.84 MHAOvlFilter::overlap_save_ﬁlterbank_t Class Reference ............199

4.85 MHAParser::base_t Class Reference ........................200

4.86 MHAParser::bool_mon_t Class Reference .....................203

4.87 MHAParser::bool_t Class Reference ........................204

4.88 MHAParser::commit_t<receiver_t >Class Template Reference .........205

4.89 MHAParser::complex_mon_t Class Reference ...................206

4.90 MHAParser::complex_t Class Reference ......................207

4.91 MHAParser::ﬂoat_mon_t Class Reference .....................207

4.92 MHAParser::ﬂoat_t Class Reference ........................208

4.93 MHAParser::int_mon_t Class Reference ......................210

4.94 MHAParser::int_t Class Reference ..........................211

4.95 MHAParser::keyword_list_t Class Reference ....................212

4.96 MHAParser::kw_t Class Reference .........................214

4.97 MHAParser::mcomplex_mon_t Class Reference ..................215

4.98 MHAParser::mcomplex_t Class Reference .....................216

4.99 MHAParser::mﬂoat_mon_t Class Reference ....................217

4.100 MHAParser::mﬂoat_t Class Reference .......................218

4.101 MHAParser::mhapluginloader_t Class Reference .................220

4.102 MHAParser::monitor_t Class Reference .......................220

4.103 MHAParser::parser_t Class Reference .......................220

4.104 MHAParser::range_var_t Class Reference .....................223

4.105 MHAParser::string_mon_t Class Reference ....................224

4.106 MHAParser::string_t Class Reference ........................225

4.107 MHAParser::variable_t Class Reference ......................226

4.108 MHAParser::vcomplex_mon_t Class Reference ..................226

4.109 MHAParser::vcomplex_t Class Reference .....................227

4.110 MHAParser::vﬂoat_mon_t Class Reference .....................228

4.111 MHAParser::vﬂoat_t Class Reference ........................229

4.112 MHAParser::vint_mon_t Class Reference ......................231

4.113 MHAParser::vint_t Class Reference .........................232

4.114 MHAParser::vstring_mon_t Class Reference ....................233

CONTENTS vii

4.115 MHAParser::vstring_t Class Reference .......................234

4.116 MHAParser::window_t Class Reference .......................234

4.117 MHAPlugin::conﬁg_t<runtime_cfg_t >Class Template Reference ........236

4.118 MHAPlugin::plugin_t<runtime_cfg_t >Class Template Reference ........238

4.119 mhaserver_t Class Reference ............................241

4.120 MHASignal::async_rmslevel_t Class Reference ..................242

4.121 MHASignal::delay_t Class Reference ........................243

4.122 MHASignal::delay_wave_t Class Reference ....................244

4.123 MHASignal::doublebuffer_t Class Reference ....................244

4.124 MHASignal::hilbert_t Class Reference .......................246

4.125 MHASignal::loop_wavefragment_t Class Reference ................247

4.126 MHASignal::matrix_t Class Reference ........................249

4.127 MHASignal::minphase_t Class Reference .....................255

4.128 MHASignal::quantizer_t Class Reference ......................256

4.129 MHASignal::ringbuffer_t Class Reference ......................257

4.130 MHASignal::schroeder_t Class Reference .....................259

4.131 MHASignal::spectrum_t Class Reference ......................261

4.132 MHASignal::subsample_delay_t Class Reference .................265

4.133 MHASignal::uint_vector_t Class Reference .....................266

4.134 MHASignal::waveform_t Class Reference ......................268

4.135 MHATableLookup::xy_table_t Class Reference ...................275

4.136 MHAWindow::bartlett_t Class Reference ......................277

4.137 MHAWindow::base_t Class Reference .......................278

4.138 MHAWindow::blackman_t Class Reference .....................279

4.139 MHAWindow::fun_t Class Reference ........................280

4.140 MHAWindow::hamming_t Class Reference .....................281

4.141 MHAWindow::hanning_t Class Reference .....................282

4.142 MHAWindow::rect_t Class Reference ........................283

4.143 MHAWindow::user_t Class Reference ........................284

4.144 PluginLoader::fourway_processor_t Class Reference ...............285

5 File Documentation 287

5.1 mha.h File Reference .................................287

5.2 mha_algo_comm.h File Reference .........................289

5.3 mha_defs.h File Reference ..............................290

5.4 mha_error.cpp File Reference ............................290

5.5 mha_ﬁlter.hh File Reference .............................291

5.6 mha_parser.hh File Reference ............................293

5.7 mha_plugin.hh File Reference ............................296

5.8 mha_signal.hh File Reference ............................297

5.9 mha_tablelookup.hh File Reference .........................306

Index 307

1 Overview 1

1 Overview

The HörTech Open Master Hearing Aid (openMHA), is a development and evaluation software

platform that is able to execute hearing aid signal processing in real-time on standard computing

hardware with a low delay between sound input and output.

1.1 Structure

The openMHA can be split into four major components :

•The openMHA command line application (MHA) (p. 33)

•Signal processing plugins (p. 6)

• Audio input-output (IO) plugins (see io_ﬁle_t (p. 100), MHAIOJack (p. 71), io_parser_t

(p. 102), io_tcp_parser_t (p. 105))

•The openMHA toolbox library (p. 34)

openMHA

plugins IO

audio backend

(Jack, File, TCP)

MHAlibopenmha

control applications

(e.g., Octave)

Figure 1 openMHA structure

The openMHA command line application (MHA) (p. 33) acts as a plugin host. It can load

signal processing plugins as well as audio input-output (IO) plugins. Additionally, it provides

the command line conﬁguration interface and a TCP/IP based conﬁguration interface. Several

IO plugins exist: For real-time signal processing, commonly the openMHA MHAIOJack (p. 71)

plugin (see plugins' manual) is used, which provides an interface to the Jack Audio Connection

Kit (JACK). Other IO plugins provide audio ﬁle access or TCP/IP-based processing.

openMHA plugins (p. 6) provide the audio signal processing capabilities and audio signal han-

dling. Typically, one openMHA plugin implements one speciﬁc algorithm. The complete virtual

hearing aid signal processing can be achieved by a combination of several openMHA plugins.

2 CONTENTS

1.2 Platform Services and Conventions

The openMHA platform offers some services and conventions to algorithms implemented in

plugins, that make it especially well suited to develop hearing aid algorithms, while still support-

ing general-purpose signal processing.

1.2.1 Audio Signal Domains

As in most other plugin hosts, the audio signal in the openMHA is processed in audio chunks.

However, plugins are not restricted to propagate audio signal as blocks of audio samples in the

time domain another option is to propagate the audio signal in the short time Fourier transform

(STFT) domain, i.e. as spectra of blocks of audio signal, so that not every plugin has to perform

its own STFT analysis and synthesis. Since STFT analysis and re-synthesis of acceptable

audio quality always introduces an algorithmic delay, sharing STFT data is a necessity for a

hearing aid signal processing platform, because the overall delay of the complete processing

has to be as short as possible.

Similar to some other platforms, the openMHA allows also arbitrary data to be exchanged

between plugins through a mechanism called algorithm communication variables (p. 27) or

short "AC vars". This mechanism is commonly used to share data such as ﬁlter coefﬁcients or

ﬁlter states.

1.2.2 Real-Time Safe Complex Conﬁguration Changes

Hearing aid algorithms in the openMHA can export conﬁguration settings that may be changed

by the user at run time.

To ensure real-time safe signal processing, the audio processing will normally be done in a

signal processing thread with real-time priority, while user interaction with conﬁguration pa-

rameters would be performed in a conﬁguration thread with normal priority, so that the audio

processing does not get interrupted by conﬁguration tasks. Two types of problems may occur

when the user is changing parameters in such a setup:

• The change of a simple parameter exposed to the user may cause an involved recalcu-

lation of internal runtime parameters that the algorithm actually uses in processing. The

duration required to perform this recalculation may be a signiﬁcant portion of (or take even

longer than) the time available to process one block of audio signal. In hearing aid usage,

it is not acceptable to halt audio processing for the duration that the recalculation may

require.

• If the user needs to change multiple parameters to reach a desired conﬁguration state

of an algorithm from the original conﬁguration state, then it may not be acceptable that

processing is performed while some of the parameters have already been changed while

others still retain their original values. It is also not acceptable to interrupt signal process-

ing until all pending conﬁguration changes have been performed.

1.2 Platform Services and Conventions 3

The openMHA provides a mechanism in its toolbox library to enable real-time safe conﬁguration

changes in openMHA plugins:

Basically, existing runtime conﬁgurations are used in the processing thread until the work of

creating an updated runtime conﬁguration has been completed in the conﬁguration thread.

In hearing aids, it is more acceptable to continue to use an outdated conﬁguration for a few

more milliseconds than blocking all processing.

The openMHA toolbox library provides an easy-to-use mechanism to integrate real-time safe

runtime conﬁguration updates into every plugin.

1.2.3 Plugins can Themselves Host Other Plugins

An openMHA plugin can itself act as a plugin host. This allows to combine analysis and re-

synthesis methods in a single plugin. We call plugins that can themselves load other plugins

“bridge plugins'' in the openMHA.

When such a bridge plugin is then called by the openMHA to process one block of signal, it

will ﬁrst perform its analysis, then invoke (as a function call) the signal processing in the loaded

plugin to process the block of signal in the analysis domain, wait to receive a processed block of

signal in the analysis domain back from the loaded plugin when the signal processing function

call to that plugin returns, then perform the re-synthesis transform, and ﬁnally return the block

of processed signal in the original domain back to the caller of the bridge plugin.

1.2.4 Central Calibration

The purpose of hearing aid signal processing is to enhance the sound for hearing impaired

listeners. Hearing impairment generally means that people suffering from it have increased

hearing thresholds, i.e. soft sounds that are audible for normal hearing listeners may be im-

perceptible for hearing impaired listeners. To provide accurate signal enhancement for hearing

impaired people, hearing aid signal processing algorithms have to be able to determine the

absolute physical sound pressure level corresponding to a digital signal given to any openM←-

HA plugin for processing. Inside the openMHA, we achieve this with the following convention:

The single-precision ﬂoating point time-domain sound signal samples, that are processed in-

side the openMHA plugins in blocks of short durations, have the physical pressure unit Pascal

(1Pa = 1N/m2). With this convention in place, all plugins can determine the absolute phys-

ical sound pressure level from the sound samples that they process. A derived convention is

employed in the spectral domain for STFT signals. Due to the dependency of the calibration

on the hardware used, it is the responsibility of the user of the openMHA to perform calibration

measurements and adapt the openMHA settings to make sure that this calibration convention is

met. We provide the plugin transducers which can be conﬁgured to perform the necessary

signal adjustments.

4 CONTENTS

2 Module Documentation

2.1 Concept of Variables and Data Exchange in the openMHA

Accessibility of conﬁguration variables and data exchange between plugins (processing blocks)

are an important issue in the openMHA.

In general, variable types in the openMHA are distinguished by their different access methods.

The variable types in the openMHA are:

•Conﬁguration variables : Read and write accesses are possible through the openM←-

HA conﬁguration language interface. Conﬁguration variables are implemented as C++

classes with a public data member of the underlying C type. Conﬁguration variables

can be read and modiﬁed from “outside'' using the conﬁguration language. The plugin

which provides the conﬁguration variable can use the exposed data member directly. All

accesses through the openMHA conﬁguration language are checked for data type, valid

range, and access restrictions.

•Monitor variables : Read access is possible through the openMHA conﬁguration lan-

guage. Write access is only possible from the C++ code. Internally, monitor variables

have a similar C++ class interface as conﬁguration variables.

•AC variables (algorithm communication variables (p. 27)): Any C or C++ data struc-

ture can be shared within an openMHA chain. Access management and name space

is realised in openMHA chain plugin ('mhachain'). AC variables are not available to the

openMHA conﬁguration language interface, although a read-only converter plugin acmon

is available.

•Runtime conﬁguration : Algorithms usually derive more parameters (runtime conﬁgura-

tion) from the openMHA conﬁguration language variables. When a conﬁguration variable

changes through conﬁguration language write access, then the runtime conﬁguration has

to be recomputed. Plugin developers are encouraged to encapsulate the runtime con-

ﬁguration in a C++ class, which recomputes the runtime conﬁguration from conﬁguration

variables in the constructor. The openMHA supports lock-free and thread-safe replace-

ment of the runtime conﬁguration instance (see example5.cpp (p. 20) and references

therein).

Plugin specific Chain specific

variables

monitors

runtime configuation

(e.g. C++ class)

algorithm communication

(AC variables)

Configuration language

Figure 2 Variable types in the openMHA

The C++ data types are shown in the ﬁgure below. These variables can be accessed via the

openMHA host application using the openMHA conﬁguration language. For more details see

'Application engineers' manual'.

2.1 Concept of Variables and Data Exchange in the openMHA 5

MHAParser elements

Monitors Variables parser_t

kw_t

string_t

vstring_t

bool_t

int_t

float_t

vfloat_t

mfloat_t

Numeric variables

string_mon_t

vstring_mon_t

int_mon_t

float_mon_t

vfloat_mon_t

vint_tvint_mon_t

Numeric monitors

complex_t

vcompelx_t

complex_mon_t

vcomplex_mon_t

mfloat_mon_t

Figure 3 MHAParser elements

6 CONTENTS

2.2 The openMHA Plugins (programming interface)

An openMHA plugin is the signal processing unit, usually an algorithm.

Classes

• class MHAPlugin::plugin_t<runtime_cfg_t >

The template class for C++ openMHA plugins.

Macros

• #deﬁne MHAPLUGIN_CALLBACKS_PREFIX(preﬁx, classname, indom, outdom)

C++ wrapper macro for the plugin interface.

• #deﬁne MHAPLUGIN_CALLBACKS(plugname, classname, indom, outdom) MHAPLU←-

GIN_CALLBACKS_PREFIX(MHA_STATIC_ ## plugname ## _,classname,indom,outdom)

C++ wrapper macro for the plugin interface.

• #deﬁne MHAPLUGIN_DOCUMENTATION(plugname, cat, doc) MHAPLUGIN_DOCU←-

MENTATION_PREFIX(MHA_STATIC_ ## plugname ## _,cat,doc)

Wrapper macro for the plugin documentation interface.

2.2.1 Detailed Description

openMHA plugins can be combined into processing chains. One of the conﬁgured chains can

be selected for output which allows direct comparison of single algorithms or complex signal

processing conﬁgurations. Algorithms within one chain can communicate with each other by

sharing some of their variables, see section Communication between algorithms (p. 27).

The openMHA plugins can use the openMHA conﬁguration language for their conﬁguration.

If they do so, the conﬁguration can be changed through the framework even at run time. A

description of this language can be found in section The openMHA conﬁguration language

(p. 33). If the algorithms should make use of the openMHA conﬁguration language, they need

to be written in C++ rather than pure C.

In the openMHA package a set of example plugins is included. These examples are the base

of a step by step tutorial on how to write an openMHA plugin. See section Writing openMHA

Plugins. A step-by-step tutorial (p. 10) for detailes.

openMHA plugins communicate with the openMHA using a simple ANSI-C interface. This

way it is easy to mix plugins compiled with different C++ compilers. For convenience, we

provide C++ classes which can be connected to the C++ interface. We strongly recommend

the usage of these C++ wrappers. They include out-of-the box support exporting variables to

the conﬁguration interface and for thread safe conﬁguration update.

The openMHA C++ plugin interface consists of a few number of method prototypes:

The output domain (spectrum or waveform) of an openMHA plugin will typically be the same as

the input domain:

2.2 The openMHA Plugins (programming interface) 7

•mha_wave_t (p. 154)∗process(mha_wave_t (p. 154)∗): pure waveform processing

•mha_spec_t (p. 141)∗process(mha_spec_t (p. 141)∗): pure spectral processing

But it is also possible to implement domain transformations (from the time domain into spectrum

or vice versa). The corresponding method signatures are:

•mha_spec_t (p. 141)∗process(mha_wave_t (p. 154)∗): Domain transformation from

waveform to spectrum

•mha_wave_t (p. 154)∗process(mha_spec_t (p. 141)∗): Domain transformation from

spectrum to waveform

For preparation and release of a plugin, the methods

• void prepare(mhaconﬁg_t (p. 155) &) and

• void release(void)

have to be implemented. The openMHA will call the process() method only ater the pre-

pare method has returned and before release() is invoked. It is guarantteed by the open←-

MHA framework that signal processing is performed only between calls of prepare() and

release(). Each call of prepare() is followed by a call of release() (after some optional

signal processing).

For conﬁguration purposes, the plugin class has to export a method called parse() which

implements the openMHA conﬁguration language. We strongly recommend that you do not

implement this method yourself, but by inheriting from the class MHAParser::parser_t (p. 220)

from the openMHA toolbox, directly or indirectly (inheriting from a class that itself inherits from

MHAParser::parser_t (p. 220)).

2.2.2 Connecting the C++ class with the C Interface

A C++ class which provides the appropriate methods can be used as an openMHA Plugin by

connecting it to the C interface using the MHAPLUGIN_CALLBACKS (p. 8) macro.

The openMHA Toolbox library provides a base class MHAPlugin::plugin_t (p. 238)<T>(a

template class) which can be used as the base class for a plugin class. This base class im-

plements some necessary features for openMHA plugin developers like integration into the

openMHA conﬁguration language environment (it inherits from MHAParser::parser_t (p. 220))

and thread-safe runtime conﬁguration update.

8 CONTENTS

2.2.3 Error reporting

When your plugin detects a situation that it cannot handle, like input signal of the wrong signal

domain at preparation time, unsupported number of input channels at preparation time, unsup-

ported combinations of values in the plugin's variables during conﬁguration, it should throw a

C++ exception. The exception should be of type MHAError. Exceptions of this type are caught

by the MHAPLUGIN_CALLBACKS (p. 8) macro for further error Reporting.

Throwing exceptions in response to unsupported conﬁguration changes does not stop the sig-

nal processing. The openMHA conﬁguration language parser will restore the previous value

of that variable and report an error to the conﬁgurator, while the signal processing continues.

Throwing exceptions from the signal processing thread will terminate the signal processing.

Therefore, you should generally avoid throwing exceptions from the process method. Only do

this if you detected a defect in your plugin, and then you should include enough information in

the error message to be able to ﬁx the defect.

2.2.4 Contents of the openMHA Plugin programming interface

2.2.5 Macro Deﬁnition Documentation

2.2.5.1 #deﬁne MHAPLUGIN_CALLBACKS_PREFIX( preﬁx, classname, indom, outdom )

Parameters

classname The name of the plugin class

indom Input domain (wave or spec)

outdom Output domain (wave or spec)

This macro deﬁnes all required openMHA Plugin interface functions and passes calls of these

functions to the corresponding member functions of the class ‘classname'. The parameters

‘indom' and ‘outdom' specify the input and output domain of the processing method. The

MHAInit() and MHADestroy() functions will create or destroy an instance of the class. The

approriate member functions have to be deﬁned in the class. It is suggested to make usage of

the MHAPlugin::plugin_t (p. 238) template class. Exceptions of type MHA_Error (p. 132) are

caught and transformed into apropriate error codes with their corresponding error messages.

2.2.5.2 #deﬁne MHAPLUGIN_CALLBACKS( plugname, classname, indom, outdom

)MHAPLUGIN_CALLBACKS_PREFIX(MHA_STATIC_ ## plugname ##

_,classname,indom,outdom)

Parameters

plugname The ﬁle name of the plugin without the .so or .dll extension

classname The name of the plugin class

indom Input domain (wave or spec)

outdom Output domain (wave or spec)

2.2 The openMHA Plugins (programming interface) 9

This macro deﬁnes all required openMHA Plugin interface functions and passes calls of these

functions to the corresponding member functions of the class ‘classname'. The parameters

‘indom' and ‘outdom' specify the input and output domain of the processing method. The MHA←-

Init() and MHADestroy() functions will create or destroy an instance of the class. The approriate

member functions have to be deﬁned in the class. It is suggested to make usage of the MH←-

APlugin::plugin_t (p. 238) template class. Exceptions of type MHA_Error (p. 132) are caught

and transformed into apropriate error codes with their corresponding error messages.

2.2.5.3 #deﬁne MHAPLUGIN_DOCUMENTATION( plugname, cat, doc

) MHAPLUGIN_DOCUMENTATION_PREFIX(MHA_STATIC_ ## plugname ## _,cat,doc)

Parameters

plugin The ﬁle name of the plugin without the .so or .dll extension

cat Space separated list of categories to which belong the plugin (as const char∗)

doc Documentation of the plugin (as const char∗)

This macro deﬁnes the openMHA Plugin interface function for the documentation. The cate-

gories can be any space seperated list of category names. An empty string will categorize the

plugin in the category 'other'.

The documentation should contain a description of the plugin including a description of the

underlying models, and a paragraph containing hints for usage. The text should be LaTeX

compatible (e.g., avoid or quote underscores in the text part); equations should be formatted

as LaTeX.

10 CONTENTS

2.3 Writing openMHA Plugins. A step-by-step tutorial

A step-by-step tutorial on writing openMHA plugins.

openMHA contains a small number of example plugins as C++ source code. They are meant to

help developers in understanding the concepts of openMHA plugin programming starting from

the simplest example and increasing in complexity. This tutorial explains the basic parts of the

example ﬁles.

2.3.1 example1.cpp

The example plugin ﬁle example1.cpp demonstrates the easiest way to implement an open←-

MHA Plugin. It attenuates the sound signal in the ﬁrst channel by multiplying the sound samples

with a factor. The plugin class MHAPlugin::plugin_t (p. 238) exports several methods, but only

two of them need a non-empty implementation: prepare() method is a pure virtual function

and process() is called when signal processing starts.

#include "mha_plugin.hh"

class example1_t : public MHAPlugin::plugin_t<int> {

public:

example1_t(algo_comm_t & ac,

const std::string & chain_name,

const std::string & algo_name)

: MHAPlugin::plugin_t<int>("",ac)

{/*Do nothing in constructor */}

void release(void)

{/*Do nothing in release */}

Every plugin implementation should include the 'mha_plugin.hh (p. 296)' header ﬁle. C++

helper classes for plugin development are declared in this header ﬁle, and most header ﬁles

needed for plugin development are included by mha_plugin.hh (p. 296).

The class plugin1_t inherits from the class MHAPlugin::plugin_t (p. 238), which then inher-

its from MHAParser::parser_t (p. 220) – the conﬁguration language interface in the method

"parse". Our plugin class therefore exports the working "parse" method inherited from MHA←-

Parser::parser_t (p. 220), and the plugin is visible in the openMHA conﬁguration tree.

The constructor has to accept 3 parameters of correct types. In this simple example, we do not

make use of them.

The release() method is used to free resources after signal processing. In this simple ex-

ample, we do not allocate resources, so there is no need to free them.

2.3.1.1 The prepare method

void prepare(mhaconfig_t & signal_info)

{

if (signal_info.domain != MHA_WAVEFORM)

throw MHA_Error(__FILE__, __LINE__,

"This plugin can only process waveform signals.");

if (signal_info.channels < 1)

throw MHA_Error(__FILE__,__LINE__,

"This plugin requires at least one input channel.");

}

2.3 Writing openMHA Plugins. A step-by-step tutorial 11

Parameters

signal_info Contains information about the input signal's parameters, see mhaconﬁg_t

(p. 155).

The prepare() method of the plugin is called before the signal processing starts, when the

input signal parameters like domain, number of channels, frames per block, and sampling rate

are known. The prepare() method can check these values and raise an exception if the

plugin cannot cope with them, as is done here. The plugin can also change these values if the

signal processing performed in the plugin results in an output signal with different parameters.

This plugin does not change the signal's parameters, therefore they are not modiﬁed here.

2.3.1.2 The signal processing method

mha_wave_t *process(mha_wave_t *signal)

{

unsigned int channel = 0; // channels and frames counting starts with 0

float factor = 0.1f;

unsigned int frame;

// Scale channel number "channel" by "factor":

for(frame = 0; frame < signal->num_frames; frame++) {

// Waveform channels are stored interleaved.

signal->buf[signal->num_channels *frame + channel] *= factor;

}

// Algorithms may process data in-place and return the input signal

// structure as their output signal:

return signal;

}

};

Parameters

signal Pointer to the input signal structure mha_wave_t (p. 154).

Returns

Pointer to the output signal structure. The input signal structure may be reused if the

signal has the same domain and dimensions.

The plugin works with time domain input signal (indicated by the data type mha_wave_t (p. 154)

of the process method's parameter). It scales the ﬁrst channel by a factor of 0.1. The output

signal reuses the structure that previously contained the input signal (in-place processing).

2.3.1.3 Connecting the C++ class with the C plugin interface

Plugins have to export C functions as their interface (to avoid C++ name-mangling issues and

other incompatibilities when mixing plugins compiled with different C++ compilers).

MHAPLUGIN_CALLBACKS(example1,example1_t,wave,wave)

12 CONTENTS

This macro takes care of accessing the C++ class from the C functions required as the plugin's

interface. It implements the C funtions and calls the corresponding C++ instance methods.

Plugin classes should be derived from the template class MHAPlugin::plugin_t (p. 238) to be

compatible with the C interface wrapper.

This macro also catches C++ exceptions of type MHA_Error (p. 132), when raised in the meth-

ods of the plugin class, and reports the error using an error ﬂag as the return value of the

underlying C function. It is therefore important to note that only C++ exceptions of type MH←-

A_Error (p. 132) may be raised by your plugin. If your code uses different Exception classes,

you will have to catch them yourself before control leaves your plugin class, and maybe report

the error by throwing an instance of MHA_Error (p. 132). This is important, because: (1) C++

exceptions cannot cross the plugin interface, which is in C, and (2) there is no error handling

code for your exception classes in the openMHA framework anyways.

2.3.2 example2.cpp

This is another simple example of openMHA plugin written in C++. This plugin also scales one

channel of the input signal, working in the time domain. The scale factor and which channel to

scale (index number) are made accessible to the conﬁguration language.

The algorithm is again implemented as a C++ class.

class example2_t : public MHAPlugin::plugin_t<int> {

MHAParser::int_t scale_ch;

MHAParser::float_t factor;

public:

example2_t(algo_comm_t & ac,

const std::string & chain_name,

const std::string & algo_name);

void prepare(mhaconfig_t & signal_info);

void release(void);

mha_wave_t *process(mha_wave_t *signal);

};

Parameters

scale_ch – the channel number to be scaled

factor – the scale factor of the scaling.

This class again inherits from the template class MHAPlugin::plugin_t (p. 238) for intergration

with the openMHA conﬁguration language. The two data members serve as externally visible

conﬁguration variables. All methods of this class have a non-empty implementation.

2.3.2.1 Constructor

example2_t::example2_t(algo_comm_t & ac,

const std::string & chain_name,

2.3 Writing openMHA Plugins. A step-by-step tutorial 13

const std::string & algo_name)

: MHAPlugin::plugin_t<int>("This plugin multiplies the sound signal"

" in one audio channel by a factor",ac),

scale_ch("Index of audio channel to scale. Indices start from 0.",

"0",

"[0,["),

factor("The scaling factor that is applied to the selected channel.",

"0.1",

"[0,[")

{

insert_item("channel", &scale_ch);

insert_item("factor", &factor);

}

The constructor invokes the superclass constructor with a string parameter. This string pa-

rameter serves as the help text that describes the functionality of the plugin. The constructor

registers conﬁguration variables with the openMHA conﬁguration tree and sets their default val-

ues and permitted ranges. The minimum permitted value for both variables is zero, and there is

no maximum limit (apart from the limitations of the underlying C data type). The conﬁguration

variables have to be registered with the parser node instance using the MHAParser::parser←-

_t::insert_item (p. 222) method.

2.3.2.2 The prepare method

void example2_t::prepare(mhaconfig_t & signal_info)

{

if (signal_info.domain != MHA_WAVEFORM)

throw MHA_Error(__FILE__, __LINE__,

"This plugin can only process waveform signals.");

// The user may have configured scale_ch before prepare is called.

// Check that the configured channel is present in the input signal.

if (signal_info.channels <= unsigned(scale_ch.data))

throw MHA_Error(__FILE__,__LINE__,

"This plugin requires at least %d input channels.",

scale_ch.data + 1);

// Adjust the range of the channel configuration variable so that it

// cannot be set to an out-of-range value during processing.

using MHAParser::StrCnv::val2str;

scale_ch.set_range("[0," + val2str(int(signal_info.channels)) + "[");

}

Parameters

signal_info – contains information about the input signal's parameters, see mhaconﬁg_t

(p. 155).

The user may have changed the conﬁguration variables before preparing the openMHA plugin.

A consequence of this is that it is not sufﬁcient any more to check if the input signal has at least

1 audio channel.

Instead, this prepare method checks that the input signal has enough channels so that the cur-

rent value of scale_ch.data is a valid channel index, i.e. 0 ≤scale_ch.data <signal←-

_info.channels. The prepare method does not have to check that 0 ≤scale_ch.data,

since this is guaranteed by the valid range setting of the conﬁguration variable.

14 CONTENTS

The prepare method then modiﬁes the valid range of the scale_ch variable, it modiﬁes the up-

per bound so that the user cannot set the variable to a channel index higher than the available

channels. Setting the range is done using a string parameter. The prepare method contate-

nates a string of the form "[0,n[". n is the number of channels in the input signal, and is used

here as an exclusive upper boundary. To convert the number of channels into a string, a helper

function for string conversion from the openMHA Toolbox is used. This function is overloaded

and works for several data types.

It is safe to assume that the value of conﬁguration variables does not change while the pre-

pare method executes, since openMHA preparation is triggered from a conﬁguration language

command, and the openMHA conﬁguration language parser is busy and cannot accept other

commands until all openMHA plugins are prepared (or one of them stops the process by raising

an exception). As we will see later in this tutorial, the same assumption cannot be made for the

process method.

2.3.2.3 The release method

void example2_t::release(void)

{

scale_ch.set_range("[0,[");

}

The release method should undo the state changes that were performed by the prepare

method. In this example, the prepare method has reduced the valid range of the scale_←-

ch, so that only valid channels could be selected during signal processing.

The release method reverts this change by setting the valid range back to its original value,

"[0,[".

2.3.2.4 The signal processing method

mha_wave_t *example2_t::process(mha_wave_t *signal)

{

unsigned int frame;

for(frame = 0; frame < signal->num_frames; frame++)

value(signal,frame,scale_ch.data) *= factor.data;

return signal;

}

The processing function uses the current values of the conﬁguration variables to scale every

frame in the selected audio channel.

Note that the value of each conﬁguration variable can change while the processing method

executes, since the process method usually executes in a different thread than the conﬁguration

interface.

For this simple plugin, this is not a problem, but for more advanced plugins, it has to be taken

into consideration. The next section takes a closer look at the problem.

2.3 Writing openMHA Plugins. A step-by-step tutorial 15

Consistency

Assume that one thread reads the value stored in a variable while another thread writes a new

value to that variable concurrently. In this case, you may have a consistency problem. You

would perhaps expect that the value retrieved from the variable either (a) the old value, or (b)

the new value, but not (c) something else. Yet generally case (c) is a possibility.

Fortunately, for some data types on PC systems, case (c) cannot happen. These are 32bit

wide data types with a 4-byte alignment. Therefore, the values in MHAParser::int_t (p. 211)

and MHAParser::ﬂoat_t (p. 208) are always consistent, but this is not the case for vectors,

strings, or complex values. With these, you can get a mixture of the bit patterns of old and new

values, or you can even cause a memory access violation in case a vector or string grows and

has to be reallocated to a different memory address.

There is also a consistency problem if you take the combination of two "safe" datatypes. The

openMHA provides a mechanism that can cope with these types of problems. This thread-safe

runtime conﬁguration update mechanism is introduced in example 5.

2.3.3 example3.cpp

This example introduces the openMHA Event mechanism. Plugins that provide conﬁguration

variable can receive a callback from the parser base class when a conﬁguration variable is

accessed through the conﬁguration language interface.

The third example performes the same processing as before, but now only even channel indices

are permitted when selecting the audio channel to scale. This restriction cannot be ensured by

setting the range of the channel index conﬁguration variable. Instead, the event mechanism of

openMHA conﬁguration variables is used. Conﬁguration variables emit 4 different events, and

your plugin can connect callback methods that are called when the events are triggered. These

events are:

writeaccess

• triggered on write access to a conﬁguration variable.

valuechanged

• triggered when write access to a conﬁguration variable actually changes the value of this

variable.

readaccess

• triggered after the value of the conﬁguration variable has been read.

prereadaccess

• triggered before the value of a conﬁguration variable is read, i.e. the value of the re-

quested variable can be changed by the callback to implement computation on demand.

All of these callbacks are executed in the conﬁguration thread. Therefore, the callback imple-

mentation does not have to be realtime-safe. No other updates of conﬁguration language vari-

ables through the conﬁguration language can happen in parallel, but your processing method

can execute in parallel and may change values.

16 CONTENTS

2.3.3.1 Data member declarations

class example3_t : public MHAPlugin::plugin_t<int> {

MHAParser::int_t scale_ch;

MHAParser::float_t factor;

MHAParser::int_mon_t prepared;

MHAEvents::patchbay_t<example3_t> patchbay;

This plugin exposes another conﬁguration variable, "prepared", that keeps track of the pre-

pared state of the plugin. This is a read-only (monitor) integer variable, i.e. its value can only

be changed by your plugin's C++ code. When using the conﬁguration language interface, the

value of this variable can only be read, but not changed.

The patchbay member is an instance of a connector class that connects event sources with

callbacks.

2.3.3.2 Method declarations

/*Callbacks triggered by Events */

void on_scale_ch_writeaccess();

void on_scale_ch_valuechanged();

void on_scale_ch_readaccess();

void on_prereadaccess();

public:

example3_t(algo_comm_t & ac,

const std::string & chain_name,

const std::string & algo_name);

void prepare(mhaconfig_t & signal_info);

void release(void);

mha_wave_t *process(mha_wave_t *signal);

};

This plugin exposes 4 callback methods that are triggered by events. Multiple events (from the

same or different conﬁguration variables) can be connected to the same callback method, if

desired.

This example plugin uses the valuechanged event to check that the scale_ch conﬁguration

variable is only set to valid values.

The other callbacks only cause log messages to stdout, but the comments in the logging call-

backs give a hint when listening on the events would be useful.

2.3 Writing openMHA Plugins. A step-by-step tutorial 17

2.3.3.3 Example 3 constructor

example3_t::example3_t(algo_comm_t & ac,

const std::string & chain_name,

const std::string & algo_name)

: MHAPlugin::plugin_t<int>("This plugin multiplies the sound signal"

" in one audio channel by a factor",ac),

scale_ch("Index of audio channel to scale. Indices start from 0."

" Only channels with even indices may be scaled.",

"0",

"[0,["),

factor("The scaling factor that is applied to the selected channel.",

"0.1",

"[0,["),

prepared("State of this plugin: 0 = unprepared, 1 = prepared")

{

insert_item("channel", &scale_ch);

insert_item("factor", &factor);

prepared.data = 0;

insert_item("prepared", &prepared);

patchbay.connect(&scale_ch.writeaccess, this,

&example3_t::on_scale_ch_writeaccess);

patchbay.connect(&scale_ch.valuechanged, this,

&example3_t::on_scale_ch_valuechanged);

patchbay.connect(&scale_ch.readaccess, this,

&example3_t::on_scale_ch_readaccess);

patchbay.connect(&scale_ch.prereadaccess, this,

&example3_t::on_prereadaccess);

patchbay.connect(&factor.prereadaccess, this,

&example3_t::on_prereadaccess);

patchbay.connect(&prepared.prereadaccess, this,

&example3_t::on_prereadaccess);

}

The constructor of monitor variables does not take a parameter for setting the initial value.

The single parameter here is the help text describing the contents of the read-only variable.

If the initial value should differ from 0, then the .data member of the conﬁguration variable

has to be set to the initial value in the plugin constructor's body explicitly, as is done here for

demonstration although the initial value of this monitor variable is 0.

Events and callback methods are then connected using the patchbay member variable.

2.3.3.4 The prepare method

void example3_t::prepare(mhaconfig_t & signal_info)

{

if (signal_info.domain != MHA_WAVEFORM)

throw MHA_Error(__FILE__, __LINE__,

"This plugin can only process waveform signals.");

// The user may have configured scale_ch before prepare is called.

// Check that the configured channel is present in the input signal.

if (signal_info.channels <= unsigned(scale_ch.data))

throw MHA_Error(__FILE__,__LINE__,

"This plugin requires at least %d input channels.",

scale_ch.data + 1);

// bookkeeping

prepared.data = 1;

}

The prepare method checks wether the current setting of the scale_ch variable is possible with

the input signal dimension. It does not adjust the range of the variable, since the range alone

is not sufﬁcient to ensure all future settings are also valid: The scale channel index has to be

even.

18 CONTENTS

2.3.3.5 The release method

void example3_t::release(void)

{

prepared.data = 0;

}

The release method is needed for tracking the prepared state only in this example.

2.3.3.6 The signal processing method

mha_wave_t *example3_t::process(mha_wave_t *signal)

{

unsigned int frame;

for(frame = 0; frame < signal->num_frames; frame++)

value(signal,frame,scale_ch.data) *= factor.data;

return signal;

}

The signal processing member function is the same as in example 2.

2.3.3.7 The callback methods

void example3_t::on_scale_ch_writeaccess()

{

printf("Write access: Attempt to set scale_ch=%d.\n", scale_ch.data);

// Can be used to track any writeaccess to the configuration, even

// if it does not change the value. E.g. setting the name of the

// sound file in a string configuration variable can cause a sound

// file player plugin to start playing the sound file from the

// beginning.

}

void example3_t::on_scale_ch_valuechanged()

{

if (scale_ch.data & 1)

throw MHA_Error(__FILE__,__LINE__,

"Attempt to set scale_ch to non-even value %d",

scale_ch.data);

// Can be used to recompute a runtime configuration only if some

// configuration variable actually changed.

}

void example3_t::on_scale_ch_readaccess()

{

printf("scale_ch has been read.\n");

// A configuration variable used as an accumulator can be reset

// after it has been read.

}

void example3_t::on_prereadaccess()

{

printf("A configuration language variable is about to be read.\n");

// Can be used to compute the value on demand.

}

MHAPLUGIN_CALLBACKS(example3,example3_t,wave,wave)

When the writeaccess or valuechanged callbacks throw an MHAError exception, then the

change made to the value of the conﬁguration variable is reverted.

If multiple event sources are connected to a single callback method, then it is not possible

to determine which event has caused the callback to execute. Often, this information is not

crucial, i.e. when the answer to a change of any variable in a set of variables is the same, e.g.

the recomputation of a new runtime conﬁguration that takes all variables of this set as input.

2.3 Writing openMHA Plugins. A step-by-step tutorial 19

2.3.4 example4.cpp

This plugin is the same as example 3 except that it works on the spectral domain (STFT).

2.3.4.1 The Prepare method

void example4_t::prepare(mhaconfig_t & signal_info)

{

if (signal_info.domain != MHA_SPECTRUM)

throw MHA_Error(__FILE__, __LINE__,

"This plugin can only process spectrum signals.");

// The user may have configured scale_ch before prepare is called.

// Check that the configured channel is present in the input signal.

if (signal_info.channels <= unsigned(scale_ch.data))

throw MHA_Error(__FILE__,__LINE__,

"This plugin requires at least %d input channels.",

scale_ch.data + 1);

// bookkeeping

prepared.data = 1;

}

The prepare method now checks that the signal domain is MHA_SPECTRUM.

2.3.4.2 The signal processing method

mha_spec_t *example4_t::process(mha_spec_t *signal)

{

unsigned int bin;

// spectral signal is stored non-interleaved.

mha_complex_t *channeldata =

signal->buf + signal->num_frames *scale_ch.data;

for(bin = 0; bin < signal->num_frames; bin++)

channeldata[bin] *= factor.data;

return signal;

}

The signal processing member function works on the spectral signal instead of the wave signal

as before.

The mha_spec_t (p. 141) instance stores the complex (mha_complex_t (p. 123)) spectral sig-

nal for positive frequences only (since the waveform signal is always real). The num_frames

member of mha_spec_t (p. 141) actually denotes the number of STFT bins.

Please note that different from mha_wave_t (p. 154), a multichannel signal in mha_spec_t

(p. 141) is stored non-interleaved in the signal buffer.

Some arithmetic operations are deﬁned on struct mha_complex_t (p. 123) to facilitate efﬁcient

complex computations. The ∗=operator used here (deﬁned for real and for complex arguments)

is one of them.

2.3.4.3 Connecting the C++ class with the C plugin interface

MHAPLUGIN_CALLBACKS(example4,example4_t,spec,spec)

When connecting a class that performs spectral processing with the C interface, use spec

instead of wave as the domain indicator.

20 CONTENTS

2.3.5 example5.cpp

Many algorithms use complex operations to transform the user space variables into run time

conﬁgurations. If this takes a noticeable time (e.g. more than 100-500 µsec), the update of

the runtime conﬁguration can not take place in the real time processing thread. Furthermore,

the parallel access to complex structures may cause unpredictable results if variables are read

while only parts of them are written to memory (cf. section Consistency (p. 15)). To handle

these situations, a special C++ template class MHAPlugin::plugin_t (p. 238) was designed.

This class helps keeping all access to the conﬁguration language variables in the conﬁguration

thread rather than in the processing thread.

The runtime conﬁguration class example5_t is the parameter of the template class MHA←-

Plugin::plugin_t (p. 238). Its constructor converts the user variables into a runtime conﬁgu-

ration. Because the constructor executes in the conﬁguration thread, there is no harm if the

constructor takes a long time. All other member functions and data members of the runtime

conﬁgurations are accessed only from the signal processing thread (real-time thread).

class example5_t {

public:

example5_t(unsigned int,unsigned int,mha_real_t);

mha_spec_t*process(mha_spec_t*);

private:

unsigned int channel;

mha_real_t scale;

};

The plugin interface class inherits from the plugin template class MHAPlugin::plugin_←-

t(p. 238), parameterised by the runtime conﬁguration. Conﬁguration changes (write access

to the variables) will emit a write access event of the changed variables. These events can be

connected to member functions of the interface class by the help of a MHAEvents::patchbay←-

_t (p. 156) instance.

class plugin_interface_t : public MHAPlugin::plugin_t<example5_t> {

public:

plugin_interface_t(const algo_comm_t&,const std::string&,const std::string&);

mha_spec_t*process(mha_spec_t*);

void prepare(mhaconfig_t&);

private:

void update_cfg();

/*integer variable of MHA-parser: */

MHAParser::int_t scale_ch;

/*float variable of MHA-parser: */

MHAParser::float_t factor;

/*patch bay for connecting configuration parser

events with local member functions: */

MHAEvents::patchbay_t<plugin_interface_t> patchbay;

};

The constructor of the runtime conﬁguration analyses and validates the user variables. If the

conﬁguration is invalid, an exception of type MHA_Error (p. 132) is thrown. This will cause

the openMHA conﬁguration language command which caused the change to fail: The modiﬁed

conﬁguration language variable is then reset to its original value, and the error message will

contain the message string of the MHA_Error (p. 132) exception.

2.3 Writing openMHA Plugins. A step-by-step tutorial 21

example5_t::example5_t(unsigned int ichannel,

unsigned int numchannels,

mha_real_t iscale)

: channel(ichannel),scale(iscale)

{

if( channel >= numchannels )

throw MHA_Error(__FILE__,__LINE__,

"Invalid channel number %d (only %d channels configured).",

channel,numchannels);

}

In this example, the run time conﬁguration class example5_t has a signal processing member

function. In this function, the selected channel is scaled by the given scaling factor.

mha_spec_t*example5_t::process(mha_spec_t*spec)

{

/*Scale channel number "scale_ch" by "factor": */

for(unsigned int fr = 0; fr < spec->num_frames; fr++){

spec->buf[fr + channel *spec->num_frames].re *= scale;

spec->buf[fr + channel *spec->num_frames].im *= scale;

}

return spec;

}

The constructor of the example plugin class is similar to the previous examples. A callback

triggered on write access to the variables is registered using the MHAEvents::patchbay_t

(p. 156) instance.

plugin_interface_t::plugin_interface_t(

const algo_comm_t& iac,

const std::string&,const std::string&)

: MHAPlugin::plugin_t<example5_t>("example plugin configuration structure",iac),

/*initialzing variable 'scale_ch' with MHAParser::int_t(char*name, .... ) */

scale_ch("channel number to be scaled","0","[0,["),

/*initialzing variable 'factor' with MHAParser::float_t(char*name, .... ) */

factor("scale factor","1.0","[0,2]")

{

/*Register variables to the configuration parser: */

insert_item("channel",&scale_ch);

insert_item("factor",&factor);

*On write access to the parser variables a notify callback of

*this class will be called. That funtion will update the runtime

*configuration.

patchbay.connect(&scale_ch.writeaccess,this,&plugin_interface_t::update_cfg);

patchbay.connect(&factor.writeaccess,this,&plugin_interface_t::update_cfg);

}

The processing function can gather the latest valid runtime conﬁguration by a call of poll_←-

config. On success, the class member cfg points to this conﬁguration. On error, if there is

no usable runtime conﬁguration instance, an exception is thrown. In this example, the prepare

method ensures that there is a valid runtime conﬁguration, so that in this example, no error can

be raised at this point. The prepare method is always executed before the process method is

called. The runtime conﬁguration class in this example provides a signal processing method.

The process method of the plugin interface calls the process method of this instance to perform

the actual signal processing.

22 CONTENTS

mha_spec_t*plugin_interface_t::process(mha_spec_t*spec)

{

poll_config();

return cfg->process(spec);

}

The prepare method ensures that a valid runtime conﬁguration exists by creating a new runtime

conﬁguration from the current conﬁguration language variables. If the conﬁguraion is invalid,

then an exception of type MHA_Error (p. 132) is raised and the preparation of the openMHA

fails with an error message.

void plugin_interface_t::prepare(mhaconfig_t& tfcfg)

{

if( tfcfg.domain != MHA_SPECTRUM )

throw MHA_Error(__FILE__,__LINE__,

"Example5: Only spectral processing is supported.");

/*remember the transform configuration (i.e. channel numbers): */

tftype = tfcfg;

/*make sure that a valid runtime configuration exists: */

update_cfg();

}

The update_cfg member function is called when the value of a conﬁguration language variable

changes, or from the prepare method. It allocates a new runtime conﬁguration and registers it

for later access from the real time processing thread. The function push_conﬁg (p. 238) stores

the conﬁguration in a FiFo queue of runtime conﬁgurations. Once they are inserted in the FiFo,

the MHAPlugin::plugin_t (p. 238) template is responsible for deleting runtime conﬁguration

instances stored in the FiFo. You don't need to keep track of the created instances, and you

must not delete them yourself.

void plugin_interface_t::update_cfg()

{

if( tftype.channels )

push_config(new example5_t(scale_ch.data,tftype.channels,factor.data));

}

In the end of the example code ﬁle, the macro MHAPLUGIN_CALLBACKS (p. 8) deﬁnes all

ANSI-C interface functions and passes them to the corresponding C++ class member func-

tions (partly deﬁned by the MHAPlugin::plugin_t (p. 238) template class). All exceptions of

type MHA_Error (p. 132) are caught and transformed into an appropriate error code and error

message.

MHAPLUGIN_CALLBACKS(example5,plugin_interface_t,spec,spec)

2.3 Writing openMHA Plugins. A step-by-step tutorial 23

2.3.6 example6.cpp

This last example is the same as the previous one, but it additionally creates an 'Algorithm

Communication Variable' (AC variable). It calculates the RMS level of a given channel and

stores it into this variable. The variable can be accessed by any other algorithm in the same

chain. To store the data onto disk, the 'acsave' plugin can be used. 'acmon' is a plugin which

converts AC variables into parsable monitor variables.

In the constructor of the plugin class the variable rmsdb is registered under the name

example6_rmslev as a one-dimensional AC variable of type ﬂoat. For registration of other

types, read access and other detailed informations please see Communication between al-

gorithms (p. 27).

example6_t::example6_t(const algo_comm_t& iac,

const std::string&,const std::string&)

: MHAPlugin::plugin_t<cfg_t>("example plugin configuration structure",iac),

/*initialzing variable 'channel_no' with MHAParser::int_t(char*name, .... ) */

channel_no("channel in which the RMS level is measured","0","[0,[")

{

/*Register variables to the configuration parser: */

insert_item("channel",&channel_no);

*On write access to the parser variables a notify callback of

*this class will be called. That funtion will update the runtime

*configuration.

patchbay.connect(&channel_no.writeaccess,this,&example6_t::update_cfg);

*Propagate the level variable to all algorithms in the

*processing chain. If multiple instances of this algorithm are

*required, than it is necessary to use different names for this

*variable (i.e. prefixing the name with the algorithm name

*passed to MHAInit).

ac.insert_var_float( ac.handle, "example6_rmslev", &rmsdb );

}

24 CONTENTS

2.3.7 Debugging openMHA plugins

Suppose you would want to step through the code of your openMHA plugin with a debug-

ger. This example details how to use the linux gdb debugger to inspect the example6_←-

t::prepare() and example6_t::process() routines of example6.cpp (p. 23) example

First, make sure that your plugin is compiled with the compiler option to include debugging

symbols: Apply the -ggdb switch to all gcc, g++ invocations.

Once the plugin is compiled, with debugging symbols, create a test conﬁguration. For example

6, assuming there is an audio ﬁle named input.wav in your working directory, you could create

a conﬁguration ﬁle named ‘debugexample6.cfg', with the following content:

# debugexample6.cfg

fragsize = 64

srate = 44100

nchannels_in = 2

iolib = MHAIOFile

io.in = input.wav

io.out = output.wav

mhalib = example6

mha.channel = 1

cmd=start

Assuming all your binaries and shared-object libraries are in your ‘bin' directory (see READ←-

ME.md), you could start gdb using

$ export MHA_LIBRARY_PATH=$PWD/bin

$ gdb $MHA_LIBRARY_PATH/mha

Set breakpoints in prepare and process methods, and start execution. Note that specifying

the breakpoint by symbol (example6_t::prepare) does not yet work, as the symbol lives in

the openMHA plugin that has not yet been loaded. Specifying by line number works, however.

Specifying the breakpoint by symbol also works once the plugin is loaded (i.e. when the debug-

ger stops in the ﬁrst break point). You can set the breakpoints like this (example shown here is

run in gdb version 7.11.1):

(gdb) run ?read:debugexample6.cfg

Starting program: {openMHA_directory}/bin/mha ?read:debugexample6.cfg

[Thread debugging using libthread_db enabled]

Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".

The Open Master Hearing Aid (openMHA) server

This program comes with ABSOLUTELY NO WARRANTY; for details see file COPYING.

This is free software, and you are welcome to redistribute it

under the terms of the GNU AFFERO GENERAL PUBLIC LICENSE, Version 3;

for details see file COPYING.

Breakpoint 1, example6_t::prepare (this=0x6478b0, tfcfg=...)

at example6.cpp:192

192 if( tfcfg.domain != MHA_WAVEFORM )

(gdb) b example6.cpp:162

Breakpoint 2 at 0x7ffff589744a: file example6.cpp, line 162.

(gdb) c

Continuing.

2.3 Writing openMHA Plugins. A step-by-step tutorial 25

Where ‘{openMHA_directory}' is the directory where openMHA is located (which should also

be your working directory in this case). Next stop is the process() method. You can now

examine and change the variables, step through the program as needed (using, for example ‘n'

to step in the next line):

Breakpoint 2, example6_t::process (this=0x7ffff6a06c0d, wave=0x10a8b550)

at example6.cpp:162

162 {

(gdb) n

163 poll_config();

(gdb)

26 CONTENTS

2.4 The MHA Framework interface

2.5 Communication between algorithms 27

2.5 Communication between algorithms

Algorithms within one chain can share variables for communication with other algorithms.

Collaboration diagram for Communication between algorithms:

The openMHA Toolbox

library

Communication between

algorithms mha_algo_comm.h

Files

• ﬁle mha_algo_comm.h

Header ﬁle for Algorithm Communication.

Namespaces

•MHA_AC

Functions and classes for Algorithm Communication (AC) support.

Classes

• class MHA_AC::spectrum_t

Insert a MHASignal::spectrum_t (p. 261) class into the AC space.

• class MHA_AC::waveform_t

Insert a MHASignal::waveform_t (p. 268) class into the AC space.

• class MHA_AC::int_t

Insert a integer variable into the AC space.

• class MHA_AC::ﬂoat_t

Insert a ﬂoat point variable into the AC space.

• class MHA_AC::double_t

Insert a double precision ﬂoating point variable into the AC space.

• class MHA_AC::ac2matrix_t

Copy AC variable to a matrix.

• class MHA_AC::acspace2matrix_t

Copy all or a subset of all numeric AC variables into an array of matrixes.

• struct algo_comm_t

A reference handle for algorithm communication variables.

• struct comm_var_t

Algorithm communication variable structure.

28 CONTENTS

Functions

•mha_spec_t MHA_AC::get_var_spectrum (algo_comm_t ac, const std::string &name)

Convert an AC variable into a spectrum.

•mha_wave_t MHA_AC::get_var_waveform (algo_comm_t ac, const std::string

&name)

Convert an AC variable into a waveform.

• int MHA_AC::get_var_int (algo_comm_t ac, const std::string &name)

Return value of an integer scalar AC variable.

• ﬂoat MHA_AC::get_var_ﬂoat (algo_comm_t ac, const std::string &name)

Return value of an ﬂoating point scalar AC variable.

• std::vector<ﬂoat >MHA_AC::get_var_vﬂoat (algo_comm_t ac, const std::string

&name)

Return value of an ﬂoating point vector AC variable as standard vector of ﬂoats.

2.5.1 Detailed Description

This mechanism allows interaction between algorithms (i.e. separation of noise estimation

and noise reduction algorithms, combination of dynamic compression and noise estimation).

Through a set of simple C functions, algorithms can propagate variables of any type, even C++

classes, to other algorithms.

An algorithm communication handle (algo_comm_t (p. 88)) is passed at initialisation time to the

constructor of each plugin class constructor (p. 238). This handle contains a reference handle,

algo_comm_t::handle (p. 88), and a number of function pointers, algo_comm_t::insert_var

(p. 88) etc.. An algorithm communication variable is an object of type comm_var_t (p. 95).

For AC variables of numeric types, openMHA Plugins for conversion into parsable monitor

variables, acmon, and storage into Matlab or text ﬁles, acsave, are available.

2.5.2 Function Documentation

2.5.2.1 mha_spec_t MHA_AC::get_var_spectrum ( algo_comm_t ac, const std::string & name )

This function reads an AC variable and tries to convert it into a valid spectrum. The Spectrum

variable is granted to be valid only for one call of the processing function.

Parameters

ac AC handle

name Name of the variable

2.5 Communication between algorithms 29

Returns

Spectrum structure

2.5.2.2 mha_wave_t MHA_AC::get_var_waveform ( algo_comm_t ac, const std::string & name )

This function reads an AC variable and tries to convert it into a valid waveform. The waveform

variable is granted to be valid only for one call of the processing function.

Parameters

ac AC handle

name Name of the variable

Returns

waveform structure

2.5.2.3 int MHA_AC::get_var_int ( algo_comm_t ac, const std::string & name )

Parameters

ac AC handle

name Name of the variable

Returns

Variable value

2.5.2.4 ﬂoat MHA_AC::get_var_ﬂoat ( algo_comm_t ac, const std::string & name )

Parameters

ac AC handle

name Name of the variable

Returns

Variable value

2.5.2.5 std::vector<ﬂoat >MHA_AC::get_var_vﬂoat ( algo_comm_t ac, const std::string & name )

30 CONTENTS

Parameters

ac AC handle

name Name of the variable

Returns

Variable value

2.6 Error handling in the openMHA 31

2.6 Error handling in the openMHA

Errors are reported to the user via the MHA_Error (p. 132) exception.

Collaboration diagram for Error handling in the openMHA:

The openMHA Toolbox

library Error handling in the

openMHA

Classes

• class MHA_Error

Error reporting exception class.

Macros

• #deﬁne MHA_ErrorMsg(x) MHA_Error(__FILE__,__LINE__,"%s",x)

Throw an openMHA error with a text message.

• #deﬁne MHA_assert(x) if(!(x)) throw MHA_Error(__FILE__,__LINE__,"\"%s\" is

false.",#x)

Assertion macro, which throws an MHA_Error (p. 132).

• #deﬁne MHA_assert_equal(a, b) if( a != b ) throw MHA_Error(__FILE__,__LINE__←-

,"\"%s == %s\" is false (%s = %g, %s = %g).",#a,#b,#a,(double)(a),#b,(double)(b))

Equality assertion macro, which throws an MHA_Error (p. 132) with the values.

Functions

• void mha_debug (const char ∗fmt,...)

Print an info message (stderr on Linux, OutputDebugString in Windows).

2.6.1 Detailed Description

2.6.2 Macro Deﬁnition Documentation

2.6.2.1 #deﬁne MHA_ErrorMsg( x)MHA_Error(__FILE__,__LINE__,"%s",x)

32 CONTENTS

Parameters

xText message.

2.6.2.2 #deﬁne MHA_assert( x) if(!(x)) throw MHA_Error(__FILE__,__LINE__,"\"%s\" is false.",#x)

Parameters

xBoolean expression which should be true.

2.6.2.3 #deﬁne MHA_assert_equal( a, b ) if( a != b ) throw MHA_Error(__FILE__,__LINE__,"\"%s ==

%s\" is false (%s = %g, %s = %g).",#a,#b,#a,(double)(a),#b,(double)(b))

Parameters

aNumeric expression which can be converted to double (for printing).

bNumeric expression which should be equal to a

2.7 The openMHA conﬁguration language 33

2.7 The openMHA conﬁguration language

openMHA Plugins that should use the openMHA conﬁguration language for their conﬁguration

have to be implemented in C++ and need to include mha_parser.hh (p. 293).

All required classes and functions for parser access are declared in the namespace MH←-

AParser (p. 76). The plugin class should be derived from the class MHAParser::parser_t

(p. 220) (or MHAPlugin::plugin_t (p. 238)), which symbolises a sub-parser node in the open←-

MHA script hierarchy. Variables of many types can be registered to the sub-parser node by

calling the member function insert_item (p. 222).

The openMHA Plugin template class MHAPlugin::plugin_t (p. 238) together with the Plugin

macro MHAPLUGIN_CALLBACKS (p. 8) provide the callback mappings and correct inheri-

tance. If your plugin is based on that template class, you simply have to use the insert_item

command to give access to your variables, everything else is managed internally.

A complete list of all openMHA script items is given in the description of the MHAParser (p. 76)

namespace.

34 CONTENTS

2.8 The openMHA Toolbox library

The openMHA toolbox is a static C++ library which makes it more comfortable to develop

openMHA plugins.

Collaboration diagram for The openMHA Toolbox library:

Vector and matrix processing

toolbox

Complex arithmetics

in the openMHA

The openMHA Toolbox

library

MHASignal

Error handling in the

openMHA

Fast Fourier Transform

functions

Communication between

algorithms

mha_algo_comm.h

Modules

•Error handling in the openMHA

Errors are reported to the user via the MHA_Error (p. 132) exception.

•Vector and matrix processing toolbox

The vector and matrix processing toolbox consists of a number of classes deﬁned in the

namespace MHASignal (p. 81), and many functions and operators for use with the structures

mha_wave_t (p. 154) and mha_spec_t (p. 141).

•Complex arithmetics in the openMHA

•Fast Fourier Transform functions

Files

• ﬁle mha_algo_comm.h

Header ﬁle for Algorithm Communication.

• ﬁle mha_ﬁlter.hh

Header ﬁle for IIR ﬁlter classes.

• ﬁle mha_signal.hh

Header ﬁle for audio signal handling and processing classes.

• ﬁle mha_tablelookup.hh

Header ﬁle for table lookup classes.

2.8 The openMHA Toolbox library 35

Namespaces

•MHAOvlFilter

Namespace for overlapping FFT based ﬁlter bank classes and functions.

•MHAFilter

Namespace for IIR and FIR ﬁlter classes.

•MHAParser

Name space for the openMHA-Parser conﬁguration language.

•MHASignal

Namespace for audio signal handling and processing classes.

•MHATableLookup

Namespace for table lookup classes.

2.8.1 Detailed Description

It contains the openMHA script language classes.

36 CONTENTS

2.9 Vector and matrix processing toolbox

The vector and matrix processing toolbox consists of a number of classes deﬁned in the names-

pace MHASignal (p. 81), and many functions and operators for use with the structures mha←-

_wave_t (p. 154) and mha_spec_t (p. 141).

Collaboration diagram for Vector and matrix processing toolbox:

Vector and matrix processing

toolbox The openMHA Toolbox

library

MHASignal

Namespaces

•MHASignal

Namespace for audio signal handling and processing classes.

•MHAWindow

Collection of Window types.

Classes

• struct mha_wave_t

Waveform signal structure.

• struct mha_spec_t

Spectrum signal structure.

• struct mha_audio_descriptor_t

Description of an audio fragment (planned as a replacement of mhaconﬁg_t (p. 155)).

• struct mha_audio_t

An audio fragment in the openMHA (planned as a replacement of mha_wave_t (p. 154) and

mha_spec_t (p. 141)).

• class MHASignal::spectrum_t

a signal processing class for spectral data (based on mha_spec_t (p. 141))

• class MHASignal::waveform_t

signal processing class for waveform data (based on mha_wave_t (p. 154))

• class MHASignal::doublebuffer_t

Double-buffering class.

• class MHASignal::hilbert_t

Hilbert transformation of a waveform segment.

• class MHASignal::minphase_t

Minimal phase function.

• class MHASignal::uint_vector_t

2.9 Vector and matrix processing toolbox 37

Vector of unsigned values, used for size and index description of n-dimensional matrixes.

• class MHASignal::matrix_t

n-dimensional matrix with real or complex ﬂoating point values.

• class MHAParser::window_t

MHA conﬁguration interface for a window function generator.

• class MHASignal::delay_wave_t

Delayline containing wave fragments.

• class MHASignal::async_rmslevel_t

Class for asynchronous level metering.

Typedefs

• typedef ﬂoat mha_real_t

openMHA type for real numbers

Functions

•mha_wave_t range (mha_wave_t s, unsigned int k0, unsigned int len)

Return a time interval from a waveform chunk.

•mha_spec_t channels (mha_spec_t s, unsigned int ch_start, unsigned int nch)

Return a channel interval from a spectrum.

• void MHASignal::for_each (mha_wave_t ∗s, mha_real_t(∗fun)(mha_real_t))

Apply a function to each element of a mha_wave_t (p. 154).

•mha_real_t MHASignal::lin2db (mha_real_t x)

Conversion from linear scale to dB (no SPL reference)

•mha_real_t MHASignal::db2lin (mha_real_t x)

Conversion from dB scale to linear (no SPL reference)

•mha_real_t MHASignal::pa2dbspl (mha_real_t x)

Conversion from linear Pascal scale to dB SPL.

•mha_real_t MHASignal::pa22dbspl (mha_real_t x, mha_real_t eps=1e-20f)

Conversion from squared Pascal scale to dB SPL.

•mha_real_t MHASignal::dbspl2pa (mha_real_t x)

Conversion from dB SPL to linear Pascal scale.

•mha_real_t MHASignal::smp2sec (mha_real_t n, mha_real_t srate)

conversion from samples to seconds

•mha_real_t MHASignal::sec2smp (mha_real_t sec, mha_real_t srate)

conversion from seconds to samples

•mha_real_t MHASignal::bin2freq (mha_real_t bin, unsigned fftlen, mha_real_t srate)

conversion from fft bin index to frequency

•mha_real_t MHASignal::freq2bin (mha_real_t freq, unsigned fftlen, mha_real_t srate)

conversion from frequency to fft bin index

•mha_real_t MHASignal::smp2rad (mha_real_t samples, unsigned bin, unsigned fftlen)

conversion from delay in samples to phase shift

38 CONTENTS

•mha_real_t MHASignal::rad2smp (mha_real_t phase_shift, unsigned bin, unsigned

fftlen)

conversion from phase shift to delay in samples

•template<class elem_type >

std::vector<elem_type >MHASignal::dupvec (std::vector<elem_type >vec, unsigned

Duplicate last vector element to match desired size.

•template<class elem_type >

std::vector<elem_type >MHASignal::dupvec_chk (std::vector<elem_type >vec, un-

signed n)

Duplicate last vector element to match desired size, check for dimension.

• bool equal_dim (const mha_wave_t &a, const mha_wave_t &b)

Test for equal dimension of waveform structures.

• bool equal_dim (const mha_wave_t &a, const mhaconﬁg_t &b)

Test for match of waveform dimension with mhaconﬁg structure.

• bool equal_dim (const mha_spec_t &a, const mha_spec_t &b)

Test for equal dimension of spectrum structures.

• bool equal_dim (const mha_spec_t &a, const mhaconﬁg_t &b)

Test for match of spectrum dimension with mhaconﬁg structure.

• bool equal_dim (const mha_wave_t &a, const mha_spec_t &b)

Test for equal dimension of waveform/spectrum structures.

• bool equal_dim (const mha_spec_t &a, const mha_wave_t &b)

Test for equal dimension of waveform/spectrum structures.

• void integrate (mha_wave_t &s)

Numeric integration of a signal vector (real values)

• void integrate (mha_spec_t &s)

Numeric integration of a signal vector (complex values)

• unsigned int size (const mha_wave_t &s)

Return size of a waveform structure.

• unsigned int size (const mha_spec_t &s)

Return size of a spectrum structure.

• unsigned int size (const mha_wave_t ∗s)

Return size of a waveform structure.

• unsigned int size (const mha_spec_t ∗s)

Return size of a spectrum structure.

• void clear (mha_wave_t &s)

Set all values of waveform to zero.

• void clear (mha_wave_t ∗s)

Set all values of waveform to zero.

• void clear (mha_spec_t &s)

Set all values of spectrum to zero.

• void clear (mha_spec_t ∗s)

Set all values of spectrum to zero.

• void assign (mha_wave_t self, mha_real_t val)

Set all values of waveform 'self' to 'val'.

2.9 Vector and matrix processing toolbox 39

• void assign (mha_wave_t self, const mha_wave_t &val)

Set all values of waveform 'self' to 'val'.

• void assign (mha_spec_t self, const mha_spec_t &val)

Set all values of spectrum 'self' to 'val'.

• void timeshift (mha_wave_t &self, int shift)

Time shift of waveform chunk.

•mha_real_t &value (mha_wave_t ∗s, unsigned int fr, unsigned int ch)

Access an element of a waveform structure.

• const mha_real_t &value (const mha_wave_t ∗s, unsigned int fr, unsigned int ch)

Constant access to an element of a waveform structure.

•mha_complex_t &value (mha_spec_t ∗s, unsigned int fr, unsigned int ch)

Access to an element of a spectrum.

• const mha_complex_t &value (const mha_spec_t ∗s, unsigned int fr, unsigned int ch)

Constant access to an element of a spectrum.

•mha_real_t &value (mha_wave_t &s, unsigned int fr, unsigned int ch)

Access to an element of a waveform structure.

• const mha_real_t &value (const mha_wave_t &s, unsigned int fr, unsigned int ch)

Constant access to an element of a waveform structure.

•mha_complex_t &value (mha_spec_t &s, unsigned int fr, unsigned int ch)

Access to an element of a spectrum.

• const mha_complex_t &value (const mha_spec_t &s, unsigned int fr, unsigned int ch)

Constant access to an element of a spectrum.

• std::vector<ﬂoat >std_vector_ﬂoat (const mha_wave_t &)

Converts a mha_wave_t (p. 154) structure into a std::vector<ﬂoat>(interleaved order).

• std::vector<std::vector<ﬂoat > > std_vector_vector_ﬂoat (const mha_wave_t &)

Converts a mha_wave_t (p. 154) structure into a std::vector<std::vector<ﬂoat> > (outer vec-

tor represents channels).

• std::vector<std::vector<mha_complex_t > > std_vector_vector_complex (const

mha_spec_t &)

Converts a mha_spec_t (p. 141) structure into a std::vector<std::vector<mha_complex_t> >

(outer vector represents channels).

•mha_wave_t &operator+= (mha_wave_t &, const mha_real_t &)

Addition operator.

•mha_wave_t &operator+= (mha_wave_t &, const mha_wave_t &)

Addition operator.

•mha_wave_t &operator-= (mha_wave_t &, const mha_wave_t &)

Subtraction operator.

•mha_spec_t &operator-= (mha_spec_t &, const mha_spec_t &)

Subtraction operator.

•mha_wave_t &operator∗=(mha_wave_t &, const mha_real_t &)

Element-wise multiplication operator.

•mha_wave_t &operator∗=(mha_wave_t &, const mha_wave_t &)

Element-wise multiplication operator.

•mha_spec_t &operator∗=(mha_spec_t &, const mha_real_t &)

Element-wise multiplication operator.

40 CONTENTS

•mha_spec_t &operator∗=(mha_spec_t &, const mha_wave_t &)

Element-wise multiplication operator.

•mha_spec_t &operator∗=(mha_spec_t &, const mha_spec_t &)

Element-wise multiplication operator.

•mha_spec_t &operator/= (mha_spec_t &, const mha_spec_t &)

Element-wise division operator.

•mha_wave_t &operator/= (mha_wave_t &, const mha_wave_t &)

Element-wise division operator.

•mha_spec_t &operator+= (mha_spec_t &, const mha_spec_t &)

Addition operator.

•mha_spec_t &operator+= (mha_spec_t &, const mha_real_t &)

Addition operator.

•mha_wave_t &operator∧=(mha_wave_t &self, const mha_real_t &arg)

Exponent operator.

• void MHASignal::copy_channel (mha_spec_t &self, const mha_spec_t &src, unsigned

sch, unsigned dch)

Copy one channel of a source signal.

• void MHASignal::copy_channel (mha_wave_t &self, const mha_wave_t &src, un-

signed src_channel, unsigned dest_channel)

Copy one channel of a source signal.

•mha_real_t MHASignal::rmslevel (const mha_spec_t &s, unsigned int channel, un-

signed int fftlen)

Return RMS level of a spectrum channel.

•mha_real_t MHASignal::colored_intensity (const mha_spec_t &s, unsigned int chan-

nel, unsigned int fftlen, mha_real_t sqfreq_response[ ])

Colored spectrum intensity.

•mha_real_t MHASignal::maxabs (const mha_spec_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t MHASignal::rmslevel (const mha_wave_t &s, unsigned int channel)

Return RMS level of a waveform channel.

•mha_real_t MHASignal::maxabs (const mha_wave_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t MHASignal::maxabs (const mha_wave_t &s)

Find maximal absolute value.

•mha_real_t MHASignal::max (const mha_wave_t &s)

Find maximal value.

•mha_real_t MHASignal::min (const mha_wave_t &s)

Find minimal value.

•mha_real_t MHASignal::sumsqr_channel (const mha_wave_t &s, unsigned int chan-

nel)

Calculate sum of squared values in one channel.

•mha_real_t MHASignal::sumsqr_frame (const mha_wave_t &s, unsigned int frame)

Calculate sum over all channels of squared values.

• void conjugate (mha_spec_t &self)

Replace (!) the value of this mha_spec_t (p. 141) with its conjugate.

2.9 Vector and matrix processing toolbox 41

2.9.1 Detailed Description

2.9.2 Typedef Documentation

2.9.2.1 typedef ﬂoat mha_real_t

This type is expected to be allways the C-type 'ﬂoat' (IEEE 754 single).

2.9.3 Function Documentation

2.9.3.1 mha_wave_t range ( mha_wave_t s, unsigned int k0, unsigned int len )

A waveform chunk containing a time intervall of a larger waveform chunk is returned. The

number of channels remains constant. The data of the output waveform structure points to

the data of the input structure, i.e., write access to the output waveform chunk modiﬁes the

corresponding entries in the input chunk.

Parameters

sWaveform structure

k0 Index of ﬁrst value in output

len Number of frames in output

Returns

Waveform structure representing the sub-interval.

2.9.3.2 mha_spec_t channels ( mha_spec_t s, unsigned int ch_start, unsigned int nch )

Parameters

sInput spectrum

ch_start Index of ﬁrst channel in output

nch Number of channels in output

Returns

Spectrum structure representing the sub-interval.

2.9.3.3 void MHASignal::for_each ( mha_wave_t ∗s, mha_real_t(∗)(mha_real_t)fun )

[inline]

42 CONTENTS

Parameters

sPointer to a mha_wave_t (p. 154) structure

fun Function to be applied (one argument)

2.9.3.4 mha_real_t MHASignal::lin2db ( mha_real_t x)[inline]

Parameters

xLinear input.

2.9.3.5 mha_real_t MHASignal::db2lin ( mha_real_t x)[inline]

Parameters

xdB input.

2.9.3.6 mha_real_t MHASignal::pa2dbspl ( mha_real_t x)[inline]

Parameters

xLinear input.

2.9.3.7 mha_real_t MHASignal::pa22dbspl ( mha_real_t x, mha_real_t eps = 1e-20f )

[inline]

Parameters

xsquared pascal input

eps minimum squared-pascal value

2.9.3.8 mha_real_t MHASignal::dbspl2pa ( mha_real_t x)[inline]

Parameters

xLinear input.

2.9.3.9 mha_real_t MHASignal::smp2sec ( mha_real_t n, mha_real_t srate )[inline]

Parameters

nnumber of samples

srate sampling rate / Hz

2.9 Vector and matrix processing toolbox 43

2.9.3.10 mha_real_t MHASignal::sec2smp ( mha_real_t sec, mha_real_t srate )[inline]

Parameters

sec time in seconds

srate sampling rate / Hz

Returns

number of samples, generally has non-zero fractional part

2.9.3.11 mha_real_t MHASignal::bin2freq ( mha_real_t bin, unsigned fftlen, mha_real_t srate )

[inline]

Parameters

bin index of fft bin, index 0 has dc

fftlen FFT length

srate sampling frequency / Hz

Returns

frequency of fft bin / Hz

2.9.3.12 mha_real_t MHASignal::freq2bin ( mha_real_t freq, unsigned fftlen, mha_real_t srate )

[inline]

Parameters

freq frequency / Hz

fftlen FFT length

srate sampling frequency / Hz

Returns

0-based index of fft bin, generally has non-zero fractional part

2.9.3.13 mha_real_t MHASignal::smp2rad ( mha_real_t samples, unsigned bin, unsigned fftlen )

[inline]

Compute phase shift that needs to be applied to fft spectrum to achieve the desired delay.

Parameters

samples delay in samples. Positive delay: shift current signal to future.

bin index of fft bin, index 0 has dc (index 0 and nyqvist bin cannot be delayed)

fftlen FFT length

44 CONTENTS

Returns

The phase shift in radiant that needs to be applied to fft bin to achieve the desired delay. A

positive delay requires a negative phase shift. If required phase shift is >pi or <-pi, then

the desired delay cannot be applied in the fft domain with given parameters. Required

phase shifts close to pi should not be used. If bin is 0 or nyqvist, returns 0 phase shift.

2.9.3.14 mha_real_t MHASignal::rad2smp ( mha_real_t phase_shift, unsigned bin, unsigned fftlen

)[inline]

Compute delay in samples that is achieved by a phase shift.

Parameters

phase_shift phase shift in radiant

bin index of fft bin, index 0 has dc (index 0 and nyqvist bin cannot be delayed)

fftlen FFT length

Returns

The delay in samples achieved by applying the phase shift. A negative phase shift causes

a positive delay: shifts current signal to future.

2.9.3.15 template<class elem_type >std::vector<elem_type>MHASignal::dupvec ( std::vector<

elem_type >vec, unsigned n)

Parameters

vec Input vector.

nTarget number of elements.

Return values

Resized vector.

2.9.3.16 template<class elem_type >std::vector<elem_type>MHASignal::dupvec_chk ( std::vector<

elem_type >vec, unsigned n)

The input dimension can be either 1 or the target length.

Parameters

vec Input vector.

nTarget number of elements.

2.9 Vector and matrix processing toolbox 45

Return values

Resized vector.

2.9.3.17 bool equal_dim ( const mha_wave_t &a, const mha_spec_t &b)[inline]

Warning

Waveform structures mha_wave_t (p. 154) use interleaved data order, while spectrum

structures mha_spec_t (p. 141) use non-interleaved.

2.9.3.18 bool equal_dim ( const mha_spec_t &a, const mha_wave_t &b)[inline]

Warning

Waveform structures mha_wave_t (p. 154) use interleaved data order, while spectrum

structures mha_spec_t (p. 141) use non-interleaved.

2.9.3.19 void integrate ( mha_wave_t &s)

Parameters

sInput signal vector

2.9.3.20 void integrate ( mha_spec_t &s)

Parameters

sInput signal vector

2.9.3.21 void assign ( mha_wave_t self, mha_real_t val )[inline]

Parameters

self Waveform to be modiﬁed.

val Value to be assigned to all entries of waveform.

2.9.3.22 void assign ( mha_wave_t self, const mha_wave_t &val )

Parameters

self Waveform to be modiﬁed.

val Source waveform structure.

46 CONTENTS

2.9.3.23 void assign ( mha_spec_t self, const mha_spec_t &val )

Parameters

self Spectrum to be modiﬁed.

val Source spectrum.

2.9.3.24 void timeshift ( mha_wave_t &self, int shift )

Shifted areas are ﬁlled with zeros.

Parameters

self Waveform chunk to be shifted

shift Shift amount, positive values shift to later times

2.9.3.25 mha_real_t& value ( mha_wave_t ∗s, unsigned int fr, unsigned int ch )[inline]

Parameters

sWaveform structure

fr Frame number

ch Channel number

Returns

Reference to element

2.9.3.26 const mha_real_t& value ( const mha_wave_t ∗s, unsigned int fr, unsigned int ch )

[inline]

Parameters

sWaveform structure

fr Frame number

ch Channel number

Returns

Reference to element

2.9.3.27 mha_complex_t& value ( mha_spec_t ∗s, unsigned int fr, unsigned int ch )[inline]

2.9 Vector and matrix processing toolbox 47

Parameters

sSpectrum structure

fr Bin number

ch Channel number

Returns

Reference to element

2.9.3.28 const mha_complex_t& value ( const mha_spec_t ∗s, unsigned int fr, unsigned int ch )

[inline]

Parameters

sSpectrum structure

fr Bin number

ch Channel number

Returns

Reference to element

2.9.3.29 mha_real_t& value ( mha_wave_t &s, unsigned int fr, unsigned int ch )[inline]

Parameters

sWaveform structure

fr Frame number

ch Channel number

Returns

Reference to element

2.9.3.30 const mha_real_t& value ( const mha_wave_t &s, unsigned int fr, unsigned int ch )

[inline]

Parameters

sWaveform structure

fr Frame number

ch Channel number

48 CONTENTS

Returns

Reference to element

2.9.3.31 mha_complex_t& value ( mha_spec_t &s, unsigned int fr, unsigned int ch )

[inline]

Parameters

sSpectrum structure

fr Bin number

ch Channel number

Returns

Reference to element

2.9.3.32 const mha_complex_t& value ( const mha_spec_t &s, unsigned int fr, unsigned int ch )

[inline]

Parameters

sSpectrum structure

fr Bin number

ch Channel number

Returns

Reference to element

2.9.3.33 std::vector<ﬂoat>std_vector_ﬂoat ( const mha_wave_t & )

Warning

This function is not real-time safe. Do not use in signal processing thread.

2.9.3.34 std::vector<std::vector<ﬂoat> > std_vector_vector_ﬂoat ( const mha_wave_t & )

Warning

This function is not real-time safe. Do not use in signal processing thread.

2.9.3.35 std::vector<std::vector<mha_complex_t> > std_vector_vector_complex ( const

mha_spec_t & )

Warning

This function is not real-time safe. Do not use in signal processing thread.

2.9 Vector and matrix processing toolbox 49

2.9.3.36 mha_wave_t& operator∧= ( mha_wave_t &self, const mha_real_t &arg )

Warning

This overwrites the xor operator!

2.9.3.37 void MHASignal::copy_channel ( mha_spec_t &self, const mha_spec_t &src, unsigned

sch, unsigned dch )

Parameters

self Destination.

src Source

sch Source channel number

dch Destination channel number

2.9.3.38 void MHASignal::copy_channel ( mha_wave_t &self, const mha_wave_t &src, unsigned

src_channel, unsigned dest_channel )

Parameters

self Destination.

src Source

src_channel Source channel number

dest_channel Destination channel number

2.9.3.39 mha_real_t MHASignal::rmslevel ( const mha_spec_t &s, unsigned int channel, unsigned

int fftlen )

Parameters

sInput spectrum

channel Channel number to be tested

fftlen FFT length (to correctly count the level of the Nyquist bin)

Returns

RMS level in Pa

2.9.3.40 mha_real_t MHASignal::colored_intensity ( const mha_spec_t &s, unsigned int channel,

unsigned int fftlen, mha_real_t sqfreq_response[ ] )

computes the squared sum of the spectrum after ﬁltering with the frequency response

50 CONTENTS

Parameters

sInput spectrum

channel Channel number to be tested

fftlen FFT length (to correctly count the level of the Nyquist bin)

sqfreq_response A squared weighting factor for every fft bin.

Returns

sum of squares. Root of this is the colored level in Pa

2.9.3.41 mha_real_t MHASignal::maxabs ( const mha_spec_t &s, unsigned int channel )

Parameters

sInput signal

channel Channel to be tested

Returns

maximum absolute value

2.9.3.42 mha_real_t MHASignal::rmslevel ( const mha_wave_t &s, unsigned int channel )

Parameters

sInput waveform signal

channel Channel number to be tested

Returns

RMS level in Pa

2.9.3.43 mha_real_t MHASignal::maxabs ( const mha_wave_t &s, unsigned int channel )

Parameters

sInput signal

channel Channel to be tested

2.9 Vector and matrix processing toolbox 51

Returns

maximum absolute value

2.9.3.44 mha_real_t MHASignal::maxabs ( const mha_wave_t &s)

Parameters

sInput signal

Returns

maximum absolute value

2.9.3.45 mha_real_t MHASignal::max ( const mha_wave_t &s)

Parameters

sInput signal

Returns

maximum absolute value

2.9.3.46 mha_real_t MHASignal::min ( const mha_wave_t &s)

Parameters

sInput signal

Returns

maximum absolute value

2.9.3.47 mha_real_t MHASignal::sumsqr_channel ( const mha_wave_t &s, unsigned int channel )

Parameters

sInput signal

channel Channel

52 CONTENTS

Returns

Px2

2.9.3.48 mha_real_t MHASignal::sumsqr_frame ( const mha_wave_t &s, unsigned int frame )

Parameters

sInput signal

frame Frame number

Returns

Px2

2.10 Complex arithmetics in the openMHA 53

2.10 Complex arithmetics in the openMHA

Collaboration diagram for Complex arithmetics in the openMHA:

Complex arithmetics

in the openMHA

The openMHA Toolbox

library

Classes

• struct mha_complex_t

Type for complex ﬂoating point values.

Functions

•mha_complex_t &set (mha_complex_t &self, mha_real_t real, mha_real_t imag=0)

Assign real and imaginary parts to a mha_complex_t (p. 123) variable.

•mha_complex_t mha_complex (mha_real_t real, mha_real_t imag=0)

Create a new mha_complex_t (p. 123) with speciﬁed real and imaginary parts.

•mha_complex_t &set (mha_complex_t &self, const std::complex<mha_real_t >

&stdcomplex)

Assign a mha_complex_t (p. 123) variable from a std::complex.

• std::complex<mha_real_t >stdcomplex (const mha_complex_t &self)

Create a std::complex from mha_complex_t (p. 123).

•mha_complex_t &expi (mha_complex_t &self, mha_real_t angle)

replaces the value of the given mha_complex_t (p. 123) with exp(i∗b).

• double angle (const mha_complex_t &self)

Computes the angle of a complex number in the complex plane.

•mha_complex_t &operator+= (mha_complex_t &self, const mha_complex_t &other)

Addition of two complex numbers, overwriting the ﬁrst.

•mha_complex_t operator+ (const mha_complex_t &self, const mha_complex_←-

t&other)

Addition of two complex numbers, result is a temporary object.

•mha_complex_t &operator+= (mha_complex_t &self, mha_real_t other_real)

Addition of a complex and a real number, overwriting the complex.

•mha_complex_t operator+ (const mha_complex_t &self, mha_real_t other_real)

Addition of a complex and a real number, result is a temporary object.

•mha_complex_t &operator-= (mha_complex_t &self, const mha_complex_t &other)

Subtraction of two complex numbers, overwriting the ﬁrst.

54 CONTENTS

•mha_complex_t operator- (const mha_complex_t &self, const mha_complex_←-

t&other)

Subtraction of two complex numbers, result is a temporary object.

•mha_complex_t &operator-= (mha_complex_t &self, mha_real_t other_real)

Subtraction of a complex and a real number, overwriting the complex.

•mha_complex_t operator- (const mha_complex_t &self, mha_real_t other_real)

Subtraction of a complex and a real number, result is a temporary object.

•mha_complex_t &operator∗=(mha_complex_t &self, const mha_complex_t &other)

Multiplication of two complex numbers, overwriting the ﬁrst.

•mha_complex_t operator∗(const mha_complex_t &self, const mha_complex_←-

t&other)

Multiplication of two complex numbers, result is a temporary object.

•mha_complex_t &operator∗=(mha_complex_t &self, mha_real_t other_real)

Multiplication of a complex and a real number, overwriting the complex.

•mha_complex_t &expi (mha_complex_t &self, mha_real_t angle,mha_real_t factor)

replaces (!) the value of the given mha_complex_t (p. 123) with a ∗exp(i∗b)

•mha_complex_t operator∗(const mha_complex_t &self, mha_real_t other_real)

Multiplication of a complex and a real number, result is a temporary object.

•mha_real_t abs2 (const mha_complex_t &self)

Compute the square of the absolute value of a complex value.

•mha_real_t abs (const mha_complex_t &self)

Compute the absolute value of a complex value.

•mha_complex_t &operator/= (mha_complex_t &self, mha_real_t other_real)

Division of a complex and a real number, overwriting the complex.

•mha_complex_t operator/ (const mha_complex_t &self, mha_real_t other_real)

Division of a complex and a real number, result is a temporary object.

•mha_complex_t &safe_div (mha_complex_t &self, const mha_complex_t &other,

mha_real_t eps, mha_real_t eps2)

Safe division of two complex numbers, overwriting the ﬁrst.

•mha_complex_t &operator/= (mha_complex_t &self, const mha_complex_t &other)

Division of two complex numbers, overwriting the ﬁrst.

•mha_complex_t operator/ (const mha_complex_t &self, const mha_complex_←-

t&other)

Division of two complex numbers, result is a temporary object.

•mha_complex_t operator- (const mha_complex_t &self)

Unary minus on a complex results in a negative temporary object.

• bool operator== (const mha_complex_t &x, const mha_complex_t &y)

Compare two complex numbers for equality.

• bool operator!= (const mha_complex_t &x, const mha_complex_t &y)

Compare two complex numbers for inequality.

• void conjugate (mha_complex_t &self)

Replace (!) the value of this mha_complex_t (p. 123) with its conjugate.

•mha_complex_t _conjugate (const mha_complex_t &self)

Compute the cojugate of this complex value.

• void reciprocal (mha_complex_t &self)

2.10 Complex arithmetics in the openMHA 55

Replace the value of this complex with its reciprocal.

•mha_complex_t _reciprocal (const mha_complex_t &self)

compute the reciprocal of this complex value.

• void normalize (mha_complex_t &self)

Divide a complex by its absolute value, thereby normalizing it (projecting onto the unit circle).

• void normalize (mha_complex_t &self, mha_real_t margin)

Divide a complex by its absolute value, thereby normalizing it (projecting onto the unit circle),

with a safety margin.

• bool almost (const mha_complex_t &self, const mha_complex_t &other, mha_real_t

times_epsilon=1e2)

Compare two complex numbers for equality except for a small relative error.

• bool operator<(const mha_complex_t &x, const mha_complex_t &y)

Compares the absolute values of two complex numbers.

2.10.1 Detailed Description

2.10.2 Function Documentation

2.10.2.1 mha_complex_t& set ( mha_complex_t &self, mha_real_t real, mha_real_t imag =

0)[inline]

Parameters

self The mha_complex_t (p. 123) variable whose value is about to change.

real The new real part.

imag The new imaginary part.

Returns

A reference to the changed variable.

2.10.2.2 mha_complex_t mha_complex ( mha_real_t real, mha_real_t imag = 0)[inline]

Parameters

real The real part.

imag The imaginary part.

Returns

The new value.

2.10.2.3 mha_complex_t& set ( mha_complex_t &self, const std::complex<mha_real_t >&

stdcomplex )[inline]

56 CONTENTS

Parameters

self The mha_complex_t (p. 123) variable whose value is about to change.

stdcomplex The new complex value.

Returns

A reference to the changed variable.

2.10.2.4 mha_complex_t& expi ( mha_complex_t &self, mha_real_t angle )[inline]

Parameters

self The mha_complex_t (p. 123) variable whose value is about to change.

angle The angle in the complex plane [rad].

Returns

A reference to the changed variable.

2.10.2.5 double angle ( const mha_complex_t &self )[inline]

Parameters

self The complex number whose angle is needed.

Returns

The angle of a complex number in the complex plane.

2.10.2.6 mha_complex_t& expi ( mha_complex_t &self, mha_real_t angle, mha_real_t

factor )[inline]

Parameters

self The mha_complex_t (p. 123) variable whose value is about to change.

angle The imaginary exponent.

factor The absolute value of the result.

Returns

A reference to the changed variable.

2.10 Complex arithmetics in the openMHA 57

2.10.2.7 mha_real_t abs2 ( const mha_complex_t &self )[inline]

Returns

The square of the absolute value of self.

2.10.2.8 mha_real_t abs ( const mha_complex_t &self )[inline]

Returns

The absolute value of self.

2.10.2.9 mha_complex_t& safe_div ( mha_complex_t &self, const mha_complex_t &other,

mha_real_t eps, mha_real_t eps2 )[inline]

If abs(divisor) <eps, then divisor is replaced by eps. eps2 = eps∗eps.

2.10.2.10 mha_complex_t _conjugate ( const mha_complex_t &self )[inline]

Returns

A temporary object holding the conjugate value.

2.10.2.11 mha_complex_t _reciprocal ( const mha_complex_t &self )[inline]

Returns

A temporary object holding the reciprocal value.

2.10.2.12 bool almost ( const mha_complex_t &self, const mha_complex_t &other,

mha_real_t times_epsilon = 1e2 )[inline]

Parameters

self The ﬁrst complex number.

other The second complex number.

times_epsilon Permitted relative error is this number multiplied with the machine accuracy

for this Floating point format (std::numeric_limits<mha_real_t>::epsilon)

Returns

true if the relative difference is below times_epsilon ∗std::numeric_limits<mha_real_t>←-

::epsilon

58 CONTENTS

2.11 Fast Fourier Transform functions

Collaboration diagram for Fast Fourier Transform functions:

The openMHA Toolbox

library Fast Fourier Transform

functions

Typedefs

• typedef void ∗mha_fft_t

Handle for an FFT object.

Functions

•mha_fft_t mha_fft_new (unsigned int n)

Create a new FFT handle.

• void mha_fft_free (mha_fft_t h)

Destroy an FFT handle.

• void mha_fft_wave2spec (mha_fft_t h, const mha_wave_t ∗in, mha_spec_t ∗out)

Tranform waveform segment into spectrum.

• void mha_fft_wave2spec (mha_fft_t h, const mha_wave_t ∗in, mha_spec_t ∗out, bool

swaps)

Tranform waveform segment into spectrum.

• void mha_fft_spec2wave (mha_fft_t h, const mha_spec_t ∗in, mha_wave_t ∗out)

Tranform spectrum into waveform segment.

• void mha_fft_spec2wave (mha_fft_t h, const mha_spec_t ∗in, mha_wave_t ∗out, un-

signed int offset)

Tranform spectrum into waveform segment.

• void mha_fft_forward (mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut)

Complex to complex FFT (forward).

• void mha_fft_backward (mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut)

Complex to complex FFT (backward).

• void mha_fft_forward_scale (mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut)

Complex to complex FFT (forward).

• void mha_fft_backward_scale (mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut)

Complex to complex FFT (backward).

• void mha_fft_wave2spec_scale (mha_fft_t h, const mha_wave_t ∗in, mha_spec_←-

t∗out)

Tranform waveform segment into spectrum.

• void mha_fft_spec2wave_scale (mha_fft_t h, const mha_spec_t ∗in, mha_wave_←-

t∗out)

Tranform spectrum into waveform segment.

2.11 Fast Fourier Transform functions 59

2.11.1 Detailed Description

2.11.2 Typedef Documentation

2.11.2.1 typedef void∗mha_fft_t

This FFT object is used by the functions mha_fft_wave2spec and mha_fft_spec2wave. The F←-

FT back-end is the FFTW library. The back-end is completely hidden, including external header

ﬁles or linking external libraries is not required.

2.11.3 Function Documentation

2.11.3.1 mha_fft_t mha_fft_new ( unsigned int n)

Parameters

nFFT length.

Create a new FFT handle.

Parameters

nFFT length

Return values

FFT object

2.11.3.2 void mha_fft_free ( mha_fft_t h)

Parameters

hHandle to be destroyed.

Destroy an FFT handle.

Parameters

hFFT object to be removed

2.11.3.3 void mha_fft_wave2spec ( mha_fft_t h, const mha_wave_t ∗in, mha_spec_t ∗out )

60 CONTENTS

Parameters

hFFT handle.

in Input waveform segment.

out Output spectrum.

Tranform waveform segment into spectrum.

Parameters

hFFT object handle

in pointer to input waveform signal

out pointer to output spectrum signal (has to be allocated)

2.11.3.4 void mha_fft_wave2spec ( mha_fft_t h, const mha_wave_t ∗in, mha_spec_t ∗out, bool

swaps )

Like normal wave2spec, but swaps wave buffer halves before transforming if the swaps param-

eter is true.

Warning: These openMHA FFTs adopt a nonstandard scaling scheme in which the forward

transform scales by 1/N and the backward does not scale. We would recommend using the

'_scale' methods instead.

Parameters

hFFT handle.

in Input waveform segment.

out Output spectrum.

swaps Function swaps the ﬁrst and second half of the waveform buffer before the FFT

transform when this parameter is set to true.

2.11.3.5 void mha_fft_spec2wave ( mha_fft_t h, const mha_spec_t ∗in, mha_wave_t ∗out )

Warning: These openMHA FFTs adopt a nonstandard scaling scheme in which the forward

transform scales by 1/N and the backward does not scale. We would recommend using the

'_scale' methods instead.

Parameters

hFFT handle.

in Input spectrum.

out Output waveform segment.

2.11 Fast Fourier Transform functions 61

Tranform spectrum into waveform segment.

Parameters

hFFT object handle

in pointer to input spectrum

out pointer to output waveform signal (has to be allocated)

2.11.3.6 void mha_fft_spec2wave ( mha_fft_t h, const mha_spec_t ∗in, mha_wave_t ∗out,

unsigned int offset )

out may have fewer number of frames than needed for a complete iFFT. Only as many frames

are written into out as ﬁt, starting with offset offset of the complete iFFT.

Warning: These openMHA FFTs adopt a nonstandard scaling scheme in which the forward

transform scales by 1/N and the backward does not scale. We would recommend using the

'_scale' methods instead.

Parameters

hFFT handle.

in Input spectrum.

out Output waveform segment.

offset Offset into iFFT wave buffer

Tranform spectrum into waveform segment.

Only part of the iFFT is tranferred into the out buffer.

Out may have fewer number of freames than needed for a complete iFFT. Only as many frames

are written into out as ﬁt, starting with offset offset of the complete iFFT.

Parameters

hFFT object handle

in pointer to input spectrum

out pointer to output waveform signal (has to be allocated)

offset Offset into complete iFFT buffer.

2.11.3.7 void mha_fft_forward ( mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut )

sIn and sOut need to have nfft bins (please note that mha_spec_t (p. 141) typically has nfft/2+1

bins for half-complex representation).

Warning: These openMHA FFTs adopt a nonstandard scaling scheme in which the forward

transform scales by 1/N and the backward does not scale. We would recommend using the

'_scale' methods instead.

62 CONTENTS

Parameters

hFFT handle.

sIn Input spectrum.

sOut Output spectrum.

2.11.3.8 void mha_fft_backward ( mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut )

sIn and sOut need to have nfft bins (please note that mha_spec_t (p. 141) typically has nfft/2+1

bins for half-complex representation).

Warning: These openMHA FFTs adopt a nonstandard scaling scheme in which the forward

transform scales by 1/N and the backward does not scale. We would recommend using the

'_scale' methods instead.

Parameters

hFFT handle.

sIn Input spectrum.

sOut Output spectrum.

2.11.3.9 void mha_fft_forward_scale ( mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut )

sIn and sOut need to have nfft bins (please note that mha_spec_t (p. 141) typically has nfft/2+1

bins for half-complex representation).

The _scale methods use standard DFT scaling: There is no scaling in the forward transforma-

tion, and 1/N scaling for the backward.

Parameters

hFFT handle.

sIn Input spectrum.

sOut Output spectrum.

2.11.3.10 void mha_fft_backward_scale ( mha_fft_t h, mha_spec_t ∗sIn, mha_spec_t ∗sOut )

sIn and sOut need to have nfft bins (please note that mha_spec_t (p. 141) typically has nfft/2+1

bins for half-complex representation).

The _scale methods use standard DFT scaling: There is no scaling in the forward transforma-

tion, and 1/N scaling for the backward.

2.11 Fast Fourier Transform functions 63

Parameters

hFFT handle.

sIn Input spectrum.

sOut Output spectrum.

2.11.3.11 void mha_fft_wave2spec_scale ( mha_fft_t h, const mha_wave_t ∗in, mha_spec_t ∗

out )

The _scale methods use standard DFT scaling: There is no scaling in the forward transforma-

tion, and 1/N scaling for the backward.

Parameters

hFFT handle.

in Input waveform segment.

out Output spectrum.

2.11.3.12 void mha_fft_spec2wave_scale ( mha_fft_t h, const mha_spec_t ∗in, mha_wave_t ∗

out )

The _scale methods use standard DFT scaling: There is no scaling in the forward transforma-

tion, and 1/N scaling for the backward.

Parameters

hFFT handle.

in Input spectrum.

out Output waveform segment.

64 CONTENTS

3 Namespace Documentation

3.1 AuditoryProﬁle Namespace Reference

Namespace for classes and functions around the auditory proﬁle (e.g., audiogram handling)

Classes

• class fmap_t

A class to store frequency dependent data (e.g., HTL and UCL).

• class parser_t

Class to make the auditory proﬁle accessible through the parser interface.

• class proﬁle_t

The Auditory Proﬁle class.

3.1.1 Detailed Description

The auditory proﬁle as deﬁned by HearCom or BMBF Modellbasierte Hoergeraete is stored in

the class AuditoryProﬁle::proﬁle_t (p. 94). Until a complete deﬁnition is available, only the

currently needed elements are implemented.

3.2 DynComp Namespace Reference

dynamic compression related classes and functions

Classes

• class dc_afterburn_rt_t

Real-time class for after burn effect.

• class dc_afterburn_t

Afterburn class, to be deﬁned as a member of compressors.

• class dc_afterburn_vars_t

Variables for dc_afterburn_t (p. 97) class.

• class gaintable_t

Gain table class.

3.2 DynComp Namespace Reference 65

Functions

•mha_real_t interp1 (const std::vector<mha_real_t >&vX, const std::vector<mha_←-

real_t >&vY, mha_real_t X)

One-dimensional linear interpolation.

•mha_real_t interp2 (const std::vector<mha_real_t >&vX, const std::vector<mha←-

_real_t >&vY, const std::vector<std::vector<mha_real_t > > &mZ, mha_real_t X,

mha_real_t Y)

Linear interpolation in a two-dimensional ﬁeld.

3.2.1 Function Documentation

3.2.1.1 mha_real_t DynComp::interp1 ( const std::vector<mha_real_t >&vX, const std::vector<

mha_real_t >&vY, mha_real_t X)

Parameters

vX Vector with input samples.

vY Vector with values at input samples.

XInput value to be interpolated.

Return values

Interpolated value Y(X) at position X.

3.2.1.2 mha_real_t DynComp::interp2 ( const std::vector<mha_real_t >&vX, const std::vector<

mha_real_t >&vY, const std::vector<std::vector<mha_real_t > > &mZ, mha_real_t

X, mha_real_t Y)

Parameters

vX Vector with input samples, ﬁrst dimension.

vY Vector with input samples, second dimension.

mZ Field with values at input samples.

XFirst dimension of input value to be interpolated.

YSecond dimension of input value to be interpolated.

Return values

Interpolated value Z(X,Y) at position X,Y.

66 CONTENTS

3.3 MHA_AC Namespace Reference

Functions and classes for Algorithm Communication (AC) support.

Classes

• class ac2matrix_t

Copy AC variable to a matrix.

• class acspace2matrix_t

Copy all or a subset of all numeric AC variables into an array of matrixes.

• class double_t

Insert a double precision ﬂoating point variable into the AC space.

• class ﬂoat_t

Insert a ﬂoat point variable into the AC space.

• class int_t

Insert a integer variable into the AC space.

• class spectrum_t

Insert a MHASignal::spectrum_t (p. 261) class into the AC space.

• class waveform_t

Insert a MHASignal::waveform_t (p. 268) class into the AC space.

Functions

•mha_spec_t get_var_spectrum (algo_comm_t ac, const std::string &name)

Convert an AC variable into a spectrum.

•mha_wave_t get_var_waveform (algo_comm_t ac, const std::string &name)

Convert an AC variable into a waveform.

• int get_var_int (algo_comm_t ac, const std::string &name)

Return value of an integer scalar AC variable.

• ﬂoat get_var_ﬂoat (algo_comm_t ac, const std::string &name)

Return value of an ﬂoating point scalar AC variable.

• std::vector<ﬂoat >get_var_vﬂoat (algo_comm_t ac, const std::string &name)

Return value of an ﬂoating point vector AC variable as standard vector of ﬂoats.

3.3.1 Detailed Description

3.4 MHA_TCP Namespace Reference

A Namespace for TCP helper classes.

3.5 MHAEvents Namespace Reference 67

Classes

• class Async_Notify

Portable Multiplexable cross-thread notiﬁcation.

• class Client

A portable class for a tcp client connections.

• class Connection

Connection (p. 144) handles Communication between client and server, is used on both sides.

• class Event_Watcher

OS-independent event watcher, uses select on Unix and WaitForMultipleObjects on Windows.

• class Sockread_Event

Watch socket for incoming data.

• class Thread

A very simple class for portable threads.

• class Timeout_Watcher

OS-independent event watcher with internal ﬁxed-end-time timeout.

• class Wakeup_Event

A base class for asynchronous wakeup events.

Functions

• std::string STRERROR (int err)

Portable conversion from error number to error string.

• std::string HSTRERROR (int err)

Portable conversion from hostname error number to error string.

• int N_ERRNO ()

Portable access to last network error number.

• int H_ERRNO ()

Portable access to last hostname error number.

• int G_ERRNO ()

Portable access to last non-network error number.

• double dtime ()

Time access function for system's high resolution time, retrieve current time as double.

• double dtime (const struct timeval &tv)

Time access function for unix' high resolution time, converts struct timeval to double.

• struct timeval stime (double d)

Time access function for unix' high resolution time, converts time from double to struct timeval.

3.5 MHAEvents Namespace Reference

Collection of event handling classes.

68 CONTENTS

Classes

• class emitter_t

Class for emitting openMHA events.

• class patchbay_t

Patchbay which connects any event emitter with any member function of the parameter class.

3.6 MHAFilter Namespace Reference

Namespace for IIR and FIR ﬁlter classes.

Classes

• class adapt_ﬁlter_t

Adaptive ﬁlter.

• class blockprocessing_polyphase_resampling_t

A class that does polyphase resampling and takes into account block processing.

• class complex_bandpass_t

Complex bandpass ﬁlter.

• class diff_t

Differentiator class (non-normalized)

• class fftﬁlter_t

FFT based FIR ﬁlter implementation.

• class fftﬁlterbank_t

FFT based FIR ﬁlterbank implementation.

• class ﬁlter_t

Generic IIR ﬁlter class.

• class gamma_ﬂt_t

Class for gammatone ﬁlter.

• class iir_ﬁlter_t

IIR ﬁlter class wrapper for integration into parser structure.

• class iir_ord1_real_t

First order recursive ﬁlter.

• class o1_ar_ﬁlter_t

First order attack-release lowpass ﬁlter.

• class o1ﬂt_lowpass_t

First order low pass ﬁlter.

• class o1ﬂt_maxtrack_t

First order maximum tracker.

• class o1ﬂt_mintrack_t

First order minimum tracker.

• class partitioned_convolution_t

A ﬁlter class for partitioned convolution.

3.6 MHAFilter Namespace Reference 69

• class polyphase_resampling_t

A class that does polyphase resampling.

• class resampling_ﬁlter_t

Hann shaped low pass ﬁlter for resampling.

• class smoothspec_t

Smooth spectral gains, create a windowed impulse response.

• struct transfer_function_t

a structure containing a source channel number, a target channel number, and an impulse

response.

• struct transfer_matrix_t

A sparse matrix of transfer function partitionss.

Functions

• void o1_lp_coeffs (const mha_real_t tau, const mha_real_t fs, mha_real_t &c1, mha←-

_real_t &c2)

Set ﬁrst order ﬁlter coefﬁcients from time constant and sampling rate.

• void butter_stop_ord1 (double ∗A, double ∗B, double f1, double f2, double fs)

Setup a ﬁrst order butterworth band stop ﬁlter.

•MHASignal::waveform_t ∗spec2ﬁr (const mha_spec_t ∗spec, const unsigned int fftlen,

const MHAWindow::base_t &window, const bool minphase)

Create a windowed impulse response/FIR ﬁlter coefﬁcients from a spectrum.

• unsigned gcd (unsigned a, unsigned b)

greatest common divisor

• double sinc (double x)

sin(x)/x function, coping with x=0.

• std::pair<unsigned, unsigned >resampling_factors (ﬂoat source_sampling_rate, ﬂoat

target_sampling_rate, ﬂoat factor=1.0f)

Computes rational resampling factor from two sampling rates.

3.6.1 Function Documentation

3.6.1.1 void MHAFilter::o1_lp_coeffs ( const mha_real_t tau, const mha_real_t fs, mha_real_t &

c1, mha_real_t &c2 )

Parameters

tau Time constant

fs Sampling rate

Return values

c1 Recursive ﬁlter coefﬁcient

c2 Non-recursive ﬁlter coefﬁcient

70 CONTENTS

3.6.1.2 void MHAFilter::butter_stop_ord1 ( double ∗A, double ∗B, double f1, double f2, double fs )

This function calculates the ﬁlter coefﬁcients of a ﬁrst order butterworth band stop ﬁlter.

Return values

Arecursive ﬁlter coefﬁcients

Bnon recursive ﬁlter coefﬁcients

Parameters

f1 lower frequency

f2 upper frequency

fs sample frequency

3.6.1.3 MHASignal::waveform_t ∗MHAFilter::spec2ﬁr ( const mha_spec_t ∗spec, const

unsigned int fftlen, const MHAWindow::base_t &window, const bool minphase )

Parameters

spec Input spectrum

fftlen FFT length of spectrum

window Window shape (with length, e.g. initialized with MHAWindow::hanning(54)).

minphase Flag, true if original phase should be discarded and replaced by a minimal

phase function.

3.6.1.4 double MHAFilter::sinc ( double x)

This is the historical sinc function, not the normalized sinc function.

3.6.1.5 std::pair<unsigned, unsigned >MHAFilter::resampling_factors ( ﬂoat source_sampling_rate,

ﬂoat target_sampling_rate, ﬂoat factor = 1.0f )

The function will fail if either sampling_rate ∗factor is not an integer

Parameters

source_sampling_rate The original sampling rate

target_sampling_rate The desired sampling rate

factor A helper factor to use for non-integer sampling rates

3.7 MHAIOJack Namespace Reference 71

Returns

a pair that contains ﬁrst the upsampling factor and second the downsampling factor re-

quired for the speciﬁed resampling.

Exceptions

MHA_Error (p. 132)if no rational resampling factor can be found.

3.7 MHAIOJack Namespace Reference

JACK IO.

Classes

• class io_jack_t

Main class for JACK IO.

3.8 MHAJack Namespace Reference

Classes and functions for openMHA and JACK interaction.

Classes

• class client_avg_t

Generic JACK client for averaging a system response across time.

• class client_noncont_t

Generic client for synchronous playback and recording of waveform fragments.

• class client_t

Generic asynchronous JACK client.

• class port_t

Class for one channel/port.

72 CONTENTS

Functions

• void io (mha_wave_t ∗s_out, mha_wave_t ∗s_in, const std::string &name, const std←-

::vector<std::string >&p_out, const std::vector<std::string >&p_in, ﬂoat ∗srate=NULL,

unsigned int ∗fragsize=NULL, bool use_jack_transport=false)

Functional form of generic client for synchronous playback and recording of waveform frag-

ments.

• std::vector<unsigned int >get_port_capture_latency (const std::vector<std::string >

&ports)

Return the JACK port latency of ports.

• std::vector<int >get_port_capture_latency_int (const std::vector<std::string >

&ports)

Return the JACK port latency of ports.

• std::vector<unsigned int >get_port_playback_latency (const std::vector<std::string

>&ports)

Return the JACK port latency of ports.

3.8.1 Function Documentation

3.8.1.1 std::vector<unsigned int >MHAJack::get_port_capture_latency ( const std::vector<

std::string >&ports )

Parameters

ports Ports to be tested

Returns

Latency vector (one entry for each port)

3.8.1.2 std::vector<int >MHAJack::get_port_capture_latency_int ( const std::vector<std::string >&

ports )

Parameters

ports Ports to be tested

Returns

Latency vector (one entry for each port)

3.8.1.3 std::vector<unsigned int >MHAJack::get_port_playback_latency ( const std::vector<

std::string >&ports )

3.9 MHAMultiSrc Namespace Reference 73

Parameters

ports Ports to be tested

Returns

Latency vector (one entry for each port)

3.9 MHAMultiSrc Namespace Reference

Collection of classes for selecting audio chunks from multiple sources.

Classes

• class base_t

Base class for source selection.

3.9.1 Detailed Description

3.10 MHAOvlFilter Namespace Reference

Namespace for overlapping FFT based ﬁlter bank classes and functions.

Namespaces

•FreqScaleFun

Transform functions from linear scale in Hz to new frequency scales.

•ShapeFun

Shape functions for overlapping ﬁlters.

Classes

• class fftfb_t

FFT based overlapping ﬁlter bank.

• class fftfb_vars_t

Set of conﬁguration variables for FFT-based overlapping ﬁlters.

• class fspacing_t

Class for frequency spacing, used by ﬁlterbank shape generator class.

• class overlap_save_ﬁlterbank_t

A time-domain minimal phase ﬁlter bank with frequency shapes from MHAOvlFilter::fftfb_t

(p. 196).

74 CONTENTS

3.11 MHAOvlFilter::FreqScaleFun Namespace Reference

Transform functions from linear scale in Hz to new frequency scales.

Functions

•mha_real_t hz2hz (mha_real_t x)

Dummy scale transformation Hz to Hz.

•mha_real_t hz2bark (mha_real_t x)

Transformation to bark scale.

•mha_real_t hz2log (mha_real_t x)

Third octave frequency scale.

3.11.1 Function Documentation

3.11.1.1 mha_real_t MHAOvlFilter::FreqScaleFun::hz2hz ( mha_real_t x)

This function implements a dummy scale transformation (linear frequency scale).

Parameters

xInput frequency in Hz

Returns

Frequency in Hz

3.11.1.2 mha_real_t MHAOvlFilter::FreqScaleFun::hz2bark ( mha_real_t x)

This function implements a critical band rate (bark) scale.

Parameters

xInput frequency in Hz

Returns

Critical band rate in Bark

3.11.1.3 mha_real_t MHAOvlFilter::FreqScaleFun::hz2log ( mha_real_t x)

This function implements a third octave scale. Frequencies below 16 Hz are mapped to 16 Hz.

3.12 MHAOvlFilter::ShapeFun Namespace Reference 75

Parameters

xFrequency in Hz

Returns

Third octaves relative to 1000 Hz

3.12 MHAOvlFilter::ShapeFun Namespace Reference

Shape functions for overlapping ﬁlters.

Functions

•mha_real_t rect (mha_real_t x)

Filter shape function for rectangular ﬁlters.

•mha_real_t linear (mha_real_t x)

Filter shape function for sawtooth ﬁlters.

•mha_real_t hann (mha_real_t x)

Filter shape function for hanning shaped ﬁlters.

3.12.1 Function Documentation

3.12.1.1 mha_real_t MHAOvlFilter::ShapeFun::rect ( mha_real_t x)

This function creates rectangular ﬁlter shapes. The edge is exactly half way between two center

frequencies (on a given scale).

Parameters

xInput value in the range [-1,1].

Returns

Weigth function in the range [0,1]

3.12.1.2 mha_real_t MHAOvlFilter::ShapeFun::linear ( mha_real_t x)

This function creates sawtooth ﬁlter shapes. They rise linearly form 0 to 1 in the interval from

the lower neighbor center frequency to the band center frequency and from 1 to 0 in the interval

from the band center frequency to the upper neighbour band center frequency. Linear means

linear on a given frequency scale.

76 CONTENTS

Parameters

xInput value in the range [-1,1].

Returns

Weigth function in the range [0,1]

3.12.1.3 mha_real_t MHAOvlFilter::ShapeFun::hann ( mha_real_t x)

This function creates hanning window shaped ﬁlters.

Parameters

xInput value in the range [-1,1].

Returns

Weigth function in the range [0,1]

3.13 MHAParser Namespace Reference

Name space for the openMHA-Parser conﬁguration language.

Namespaces

•StrCnv

String converter namespace.

Classes

• class base_t

Base class for all parser items.

• class bool_mon_t

Monitor with string value.

• class bool_t

Variable with a boolean value ("yes"/"no")

• class commit_t

Parser variable with event-emission functionality.

• class complex_mon_t

Monitor with complex value.

3.13 MHAParser Namespace Reference 77

• class complex_t

Variable with complex value.

• class ﬂoat_mon_t

Monitor with ﬂoat value.

• class ﬂoat_t

Variable with ﬂoat value.

• class int_mon_t

Monitor variable with int value.

• class int_t

Variable with integer value.

• class keyword_list_t

Keyword list class.

• class kw_t

Variable with keyword list value.

• class mcomplex_mon_t

Matrix of complex numbers monitor.

• class mcomplex_t

Matrix variable with complex value.

• class mﬂoat_mon_t

Matrix of ﬂoats monitor.

• class mﬂoat_t

Matrix variable with ﬂoat value.

• class mhapluginloader_t

Class to create a plugin loader in a parser, including the load logic.

• class monitor_t

Base class for monitors and variable nodes.

• class parser_t

Parser node class.

• class range_var_t

Base class for all variables with a numeric value range.

• class string_mon_t

Monitor with string value.

• class string_t

Variable with a string value.

• class variable_t

Base class for variable nodes.

• class vcomplex_mon_t

Monitor with vector of complex values.

• class vcomplex_t

Vector variable with complex value.

• class vﬂoat_mon_t

Vector of ﬂoats monitor.

• class vﬂoat_t

Vector variable with ﬂoat value.

• class vint_mon_t

78 CONTENTS

Vector of ints monitor.

• class vint_t

Variable with vector<int>value.

• class vstring_mon_t

Vector of monitors with string value.

• class vstring_t

Vector variable with string values.

• class window_t

MHA conﬁguration interface for a window function generator.

Functions

• void strreplace (std::string &, const std::string &, const std::string &)

string replace function

3.13.1 Detailed Description

This namespace contains all classes which are needed for the implementation of the open←-

MHA conﬁguration language. For details on the script language itself please see section The

openMHA conﬁguration language (p. 33).

3.13.2 List of valid MHAParser items

•Sub-parser:parser_t (p. 220)

•Variables:

Numeric variables: int_t (p. 211), vint_t (p. 232), ﬂoat_t (p. 208), vﬂoat_t (p. 229),

mﬂoat_t (p. 218)

Other variables: string_t (p. 225), vstring_t (p. 234), kw_t (p. 214), bool_t (p. 204)

•Monitors:

Numeric monitors: int_mon_t (p. 210), vint_mon_t (p. 231), ﬂoat_mon_t (p. 207),

vﬂoat_mon_t (p. 228)

mﬂoat_mon_t (p. 217)

mcomplex_mon_t (p. 215)

Other monitors: bool_mon_t (p. 203), string_mon_t (p. 224), vstring_mon_t (p. 233)

Members can be inserted into the conﬁguration namespace by using MHAParser::insert_item()

or the insert_member() (p. 296) macro.

3.13.3 Function Documentation

3.13.3.1 void MHAParser::strreplace ( std::string & s, const std::string & arg, const std::string & rep )

3.14 MHAParser::StrCnv Namespace Reference 79

Parameters

starget string

arg search pattern

rep replace pattern

3.14 MHAParser::StrCnv Namespace Reference

String converter namespace.

Functions

• int num_brackets (const std::string &s)

Return number of brackets at beginning and end of string.

• void str2val (const std::string &, bool &)

Convert from string.

• void str2val (const std::string &, ﬂoat &)

Convert from string.

• void str2val (const std::string &, mha_complex_t &)

Convert from string.

• void str2val (const std::string &, int &)

Convert from string.

• void str2val (const std::string &, keyword_list_t &)

Convert from string.

• void str2val (const std::string &, std::string &)

Convert from string.

•template<class arg_t >

void str2val (const std::string &s, std::vector<arg_t >&val)

Converter for vector types.

•template<>

void str2val<mha_real_t >(const std::string &s, std::vector<mha_real_t >&v)

Converter for vector<mha_real_t>with Matlab-style expansion.

•template<class arg_t >

void str2val (const std::string &s, std::vector<std::vector<arg_t > > &val)

Converter for matrix types.

• std::string val2str (const bool &)

Convert to string.

• std::string val2str (const ﬂoat &)

Convert to string.

• std::string val2str (const mha_complex_t &)

Convert to string.

• std::string val2str (const int &)

Convert to string.

• std::string val2str (const keyword_list_t &)

80 CONTENTS

Convert to string.

• std::string val2str (const std::string &)

Convert to string.

• std::string val2str (const std::vector<ﬂoat >&)

Convert to string.

• std::string val2str (const std::vector<mha_complex_t >&)

Convert to string.

• std::string val2str (const std::vector<int >&)

Convert to string.

• std::string val2str (const std::vector<std::string >&)

Convert to string.

• std::string val2str (const std::vector<std::vector<ﬂoat > > &)

Convert to string.

• std::string val2str (const std::vector<std::vector<mha_complex_t > > &)

Convert to string.

3.14.1 Detailed Description

The functions deﬁned in this namespace manage the conversions from C++ variables to strings

and back. It was tried to keep a matlab compatible string format for vectors and vectors of

vectors.

3.14.2 Function Documentation

3.14.2.1 int MHAParser::StrCnv::num_brackets ( const std::string & s)

Parameters

sString

Returns

Number of brackets, or -1 for empty string

3.15 MHAPlugin Namespace Reference

Namespace for openMHA plugin class templates and thread-safe runtime conﬁgurations.

Classes

• class conﬁg_t

Template class for thread safe conﬁguration.

• class plugin_t

The template class for C++ openMHA plugins.

3.16 MHASignal Namespace Reference 81

3.16 MHASignal Namespace Reference

Namespace for audio signal handling and processing classes.

Classes

• class async_rmslevel_t

Class for asynchronous level metering.

• class delay_t

Class to realize a simple delay of waveform streams.

• class delay_wave_t

Delayline containing wave fragments.

• class doublebuffer_t

Double-buffering class.

• class hilbert_t

Hilbert transformation of a waveform segment.

• class loop_wavefragment_t

Copy a ﬁxed waveform fragment to a series of waveform fragments of other size.

• class matrix_t

n-dimensional matrix with real or complex ﬂoating point values.

• class minphase_t

Minimal phase function.

• class quantizer_t

Simple simulation of ﬁxpoint quantization.

• class ringbuffer_t

A ringbuffer class for time domain audio signal, which makes no assumptions with respect to

fragment size.

• class schroeder_t

Schroeder tone complex class.

• class spectrum_t

a signal processing class for spectral data (based on mha_spec_t (p. 141))

• class subsample_delay_t

implements subsample delay in spectral domain.

• class uint_vector_t

Vector of unsigned values, used for size and index description of n-dimensional matrixes.

• class waveform_t

signal processing class for waveform data (based on mha_wave_t (p. 154))

82 CONTENTS

Functions

• void for_each (mha_wave_t ∗s, mha_real_t(∗fun)(mha_real_t))

Apply a function to each element of a mha_wave_t (p. 154).

•mha_real_t lin2db (mha_real_t x)

Conversion from linear scale to dB (no SPL reference)

•mha_real_t db2lin (mha_real_t x)

Conversion from dB scale to linear (no SPL reference)

•mha_real_t pa2dbspl (mha_real_t x)

Conversion from linear Pascal scale to dB SPL.

•mha_real_t pa22dbspl (mha_real_t x, mha_real_t eps=1e-20f)

Conversion from squared Pascal scale to dB SPL.

•mha_real_t dbspl2pa (mha_real_t x)

Conversion from dB SPL to linear Pascal scale.

•mha_real_t smp2sec (mha_real_t n, mha_real_t srate)

conversion from samples to seconds

•mha_real_t sec2smp (mha_real_t sec, mha_real_t srate)

conversion from seconds to samples

•mha_real_t bin2freq (mha_real_t bin, unsigned fftlen, mha_real_t srate)

conversion from fft bin index to frequency

•mha_real_t freq2bin (mha_real_t freq, unsigned fftlen, mha_real_t srate)

conversion from frequency to fft bin index

•mha_real_t smp2rad (mha_real_t samples, unsigned bin, unsigned fftlen)

conversion from delay in samples to phase shift

•mha_real_t rad2smp (mha_real_t phase_shift, unsigned bin, unsigned fftlen)

conversion from phase shift to delay in samples

•template<class elem_type >

std::vector<elem_type >dupvec (std::vector<elem_type >vec, unsigned n)

Duplicate last vector element to match desired size.

•template<class elem_type >

std::vector<elem_type >dupvec_chk (std::vector<elem_type >vec, unsigned n)

Duplicate last vector element to match desired size, check for dimension.

• void copy_channel (mha_spec_t &self, const mha_spec_t &src, unsigned sch, un-

signed dch)

Copy one channel of a source signal.

• void copy_channel (mha_wave_t &self, const mha_wave_t &src, unsigned src_←-

channel, unsigned dest_channel)

Copy one channel of a source signal.

•mha_real_t rmslevel (const mha_spec_t &s, unsigned int channel, unsigned int fftlen)

Return RMS level of a spectrum channel.

•mha_real_t colored_intensity (const mha_spec_t &s, unsigned int channel, unsigned

int fftlen, mha_real_t sqfreq_response[ ])

Colored spectrum intensity.

•mha_real_t maxabs (const mha_spec_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t rmslevel (const mha_wave_t &s, unsigned int channel)

3.16 MHASignal Namespace Reference 83

Return RMS level of a waveform channel.

•mha_real_t maxabs (const mha_wave_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t maxabs (const mha_wave_t &s)

Find maximal absolute value.

•mha_real_t max (const mha_wave_t &s)

Find maximal value.

•mha_real_t min (const mha_wave_t &s)

Find minimal value.

•mha_real_t sumsqr_channel (const mha_wave_t &s, unsigned int channel)

Calculate sum of squared values in one channel.

•mha_real_t sumsqr_frame (const mha_wave_t &s, unsigned int frame)

Calculate sum over all channels of squared values.

• void limit (mha_wave_t &s, const mha_real_t &min, const mha_real_t &max)

Limit the singal in the waveform buffer to the range [min, max].

•template<class elem_type >

elem_type kth_smallest (elem_type array[ ], unsigned n, unsigned k)

Fast search for the kth smallest element of an array.

•template<class elem_type >

elem_type median (elem_type array[ ], unsigned n)

Fast median search.

•template<class elem_type >

elem_type mean (const std::vector<elem_type >&data, elem_type start_val)

Calculate average of elements in a vector.

•template<class elem_type >

std::vector<elem_type >quantile (std::vector<elem_type >data, const std::vector<

elem_type >&p)

Calculate quantile of elements in a vector.

• void saveas_mat4 (const mha_spec_t &data, const std::string &varname, FILE ∗fh)

Save a openMHA spectrum as a variable in a Matlab4 ﬁle.

• void saveas_mat4 (const mha_wave_t &data, const std::string &varname, FILE ∗fh)

Save a openMHA waveform as a variable in a Matlab4 ﬁle.

• void saveas_mat4 (const std::vector<mha_real_t >&data, const std::string &varname,

FILE ∗fh)

Save a ﬂoat vector as a variable in a Matlab4 ﬁle.

• void copy_permuted (mha_wave_t ∗dest, const mha_wave_t ∗src)

Copy contents of a waveform to a permuted waveform.

Variables

• unsigned long int signal_counter = 0

Signal counter to produce signal ID strings.

3.16.1 Function Documentation

3.16.1.1 void MHASignal::limit ( mha_wave_t &s, const mha_real_t &min, const mha_real_t &

max )

84 CONTENTS

Parameters

sThe signal to limit. The signal in this wave buffer is modiﬁed.

min lower limit

max upper limit

3.16.1.2 template<class elem_type >elem_type MHASignal::kth_smallest ( elem_type array[ ],

unsigned n, unsigned k)

The order of elements is altered, but not completely sorted. Using the algorithm from N. Wirth,

published in "Algorithms + data structures = programs", Prentice-Hall, 1976

Parameters

array Element array

Postcondition

The order of elements in the array is altered. array[k] then holds the result.

Parameters

nnumber of elements in array

Precondition

n>= 1

Parameters

kThe k'th smalles element is returned: k = 0 returns the minimum, k = (n-1)/2 returns the

median, k=(n-1) returns the maximum

Precondition

k<n

Returns

The kth smallest array element

3.16.1.3 template<class elem_type >elem_type MHASignal::median ( elem_type array[ ], unsigned n)

[inline]

The order of elements is altered, but not completely sorted.

3.16 MHASignal Namespace Reference 85

Parameters

array Element array

Postcondition

The order of elements in the array is altered. array[(n-1)/2] then holds the median.

Parameters

nnumber of elements in array

Precondition

n>= 1

Returns

The median of the array elements

3.16.1.4 template<class elem_type >elem_type MHASignal::mean ( const std::vector<elem_type >&

data, elem_type start_val )[inline]

Parameters

data Input vector

start_val Value for initialization of the return value before sum.

Returns

The average of the vector elements

3.16.1.5 template<class elem_type >std::vector<elem_type>MHASignal::quantile ( std::vector<

elem_type >data, const std::vector<elem_type >&p)[inline]

Parameters

data Input vector

pVector of probability values.

86 CONTENTS

Returns

Vector of quantiles of input data, one entry for each probability value.

3.16.1.6 void MHASignal::saveas_mat4 ( const mha_spec_t &data, const std::string & varname,

FILE ∗fh )

Parameters

data openMHA spectrum to be saved.

varname Matlab variable name (Matlab4 limitations on maximal length are not checked).

fh File handle to Matlab4 ﬁle.

3.16.1.7 void MHASignal::saveas_mat4 ( const mha_wave_t &data, const std::string & varname,

FILE ∗fh )

Parameters

data openMHA waveform to be saved.

varname Matlab variable name (Matlab4 limitations on maximal length are not checked).

fh File handle to Matlab4 ﬁle.

3.16.1.8 void MHASignal::saveas_mat4 ( const std::vector<mha_real_t >&data, const std::string &

varname, FILE ∗fh )

Parameters

data Float vector to be saved.

varname Matlab variable name (Matlab4 limitations on maximal length are not checked).

fh File handle to Matlab4 ﬁle.

3.16.1.9 void MHASignal::copy_permuted ( mha_wave_t ∗dest, const mha_wave_t ∗src )

Parameters

dest Destination waveform

src Source waveform

The total size of src and dest must be the same, num_frames and num_channels must be

exchanged in dest.

3.17 MHATableLookup Namespace Reference

Namespace for table lookup classes.

3.18 MHAWindow Namespace Reference 87

Classes

• class xy_table_t

Class for interpolation with non-equidistant x values.

3.18 MHAWindow Namespace Reference

Collection of Window types.

Classes

• class bartlett_t

Bartlett window.

• class base_t

Common base for window types.

• class blackman_t

Blackman window.

• class fun_t

Generic window based on a generator function.

• class hamming_t

Hamming window.

• class hanning_t

von-Hann window

• class rect_t

Rectangular window.

• class user_t

User deﬁned window.

Functions

• ﬂoat rect (ﬂoat)

Rectangular window function.

• ﬂoat bartlett (ﬂoat)

Bartlett window function.

• ﬂoat hanning (ﬂoat)

Hanning window function.

• ﬂoat hamming (ﬂoat)

Hamming window function.

• ﬂoat blackman (ﬂoat)

Blackman window function.

88 CONTENTS

4 Class Documentation

4.1 algo_comm_t Struct Reference

A reference handle for algorithm communication variables.

Public Attributes

• void ∗handle

AC variable control handle.

• int(∗insert_var )(void ∗, const char ∗,comm_var_t)

• int(∗insert_var_int )(void ∗, const char ∗, int ∗)

• int(∗insert_var_ﬂoat )(void ∗, const char ∗, ﬂoat ∗)

• int(∗remove_var )(void ∗, const char ∗)

Remove an AC variable.

• int(∗remove_ref )(void ∗, void ∗)

Remove all AC variable which refer to address.

• int(∗is_var )(void ∗, const char ∗)

Test if an AC variable exists.

• int(∗get_var )(void ∗, const char ∗,comm_var_t ∗)

Get the variable handle of an AC variable.

• int(∗get_var_int )(void ∗, const char ∗, int ∗)

Get the value of an int AC variable.

• int(∗get_var_ﬂoat )(void ∗, const char ∗, ﬂoat ∗)

Get the value of a ﬂoat AC variable.

• int(∗get_entries )(void ∗, char ∗, unsigned int)

Return a space separated list of all variable names.

• const char ∗(∗get_error )(int)

Convert AC error codes into human readable error messages.

4.1.1 Detailed Description

This structure contains a coontrol handle and a set of function pointers for sharing variables

within one processing chain. See also section Communication between algorithms (p. 27).

4.1.2 Member Data Documentation

4.1.2.1 algo_comm_t::insert_var

This function can register a variable to be shared within one chain. If a variable of this name

exists it will be overwritten.

4.1 algo_comm_t Struct Reference 89

Parameters

hAC handle

nname of variable. May not be empty. Must not contain space character. The name is

copied, therefore it is allowed that the char array pointed to gets invalid after return.

vvariable handle of type comm_var_t (p. 95)

Returns

Error code or zero on success

4.1.2.2 algo_comm_t::insert_var_int

This function can register an int variable to be shared with other algorithms. It behaves similar

to ac.insert_var.

Parameters

hAC handle

nname of variable

vpointer on the variable

Returns

Error code or zero on success

4.1.2.3 algo_comm_t::insert_var_ﬂoat

This function can register a ﬂoat variable to be shared with other algorithms. It behaves similar

to ac.insert_var.

Parameters

hAC handle

nname of variable

vpointer on the variable

Returns

Error code or zero on success

4.1.2.4 algo_comm_t::remove_var

Remove (unregister) an AC variable. After calling this function, the variable is not available to

ac.is_var or ac.get_var. The data pointer is not affected.

90 CONTENTS

Parameters

hAC handle

nname of variable to be removed

Returns

Error code or zero on success

4.1.2.5 algo_comm_t::remove_ref

This function removes all AC variables whos data ﬁeld points to the given address.

Parameters

hAC handle

paddress which should not be referred to any more

Returns

Error code or zero on success

4.1.2.6 algo_comm_t::is_var

This function tests if an AC variable of a given name exists. Use ac.get_var to get information

about the variables type and dimension.

Parameters

hAC handle

nname of variable

Returns

1 if the variable exists, 0 otherwise

4.1.2.7 algo_comm_t::get_var

This function returns the variable handle comm_var_t (p. 95) of a variable of the given name.

If no variable of that name exists, an error code is returned.

4.1 algo_comm_t Struct Reference 91

Parameters

hAC handle

nname of variable

vpointer to a AC variable object

Returns

Error code or zero on success

4.1.2.8 algo_comm_t::get_var_int

This function returns the value of an int AC variable of the given name. If no variable exists, the

variable type is mismatching or more than one entry is registered, a corresponding error code

is returned. This is a special version of ac.get_var.

Parameters

hAC handle

nname of variable

vpointer on an int variable to store the result

Returns

Error code or zero on success

4.1.2.9 algo_comm_t::get_var_ﬂoat

This function returns the value of a ﬂoat AC variable of the given name. If no variable exists, the

variable type is mismatching or more than one entry is registered, a corresponding error code

is returned. This is a special version of ac.get_var.

Parameters

hAC handle

nname of variable

vpointer on a ﬂoat variable to store the result

Returns

Error code or zero on success

92 CONTENTS

4.1.2.10 algo_comm_t::get_entries

This function returns the names of all registered variables, separated by a single space.

4.2 AuditoryProﬁle::fmap_t Class Reference 93

Parameters

hAC handle

Return values

ret Character buffer for return value

Parameters

len length of character buffer

Returns

Error code or zero on success. -1: invalid ac handle. -3: not enough room in character

buffer to store all variable names.

4.1.2.11 algo_comm_t::get_error

Parameters

eError code

Returns

Error message

4.2 AuditoryProﬁle::fmap_t Class Reference

A class to store frequency dependent data (e.g., HTL and UCL).

Inherits map<mha_real_t, mha_real_t >.

Public Member Functions

• std::vector<mha_real_t >get_frequencies () const

Return conﬁgured frequencies.

• std::vector<mha_real_t >get_values () const

Return stored values corresponding to the frequencies.

94 CONTENTS

4.3 AuditoryProﬁle::parser_t Class Reference

Class to make the auditory proﬁle accessible through the parser interface.

Inheritance diagram for AuditoryProﬁle::parser_t:

AuditoryProfile::parser_t

MHAParser::parser_t

MHAParser::base_t

Additional Inherited Members

4.4 AuditoryProﬁle::proﬁle_t Class Reference

The Auditory Proﬁle class.

Classes

• class ear_t

Class for ear-dependent parameters, e.g., audiograms or unilateral loudness scaling.

Public Member Functions

•AuditoryProﬁle::proﬁle_t::ear_t get_ear (unsigned int channel) const

Return ear information of channel number.

Public Attributes

•AuditoryProﬁle::proﬁle_t::ear_t L

Left ear data.

•AuditoryProﬁle::proﬁle_t::ear_t R

Right ear data.

4.5 AuditoryProﬁle::proﬁle_t::ear_t Class Reference 95

4.4.1 Detailed Description

See deﬁnition of auditory proﬁle

Currently only the audiogram data is stored.

4.5 AuditoryProﬁle::proﬁle_t::ear_t Class Reference

Class for ear-dependent parameters, e.g., audiograms or unilateral loudness scaling.

4.6 comm_var_t Struct Reference

Algorithm communication variable structure.

Public Attributes

• unsigned int data_type

Type of data.

• unsigned int num_entries

Number of entries.

• unsigned int stride

length of one row (C interpretation) or of one column (Fortran interpretation)

• void ∗data

Pointer to variable data.

4.6.1 Detailed Description

Algorithm communication variables (AC variables) are objects of this type. The member data

is a pointer to the variable ‘data'. This pointer has to be valid for the lifetime of this AC variable.

The member ‘data_type' can be one of the predeﬁned types or any user deﬁned type. The

member ‘num_entries' describes the number of elements of this base type stored at the pointer

address.

An AC variable can be registered with the \ref

algo_comm_t::insert_var "insert_var" function.

96 CONTENTS

4.6.2 Member Data Documentation

4.6.2.1 comm_var_t::data_type

This can be one of the predeﬁned types

• MHA_AC_CHAR

• MHA_AC_INT

• MHA_AC_MHAREAL

• MHA_AC_FLOAT

• MHA_AC_DOUBLE

• MHA_AC_MHACOMPLEX

• MHA_AC_VEC_FLOAT or any user deﬁned type with a value greater than

• MHA_AC_USER

4.7 DynComp::dc_afterburn_rt_t Class Reference

Real-time class for after burn effect.

Public Member Functions

• void burn (ﬂoat &Gin, ﬂoat Lin, unsigned int band, unsigned int channel)

gain modiﬁer method (afterburn).

4.7.1 Detailed Description

The constructor processes the parameters and creates pre-processed variables for efﬁcient

realtime processing.

4.7.2 Member Function Documentation

4.7.2.1 void DynComp::dc_afterburn_rt_t::burn ( ﬂoat & Gin, ﬂoat Lin, unsigned int band, unsigned int

channel )[inline]

Parameters

Gin Linear gain.

Lin Input level (Pascal).

band Filter band number.

4.8 DynComp::dc_afterburn_t Class Reference 97

Output level for MPO is estimated by Gin ∗Lin.

4.8 DynComp::dc_afterburn_t Class Reference

Afterburn class, to be deﬁned as a member of compressors.

Inheritance diagram for DynComp::dc_afterburn_t:

DynComp::dc_afterburn_t

DynComp::dc_afterburn

_vars_t

MHAParser::parser_t

MHAParser::base_t

MHAPlugin::config_t

< dc_afterburn_rt_t >

Additional Inherited Members

4.9 DynComp::dc_afterburn_vars_t Class Reference

Variables for dc_afterburn_t (p. 97) class.

98 CONTENTS

Inheritance diagram for DynComp::dc_afterburn_vars_t:

DynComp::dc_afterburn

_vars_t

DynComp::dc_afterburn_t

MHAParser::parser_t

MHAParser::base_t

Additional Inherited Members

4.10 DynComp::gaintable_t Class Reference

Gain table class.

Public Member Functions

•gaintable_t (const std::vector<mha_real_t >&LInput, const std::vector<mha_real_t

>&FCenter, unsigned int channels)

Constructor.

• void update (std::vector<std::vector<std::vector<mha_real_t >>>newGain)

Update gains from an external table.

•mha_real_t get_gain (mha_real_t Lin, mha_real_t Fin, unsigned int channel)

Read Gain from gain table.

•mha_real_t get_gain (mha_real_t Lin, unsigned int band, unsigned int channel)

Read Gain from gain table.

• void get_gain (const mha_wave_t &Lin, mha_wave_t &Gain)

Read Gains from gain table.

• unsigned int nbands () const

Return number of frequency bands.

• unsigned int nchannels () const

Return number of audio channels.

• std::vector<std::vector<mha_real_t > > get_iofun () const

Return current input-output function.

4.10 DynComp::gaintable_t Class Reference 99

4.10.1 Detailed Description

This gain table is intended to efﬁcient table lookup, i.e, interpolation of levels, and optional

interpolation of frequencies. Sample input levels and sample frequencies are given in the con-

structor. The gain entries can be updated with the update() (p. 99) member function via a gain

prescription rule from an auditory proﬁle.

4.10.2 Constructor & Destructor Documentation

4.10.2.1 gaintable_t::gaintable_t ( const std::vector<mha_real_t >&LInput, const std::vector<

mha_real_t >&FCenter, unsigned int channels )

Parameters

LInput Input level samples, in equivalent LTASS_combined dB SPL.

FCenter Frequency samples in Hz (e.g., center frequencies of ﬁlterbank).

channels Number of audio channels (typically 2).

4.10.3 Member Function Documentation

4.10.3.1 void gaintable_t::update ( std::vector<std::vector<std::vector<mha_real_t >>>

newGain )

Parameters

newGain New gain table entries.

Dimension change is not allowed. The number of entries are checked.

4.10.3.2 mha_real_t gaintable_t::get_gain ( mha_real_t Lin, mha_real_t Fin, unsigned int channel

)

Parameters

Lin Input level

Fin Input frequency (no match required)

channel Audio channel

4.10.3.3 mha_real_t gaintable_t::get_gain ( mha_real_t Lin, unsigned int band, unsigned int

channel )

100 CONTENTS

Parameters

Lin Input level

band Input frequency band

channel Audio channel

4.10.3.4 void gaintable_t::get_gain ( const mha_wave_t &Lin, mha_wave_t &Gain )

Parameters

Lin Input levels.

Gain Output gain.

The number of channels in Lin and Gain must match the number of bands times number of

channels in the gaintable.

4.11 expression_t Class Reference

Class for separating a string into a left hand value and a right hand value.

4.11.1 Detailed Description

A list of valid operators can be provided. After construction, the class members lval, rval and

op contain the apropriate contents.

4.12 io_ﬁle_t Class Reference

File IO.

Inheritance diagram for io_ﬁle_t:

io_file_t

MHAParser::parser_t

MHAParser::base_t

4.13 io_lib_t Class Reference 101

Public Member Functions

• void prepare (int, int)

Allocate buffers, activate FILE client and install internal ports.

• void release ()

Remove FILE client and deallocate internal ports and buffers.

Additional Inherited Members

4.12.1 Member Function Documentation

4.12.1.1 void io_ﬁle_t::prepare ( int nch_in, int nch_out )

4.12.1.2 void io_ﬁle_t::release ( )

4.13 io_lib_t Class Reference

Class for loading MHA sound IO module.

Inherits MHAParser::c_ifc_parser_t.

Public Member Functions

•io_lib_t (int fragsize, ﬂoat samplerate, IOProcessEvent_t proc_event, void ∗proc_handle,

IOStartedEvent_t start_event, void ∗start_handle, IOStoppedEvent_t stop_event, void

∗stop_handle, std::string libname)

load and initialize MHA sound io module.

•∼io_lib_t ()

Deinitialize and unload this MHA sound io module.

• void prepare (unsigned int inch, unsigned int outch)

Prepare the sound io module.

• void start ()

Tell the sound io module to start sound processing.

4.13.1 Member Function Documentation

4.13.1.1 void io_lib_t::prepare ( unsigned int inch, unsigned int outch )

After preparation, the sound io module may start the sound processing at any time (external

trigger). When the sound processing is started, the sound io module will call the start_event

callback.

102 CONTENTS

Parameters

inch number of input channels

outch number of output channels

4.13.1.2 void io_lib_t::start ( )

Some io modules need this, for others that wait for external events this method might do noth-

ing.

4.14 io_parser_t Class Reference

Main class for Parser IO.

Inheritance diagram for io_parser_t:

io_parser_t

MHAParser::parser_t

MHAParser::base_t

Public Member Functions

• void prepare (int, int)

Allocate buffers, activate JACK client and install internal ports.

• void release ()

Remove JACK client and deallocate internal ports and buffers.

4.15 io_tcp_fwcb_t Class Reference 103

Additional Inherited Members

4.14.1 Detailed Description

4.14.2 Member Function Documentation

4.14.2.1 void io_parser_t::prepare ( int nch_in, int nch_out )

4.14.2.2 void io_parser_t::release ( )

4.15 io_tcp_fwcb_t Class Reference

TCP sound-io library's interface to the framework callbacks.

Public Member Functions

•io_tcp_fwcb_t (IOProcessEvent_t proc_event, void ∗proc_handle, IOStartedEvent_←-

tstart_event, void ∗start_handle, IOStoppedEvent_t stop_event, void ∗stop_handle)

Constructor stores framework handles and initializes error numbers to 0.

• virtual ∼io_tcp_fwcb_t ()

Do-nothing destructor.

• virtual void start ()

Call the framework's start callback.

• virtual int process (mha_wave_t ∗sIn, mha_wave_t ∗&sOut)

Call the frameworks processing callback.

• virtual void set_errnos (int proc_err, int io_err)

Save error numbers to use during.

• virtual void stop ()

Call the frameworks stop callback.

Private Attributes

• IOProcessEvent_t proc_event

Pointer to signal processing callback function.

• IOStartedEvent_t start_event

Pointer to start notiﬁcation callback function.

• IOStoppedEvent_t stop_event

Pointer to stop notiﬁcation callback function.

• void ∗proc_handle

Handles belonging to framework.

• int proc_err

Errors from the processing callback and from the TCP IO itself are stored here before closing

Network handles.

104 CONTENTS

4.15.1 Detailed Description

4.15.2 Constructor & Destructor Documentation

4.15.2.1 io_tcp_fwcb_t::io_tcp_fwcb_t ( IOProcessEvent_t proc_event, void ∗proc_handle,

IOStartedEvent_t start_event, void ∗start_handle, IOStoppedEvent_t stop_event, void ∗

stop_handle )

4.15.2.2 virtual io_tcp_fwcb_t::∼io_tcp_fwcb_t ( ) [inline],[virtual]

4.15.3 Member Function Documentation

4.15.3.1 void io_tcp_fwcb_t::start ( ) [virtual]

4.15.3.2 int io_tcp_fwcb_t::process ( mha_wave_t ∗sIn, mha_wave_t ∗&sOut )[virtual]

Parameters

sIn The input sound data just received from TCP.

sOut A pointer to output sound data. Will point to the output sound data storage when the

callback ﬁnishes.

Returns

Status, an error number from the signal processing callback. If this is != 0, then the

connection should be closed.

4.15.3.3 void io_tcp_fwcb_t::set_errnos ( int proc_err, int io_err )[virtual]

See also

stop (p. 105)

Parameters

proc_err The error number from the

See also

process (p. 104) callback.

Parameters

io_err The error number from the io library itself.

4.16 io_tcp_parser_t Class Reference 105

4.15.3.4 void io_tcp_fwcb_t::stop ( ) [virtual]

Uses the error numbers set previously with

See also

set_errnos (p. 104).

4.15.4 Member Data Documentation

4.15.4.1 IOProcessEvent_t io_tcp_fwcb_t::proc_event [private]

4.15.4.2 IOStartedEvent_t io_tcp_fwcb_t::start_event [private]

Called when a new TCP connection is established or the user issues the start command while

there is a connection.

4.15.4.3 IOStoppedEvent_t io_tcp_fwcb_t::stop_event [private]

Called when the connection is closed.

4.15.4.4 void∗io_tcp_fwcb_t::proc_handle [private]

4.15.4.5 int io_tcp_fwcb_t::proc_err [private]

MHAIOTCP is notiﬁed by the server when the connection has been taken down, and calls

See also

stop (p. 105) from that callback. Within stop (p. 105), these error numbers are read again

and transmitted to the framework.

4.16 io_tcp_parser_t Class Reference

The parser interface of the IOTCP library.

Inheritance diagram for io_tcp_parser_t:

io_tcp_parser_t

MHAParser::parser_t

MHAParser::base_t

106 CONTENTS

Public Member Functions

• virtual const std::string & get_local_address () const

Read parser variable local_address, this is the address of the network interface that should

listen for incoming connections.

• virtual unsigned short get_local_port () const

Read parser variable local_port, this is the TCP port that should be used for incoming connec-

tions.

• virtual void set_local_port (unsigned short port)

Set parser variable local_port.

• virtual bool get_server_port_open () const

Return the status of the server port as it is known to the parser.

• virtual void set_server_port_open (bool open)

Inform the parser of the current status of the server socket.

• virtual bool get_connected () const

Return the parser's knowledge concerning wether there currently exists an established sound

data TCP connection or not.

• virtual void set_connected (bool connected)

Inform the parser about the existance of a sound data connection.

• virtual void set_new_peer (unsigned short port, const std::string &host)

Set parser monitor variables peer_port and peer_address, and calls set_connected(true).

•io_tcp_parser_t ()

Constructor initializes parser variables.

• virtual ∼io_tcp_parser_t ()

Do-nothing destructor.

Private Attributes

•MHAParser::string_t local_address

Lets the user set the local network interface to listen on.

•MHAParser::int_t local_port

Lets the user choose the local tcp port to listen on.

•MHAParser::int_mon_t server_port_open

Indicates wether the TCP server socket is currently open.

•MHAParser::int_mon_t connected

Indicator if there currently is a sound data connection over TCP.

•MHAParser::string_mon_t peer_address

Display the ip address of the currently connected sound data client.

•MHAParser::int_mon_t peer_port

Display the tcp port used by the current sound data client.

•MHAParser::string_t debug_ﬁlename

ﬁlename to write debugging info to (if non-empty)

• FILE ∗debug_ﬁle

ﬁle handle to write debugging info to

4.16 io_tcp_parser_t Class Reference 107

Additional Inherited Members

4.16.1 Detailed Description

4.16.2 Constructor & Destructor Documentation

4.16.2.1 io_tcp_parser_t::io_tcp_parser_t ( )

4.16.2.2 virtual io_tcp_parser_t::∼io_tcp_parser_t ( ) [inline],[virtual]

4.16.3 Member Function Documentation

4.16.3.1 virtual const std::string& io_tcp_parser_t::get_local_address ( ) const [inline],

[virtual]

Returns

A string containing the address of the local interface as it was set by the user.

4.16.3.2 unsigned short io_tcp_parser_t::get_local_port ( ) const [virtual]

Returns

The local tcp port to listen on as it was chosen by the user. The port number is between

MIN_TCP_PORT and MAX_TCP_PORT.

4.16.3.3 void io_tcp_parser_t::set_local_port ( unsigned short port )[virtual]

This is needed when it was set to 0 before: In this case, the OS chooses a free port for the TCP

server socket, and the port that it chose has to be published to the user via the parser interface.

Parameters

port The TCP port number that is currently used. In the range [MIN_TCP_PORT,

MAX_TCP_PORT], excluding 0.

Precondition

get_local_port() (p. 107) currently returns 0.

4.16.3.4 bool io_tcp_parser_t::get_server_port_open ( ) const [virtual]

108 CONTENTS

Returns

false after initialization, or the value most recently set via

See also

set_server_port_open (p. 108).

4.16.3.5 void io_tcp_parser_t::set_server_port_open ( bool open )[virtual]

Parameters

open Indicates wether the server socket has just been opened or closed.

Precondition

open may only have the value true if get_server_port_open() (p. 107) currently returns

false.

Postcondition

See also

get_server_port_open (p. 107) returns the value (p. 48) of open.

4.16.3.6 bool io_tcp_parser_t::get_connected ( ) const [virtual]

Returns

false after initialization, or the value most recently set via

See also

set_connected (p. 108).

4.16.3.7 void io_tcp_parser_t::set_connected ( bool connected )[virtual]

Parameters

connected Indicates wether there currently is a connection or not.

4.16 io_tcp_parser_t Class Reference 109

Precondition

connected must not have the same value that is currently returned by

See also

get_connected (p. 108).

Postcondition

See also

get_connected (p. 108) returns the value (p. 48) of open.

4.16.3.8 void io_tcp_parser_t::set_new_peer ( unsigned short port, const std::string & host )

[virtual]

This method should be called when a new connection is established.

Parameters

port The TCP port number used by the peer.

host The Internet host where the peer is located.

Precondition

See also

get_connected (p. 108) currently returns false.

Postcondition

See also

get_connected (p. 108) returns true.

110 CONTENTS

4.16.4 Member Data Documentation

4.16.4.1 MHAParser::string_t io_tcp_parser_t::local_address [private]

4.16.4.2 MHAParser::int_t io_tcp_parser_t::local_port [private]

4.16.4.3 MHAParser::int_mon_t io_tcp_parser_t::server_port_open [private]

4.16.4.4 MHAParser::int_mon_t io_tcp_parser_t::connected [private]

4.16.4.5 MHAParser::string_mon_t io_tcp_parser_t::peer_address [private]

4.16.4.6 MHAParser::int_mon_t io_tcp_parser_t::peer_port [private]

4.17 io_tcp_sound_t Class Reference

Sound data handling of io tcp library.

Classes

• union ﬂoat_union

This union helps in conversion of ﬂoats from host byte order to network byte order and back

again.

Public Member Functions

•io_tcp_sound_t (int fragsize, ﬂoat samplerate)

Initialize sound data handling.

• virtual ∼io_tcp_sound_t ()

Do-nothing destructor.

• virtual void prepare (int num_inchannels, int num_outchannels)

Called during prepare, sets number of audio channels and allocates sound data storage.

• virtual void release ()

Called during release.

• virtual int chunkbytes_in () const

Number of bytes that constitute one input sound chunk.

• virtual std::string header () const

Create the tcp sound header lines.

• virtual mha_wave_t ∗ntoh (const std::string &data)

Copy data received from tcp into mha_wave_t (p. 154) structure.

• virtual std::string hton (const mha_wave_t ∗s_out)

Copy sound data from the output sound structure to a string.

4.17 io_tcp_sound_t Class Reference 111

Static Private Member Functions

• static void check_sound_data_type ()

Check if mha_real_t is a usable 32-bit ﬂoating point type.

Private Attributes

• int fragsize

Number of sound samples in each channel expected and returned from processing callback.

• ﬂoat samplerate

Sampling rate.

• int num_inchannels

Number of input channels.

•MHASignal::waveform_t ∗s_in

Storage for input signal.

4.17.1 Constructor & Destructor Documentation

4.17.1.1 io_tcp_sound_t::io_tcp_sound_t ( int fragsize, ﬂoat samplerate )

Checks sound data type by calling

See also

check_sound_data_type (p. 111).

Parameters

fragsize Number of sound samples in each channel expected and returned from

processing callback.

samplerate Number of samples per second in each channel.

4.17.2 Member Function Documentation

4.17.2.1 void io_tcp_sound_t::check_sound_data_type ( ) [static],[private]

Exceptions

MHA_Error (p. 132)if mha_real_t is not compatible to 32-bit ﬂoat.

112 CONTENTS

4.17.2.2 void io_tcp_sound_t::prepare ( int num_inchannels, int num_outchannels )[virtual]

Parameters

num_inchannels Number of input audio channels.

num_outchannels Number of output audio channels.

4.17.2.3 void io_tcp_sound_t::release ( ) [virtual]

Deletes sound data storage.

4.17.2.4 int io_tcp_sound_t::chunkbytes_in ( ) const [virtual]

Returns

Number of bytes to read from TCP connection before invoking signal processing.

4.17.2.5 std::string io_tcp_sound_t::header ( ) const [virtual]

4.17.2.6 mha_wave_t ∗io_tcp_sound_t::ntoh ( const std::string & data )[virtual]

Doing network-to-host byte order swapping in the process.

Parameters

data One chunk (

See also

chunkbytes_in (p. 112)) of sound data to process.

Returns

Pointer to the sound data storage.

4.17.2.7 std::string io_tcp_sound_t::hton ( const mha_wave_t ∗s_out )[virtual]

Doing host-to-network byte order swapping while at it.

Parameters

s_out Pointer to the storage of the sound to put out.

4.18 io_tcp_sound_t::ﬂoat_union Union Reference 113

Returns

The sound data in network byte order.

4.17.3 Member Data Documentation

4.17.3.1 int io_tcp_sound_t::fragsize [private]

4.17.3.2 ﬂoat io_tcp_sound_t::samplerate [private]

Number of samples per second in each channel.

4.17.3.3 int io_tcp_sound_t::num_inchannels [private]

Number of channels expected from and returned by signal processing callback.

4.17.3.4 MHASignal::waveform_t∗io_tcp_sound_t::s_in [private]

4.18 io_tcp_sound_t::ﬂoat_union Union Reference

This union helps in conversion of ﬂoats from host byte order to network byte order and back

again.

4.18.1 Detailed Description

4.19 io_tcp_t Class Reference

The tcp sound io library.

Public Member Functions

• void prepare (int num_inchannels, int num_outchannels)

Allocate server socket and start thread waiting for sound data exchange.

• void start ()

Call frameworks start callback if there is a sound data connection at the moment.

• void stop ()

Close the current connection if there is one.

• void release ()

Close the current connection and close the server socket.

• virtual void accept_loop ()

IO thread executes this method.

• virtual void connection_loop (MHA_TCP::Connection ∗c)

IO thread executes this method for each connection.

• virtual void parse (const char ∗cmd, char ∗retval, unsigned int len)

Parser interface.

114 CONTENTS

4.19.1 Detailed Description

4.19.2 Member Function Documentation

4.19.2.1 void io_tcp_t::prepare ( int num_inchannels, int num_outchannels )

prepare opens the tcp server socket and starts the io thread that listens for audio data on the

tcp socket after doing some sanity checks

4.19.2.2 void io_tcp_t::start ( )

4.19.2.3 void io_tcp_t::stop ( )

stop IO thread

4.19.2.4 void io_tcp_t::release ( )

Stop IO thread and close server socket.

4.19.2.5 void io_tcp_t::accept_loop ( ) [virtual]

4.19.2.6 void io_tcp_t::connection_loop ( MHA_TCP::Connection ∗c)[virtual]

Parameters

cpointer to connection. connection_loop deletes connection before exiting.

4.19.2.7 virtual void io_tcp_t::parse ( const char ∗cmd, char ∗retval, unsigned int len )[inline],

[virtual]

4.20 MHA_AC::ac2matrix_t Class Reference

Copy AC variable to a matrix.

4.20 MHA_AC::ac2matrix_t Class Reference 115

Inheritance diagram for MHA_AC::ac2matrix_t:

MHA_AC::ac2matrix_t

MHASignal::matrix_t

MHASignal::uint_vector_t

Public Member Functions

•ac2matrix_t (algo_comm_t ac, const std::string &name)

Constructor.

• void update ()

Update contents of the matrix from the AC space.

• const std::string & getname () const

Return name of AC variable/matrix.

• const std::string & getusername () const

Return user speciﬁed name of AC variable/matrix.

• void insert (algo_comm_t ac)

Insert matrix into an AC space (other than source AC space)

4.20.1 Detailed Description

This class constructs a matrix of same size as an AC variable and can copy the AC variable to

itself. The update() (p. 116) function is real-time safe.

4.20.2 Constructor & Destructor Documentation

4.20.2.1 MHA_AC::ac2matrix_t::ac2matrix_t ( algo_comm_t ac, const std::string & name )

Parameters

ac AC handle

name Name of AC variable to be copied

116 CONTENTS

4.20.3 Member Function Documentation

4.20.3.1 void MHA_AC::ac2matrix_t::update ( )

This function is real-time safe. The copy operation performance is of the order of the number

of elements in the matrix.

4.20.3.2 void MHA_AC::ac2matrix_t::insert ( algo_comm_t ac )

Parameters

ac AC space handle to insert data

Note

The AC variable data buffer points to the data of the matrix. Modiﬁcations of the AC

variable directly modify the data of the matrix; after deletion of the matrix, the data buffer

is invalid.

4.21 MHA_AC::acspace2matrix_t Class Reference

Copy all or a subset of all numeric AC variables into an array of matrixes.

Public Member Functions

•acspace2matrix_t (algo_comm_t ac, const std::vector<std::string >&names)

Constructor.

•acspace2matrix_t (const MHA_AC::acspace2matrix_t &src)

Constructor with initialization from an instance.

•MHA_AC::acspace2matrix_t &operator= (const MHA_AC::acspace2matrix_t &src)

Copy all contents (deep copy).

•MHA_AC::ac2matrix_t &operator[ ] (unsigned int k)

Access operator.

• const MHA_AC::ac2matrix_t &operator[ ] (unsigned int k) const

Constant access operator.

• void update ()

Update function.

• unsigned int size () const

Number of matrixes in AC space.

• unsigned int frame () const

Actual frame number.

• void insert (algo_comm_t ac)

Insert AC space copy into an AC space (other than source AC space)

4.21 MHA_AC::acspace2matrix_t Class Reference 117

4.21.1 Constructor & Destructor Documentation

4.21.1.1 MHA_AC::acspace2matrix_t::acspace2matrix_t ( algo_comm_t ac, const std::vector<

std::string >&names )

Scan all given AC variables and allocate corresponding matrixes.

Parameters

ac AC handle.

names Names of AC variables, or empty for all.

4.21.1.2 MHA_AC::acspace2matrix_t::acspace2matrix_t ( const MHA_AC::acspace2matrix_t &src

)

Parameters

src Instance to be copied.

4.21.2 Member Function Documentation

4.21.2.1 MHA_AC::acspace2matrix_t & MHA_AC::acspace2matrix_t::operator= ( const

MHA_AC::acspace2matrix_t &src )

Parameters

src Array of matrixes to be copied.

4.21.2.2 MHA_AC::ac2matrix_t& MHA_AC::acspace2matrix_t::operator[ ] ( unsigned int k)

[inline]

Parameters

kindex into array; should not exceed size() (p. 116)-1.

Return values

Reference to matrix.

4.21.2.3 const MHA_AC::ac2matrix_t& MHA_AC::acspace2matrix_t::operator[ ] ( unsigned int k)

const [inline]

118 CONTENTS

Parameters

kindex into array; should not exceed size() (p. 116)-1.

Return values

Constant reference to matrix.

4.21.2.4 void MHA_AC::acspace2matrix_t::update ( ) [inline]

This function updates all matrixes from their corresponding AC variables. It can be called from

the MHA Framework prepare function or in the processing callback.

4.21.2.5 void MHA_AC::acspace2matrix_t::insert ( algo_comm_t ac )

Parameters

ac AC space handle to insert data

4.22 MHA_AC::double_t Class Reference

Insert a double precision ﬂoating point variable into the AC space.

Public Attributes

• double data

Floating point value variable.

4.22.1 Detailed Description

The variable is automatically removed on destruction.

4.23 MHA_AC::ﬂoat_t Class Reference

Insert a ﬂoat point variable into the AC space.

Public Member Functions

•ﬂoat_t (algo_comm_t, std::string, ﬂoat=0)

Constructor.

4.24 MHA_AC::int_t Class Reference 119

Public Attributes

• ﬂoat data

Floating point value variable.

4.23.1 Detailed Description

The variable is automatically removed on destruction.

4.24 MHA_AC::int_t Class Reference

Insert a integer variable into the AC space.

Public Attributes

• int data

Integer value variable.

4.24.1 Detailed Description

The variable is automatically removed on destruction.

4.25 MHA_AC::spectrum_t Class Reference

Insert a MHASignal::spectrum_t (p. 261) class into the AC space.

Inheritance diagram for MHA_AC::spectrum_t:

MHA_AC::spectrum_t

MHASignal::spectrum_t

mha_spec_t

120 CONTENTS

Public Member Functions

•spectrum_t (algo_comm_t ac, std::string name, unsigned int bins, unsigned int chan-

nels, bool insert_now)

Create the AC variable.

• void insert ()

Insert AC variable into AC space.

Additional Inherited Members

4.25.1 Detailed Description

The variable is automatically removed on destruction.

4.25.2 Constructor & Destructor Documentation

4.25.2.1 MHA_AC::spectrum_t::spectrum_t ( algo_comm_t ac, std::string name, unsigned int bins,

unsigned int channels, bool insert_now )

Parameters

ac AC handle

name Name of variable in AC space

bins Number of FFT bins in the waveform_t (p. 120) class

channels Number of audio channels in the waveform_t (p. 120) class

insert_now Insert implicitely in the constructor (true) or explicitely in the insert() (p. 120)

function (false)

4.26 MHA_AC::waveform_t Class Reference

Insert a MHASignal::waveform_t (p. 268) class into the AC space.

4.26 MHA_AC::waveform_t Class Reference 121

Inheritance diagram for MHA_AC::waveform_t:

MHA_AC::waveform_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

•waveform_t (algo_comm_t ac, std::string name, unsigned int frames, unsigned int

channels, bool insert_now)

Create the AC variable.

• void insert ()

Insert AC variable into AC space.

Additional Inherited Members

4.26.1 Detailed Description

The variable is automatically removed on destruction.

4.26.2 Constructor & Destructor Documentation

4.26.2.1 MHA_AC::waveform_t::waveform_t ( algo_comm_t ac, std::string name, unsigned int

frames, unsigned int channels, bool insert_now )

Parameters

ac AC handle

name Name of variable in AC space

frames Number of frames in the waveform_t (p. 120) class

channels Number of audio channels in the waveform_t (p. 120) class

insert_now Insert implicitely in the constructor (true) or explicitely in the insert() (p. 121)

function (false)

122 CONTENTS

4.27 mha_audio_descriptor_t Struct Reference

Description of an audio fragment (planned as a replacement of mhaconﬁg_t (p. 155)).

Public Attributes

• unsigned int n_samples

Number of samples.

• unsigned int n_channels

Number of audio channels.

• unsigned int n_freqs

Number of frequency bands.

• unsigned int is_complex

Flag about sample type.

•mha_real_t dt

Time distance between samples (only equidistant samples allowed)

•mha_real_t ∗cf

Center frequencies of frequency bands.

•mha_real_t ∗chdir

Hint on source direction of channel, values below zero is left, values above zero is right, zero

means unknown.

4.28 mha_audio_t Struct Reference

An audio fragment in the openMHA (planned as a replacement of mha_wave_t (p. 154) and

mha_spec_t (p. 141)).

Public Attributes

•mha_audio_descriptor_t descriptor

Dimension and description of the data.

•mha_real_t ∗rdata

Data pointer if ﬂag mha_audio_descriptor_t::is_complex (p. 122) is unset.

•mha_complex_t ∗cdata

Data pointer if ﬂag mha_audio_descriptor_t::is_complex (p. 122) is set.

4.28.1 Detailed Description

The data alignment is (t0, c0, f0),(t0, c0, f1),...,(t0, c0, ffreqs),(t0, c1, f0), . . .. This allows a di-

rect cast of the current mha_wave_t (p. 154) and mha_spec_t (p. 141) data pointers into cor-

responding mha_audio_t (p. 122) objects.

4.29 mha_channel_info_t Struct Reference 123

4.28.2 Member Data Documentation

4.28.2.1 mha_audio_descriptor_t mha_audio_t::descriptor

4.28.2.2 mha_real_t∗mha_audio_t::rdata

4.28.2.3 mha_complex_t∗mha_audio_t::cdata

4.29 mha_channel_info_t Struct Reference

Channel information structure.

Public Attributes

• int id

channel id

• char idstr [32]

channel id

• unsigned int side

side (left/right)

•mha_direction_t dir

source direction

•mha_real_t peaklevel

Peak level corresponds to this SPL (dB) level.

4.30 mha_complex_t Struct Reference

Type for complex ﬂoating point values.

Public Attributes

•mha_real_t re

Real part.

•mha_real_t im

Imaginary part.

124 CONTENTS

4.30.1 Member Data Documentation

4.30.1.1 mha_real_t mha_complex_t::re

4.30.1.2 mha_real_t mha_complex_t::im

4.31 mha_dblbuf_t<FIFO >Class Template Reference

The doublebuffer adapts blocksizes between an outer process, which provides input data and

takes output data, and an inner process, which processes the input signal and generates output

data using a different block size than the outer process.

Public Types

• typedef FIFO::value_type value_type

The datatype exchanged by the FIFO and this doublebuffer.

Public Member Functions

•mha_dblbuf_t (unsigned outer_size, unsigned inner_size, unsigned delay, unsigned

input_channels, unsigned output_channels, const value_type &delay_data)

Constructor creates FIFOs with speciﬁed delay.

• virtual void process (const value_type ∗input_signal, value_type ∗output_signal, un-

signed count)

The outer process has to call this method to propagate the input signal to the inner process,

and receives back the output signal.

• virtual void input (value_type ∗input_signal)

The inner process has to call this method to receive its input signal.

• virtual void output (const value_type ∗output_signal)

The outer process has to call this method to deliver its output signal.

Private Attributes

• unsigned outer_size

The block size used by the outer process.

• unsigned inner_size

The block size used by the inner process.

• unsigned delay

The delay introduced by bidirectional buffer size adaptation.

• unsigned ﬁfo_size

The size of each of the FIFOs.

• unsigned input_channels

The number of input channels.

4.31 mha_dblbuf_t<FIFO >Class Template Reference 125

• unsigned output_channels

The number of output channels.

• FIFO input_ﬁfo

The FIFO for transporting the input signal from the outer process to the inner process.

• FIFO output_ﬁfo

The FIFO for transporting the output signal from the inner process to the outer process.

•MHA_Error ∗inner_error

Owned copy of exception to be thrown in inner thread.

•MHA_Error ∗outer_error

Owned copy of exception to be thrown in outer thread.

4.31.1 Detailed Description

template<class FIFO>

class mha_dblbuf_t<FIFO >

This class introduces the channels concept. Input and output may have different channel

counts.

4.31.2 Constructor & Destructor Documentation

4.31.2.1 template<class FIFO >mha_dblbuf_t<FIFO >::mha_dblbuf_t ( unsigned outer_size,

unsigned inner_size, unsigned delay, unsigned input_channels, unsigned output_channels,

const value_type &delay_data )

Warning

The doublebuffer may block or raise an exception if the delay is too small. To avoid this,

the delay should be

delay >= (inner_size −gcd(inner_size, outer_size))

Parameters

outer_size The block size used by the outer process.

inner_size The block size used by the inner process.

delay The total delay

input_channels Number of input channels

output_channels Number of output channels

delay_data The delay consists of copies of this value.

126 CONTENTS

4.31.3 Member Function Documentation

4.31.3.1 template<class FIFO >void mha_dblbuf_t<FIFO >::process ( const value_type ∗

input_signal, value_type ∗output_signal, unsigned count )[virtual]

Parameters

input_signal Pointer to the input signal array.

output_signal Pointer to the output signal array.

count The number of data instances provided and expected, lower or equal to

inner_size given to constructor.

Exceptions

MHA_Error (p. 132)When count is >outer_size as given to constructor or the underlying

ﬁfo implementation detects an error.

4.31.3.2 template<class FIFO >void mha_dblbuf_t<FIFO >::input ( value_type ∗input_signal )

[virtual]

Parameters

input_signal Array where the doublebuffer can store the signal.

Exceptions

MHA_Error (p. 132)When the underlying ﬁfo implementation detects an error.

4.31.3.3 template<class FIFO >void mha_dblbuf_t<FIFO >::output ( const value_type ∗

output_signal )[virtual]

Parameters

output_signal Array from which doublebuffer reads outputsignal.

Exceptions

MHA_Error (p. 132)When the underlying ﬁfo implementation detects an error.

4.31.4 Member Data Documentation

4.31.4.1 template<class FIFO >unsigned mha_dblbuf_t<FIFO >::outer_size [private]

4.32 mha_direction_t Struct Reference 127

4.31.4.2 template<class FIFO >unsigned mha_dblbuf_t<FIFO >::inner_size [private]

4.31.4.3 template<class FIFO >unsigned mha_dblbuf_t<FIFO >::delay [private]

4.31.4.4 template<class FIFO >FIFO mha_dblbuf_t<FIFO >::input_ﬁfo [private]

4.31.4.5 template<class FIFO >FIFO mha_dblbuf_t<FIFO >::output_ﬁfo [private]

4.32 mha_direction_t Struct Reference

Channel source direction structure.

Public Attributes

•mha_real_t azimuth

azimuth in radiants

•mha_real_t elevation

elevation in radiants

•mha_real_t distance

distance in meters

4.33 mha_drifter_ﬁfo_t<T>Class Template Reference

A FIFO class for blocksize adaptation without Synchronization.

Inheritance diagram for mha_drifter_ﬁfo_t<T>:

mha_drifter_fifo_t< T >

mha_fifo_t< T >

128 CONTENTS

Public Member Functions

• virtual void write (const T ∗data, unsigned count)

write data to ﬁfo

• virtual void read (T ∗buf, unsigned count)

Read data from ﬁfo.

• virtual unsigned get_ﬁll_count () const

Return ﬁll_count, adding mha_drifter_ﬁfo_t<T>::startup_zeros (p. 131) to the number of

samples actually in the ﬁfo's buffer.

• virtual unsigned get_available_space () const

Return available space, subtracting number of mha_drifter_ﬁfo_t<T>::startup_zeros

(p. 131) from the available_space actually present in the ﬁfo's buffer.

• virtual unsigned get_des_ﬁll_count () const

The desired ﬁll count of this ﬁfo.

• virtual unsigned get_min_ﬁll_count () const

The minimum ﬁll count of this ﬁfo.

• virtual void stop ()

Called by mha_drifter_ﬁfo_t<T>::read (p. 130) or mha_drifter_ﬁfo_t<T>::write (p. 129)

when their xrun in succession counter exceeds its limit.

• virtual void starting ()

Called by mha_drifter_ﬁfo_t<T>::read (p. 130) or mha_drifter_ﬁfo_t<T>::write (p. 129)

when the respective ﬂag (mha_drifter_ﬁfo_t<T>::reader_started (p. 131) or mha_drifter←-

_ﬁfo_t<T>::writer_started (p. 131)) is about to be toggled from false to true.

•mha_drifter_ﬁfo_t (unsigned min_ﬁll_count, unsigned desired_ﬁll_count, unsigned

max_ﬁll_count)

Create drifter FIFO.

•mha_drifter_ﬁfo_t (unsigned min_ﬁll_count, unsigned desired_ﬁll_count, unsigned

max_ﬁll_count, const T &t)

Create drifter FIFO where all (initially unused) copies of T are initialized as copies of t.

Private Attributes

• const unsigned minimum_ﬁll_count

The minimum ﬁll count of this ﬁfo.

• const unsigned desired_ﬁll_count

The desired ﬁll count of the ﬁfo.

• bool writer_started

Flag set to true when write is called the ﬁrst time.

• bool reader_started

Flag set to true when read is called for the ﬁrst time.

• unsigned writer_xruns_total

The number of xruns seen by the writer since object instantiation.

• unsigned reader_xruns_total

The number of xruns seen by the reader since object instantiation.

• unsigned writer_xruns_since_start

The number of xruns seen by the writer since the last start of processing.

4.33 mha_drifter_ﬁfo_t<T>Class Template Reference 129

• unsigned reader_xruns_since_start

The number of xruns seen by the reader since the last start of processing.

• unsigned writer_xruns_in_succession

The number of xruns seen by the writer in succession.

• unsigned reader_xruns_in_succession

The number of xruns seen by the reader in succession.

• unsigned maximum_writer_xruns_in_succession_before_stop

A limit to the number of xruns seen in succession during write before the data transmission

through the FIFO is stopped.

• unsigned maximum_reader_xruns_in_succession_before_stop

A limit to the number of xruns seen in succession during read before the data transmission

through the FIFO is stopped.

•mha_ﬁfo_t<T>::value_type null_data

The value used in place of missing data.

• unsigned startup_zeros

When processing starts, that is when both mha_drifter_ﬁfo_t<T>::reader_started (p. 131)

and mha_drifter_ﬁfo_t<T>::writer_started (p. 131) are true, then ﬁrst mha_drifter_ﬁfo←-

_t<T>::desired_ﬁll_count (p. 131) instances of mha_drifter_ﬁfo_t<T>::null_data (p. 129)

are delivered to the reader.

Additional Inherited Members

4.33.1 Detailed Description

template<class T>

class mha_drifter_ﬁfo_t<T>

Features: delay concept (desired, minimum and maximum delay), drifting support by throwing

away data or inserting zeroes.

4.33.2 Constructor & Destructor Documentation

4.33.2.1 template<class T >mha_drifter_ﬁfo_t<T>::mha_drifter_ﬁfo_t ( unsigned

min_ﬁll_count, unsigned desired_ﬁll_count, unsigned max_ﬁll_count )

4.33.3 Member Function Documentation

4.33.3.1 template<class T >void mha_drifter_ﬁfo_t<T>::write ( const T ∗data, unsigned count )

[virtual]

Sets writer_started (p. 131) to true.

When processing has started, i.e. both reader_started (p. 131) and writer_started (p. 131)

are true, write speciﬁed ammount of data to the ﬁfo. If there is not enough space available, then

the exceeding data is lost and the writer xrun counters are increased.

Processing is stopped when writer_xruns_in_succession (p. 131) exceeds maximum_←-

writer_xruns_in_succession_before_stop (p. 131).

130 CONTENTS

Parameters

data Pointer to source data.

count Number of instances to copy

Reimplemented from mha_ﬁfo_t<T>(p. 136).

4.33.3.2 template<class T >void mha_drifter_ﬁfo_t<T>::read ( T ∗buf, unsigned count )

[virtual]

Sets reader_started (p. 131) to true.

When processing has started, i.e. both reader_started (p. 131) and writer_started (p. 131)

are true, then read speciﬁed ammount of data from the ﬁfo. As long as startup_zeros (p. 131)

is >0, null_data (p. 129) is delivered to the reader and startup_zeros (p. 131) is diminished.

Only when startup_zeros (p. 131) has reached 0, data is actually read from the ﬁfo's buffer.

If the read would cause the ﬁfo's ﬁll count to drop below minimum_ﬁll_count (p. 131), then

only so much data are read that minimum_ﬁll_count (p. 131) entries remain in the ﬁfo, the

missing data is replaced with null_data (p. 129), and the reader xrun counters are increased.

Processing is stopped when reader_xruns_in_succession (p. 131) exceeds maximum_←-

reader_xruns_in_succession_before_stop (p. 131).

Parameters

buf Pointer to the target buffer

count Number of instances to copy

Reimplemented from mha_ﬁfo_t<T>(p. 137).

4.33.3.3 template<class T >unsigned mha_drifter_ﬁfo_t<T>::get_ﬁll_count ( ) const

[virtual]

Reimplemented from mha_ﬁfo_t<T>(p. 135).

4.33.3.4 template<class T >unsigned mha_drifter_ﬁfo_t<T>::get_available_space ( ) const

[virtual]

TODO: uncertain if this is a good idea.

Reimplemented from mha_ﬁfo_t<T>(p. 135).

4.33.3.5 template<class T >void mha_drifter_ﬁfo_t<T>::stop ( ) [virtual]

Called by read (p. 130) or write (p. 129) when their xrun in succession counter exceeds its limit.

May also be called explicitly.

4.33 mha_drifter_ﬁfo_t<T>Class Template Reference 131

4.33.3.6 template<class T >void mha_drifter_ﬁfo_t<T>::starting ( ) [virtual]

The ﬁfo's buffer is emptied, this method resets startup_zeros (p. 131) to desired_ﬁll_count

(p. 131), and it also resets reader_xruns_since_start (p. 129) and writer_xruns_since_start

(p. 131) to 0.

4.33.4 Member Data Documentation

4.33.4.1 template<class T >const unsigned mha_drifter_ﬁfo_t<T>::minimum_ﬁll_count

[private]

4.33.4.2 template<class T >const unsigned mha_drifter_ﬁfo_t<T>::desired_ﬁll_count

[private]

The ﬁfo is initialized with this ammount of data when data transmission starts.

4.33.4.3 template<class T >bool mha_drifter_ﬁfo_t<T>::writer_started [private]

4.33.4.4 template<class T >bool mha_drifter_ﬁfo_t<T>::reader_started [private]

4.33.4.5 template<class T >unsigned mha_drifter_ﬁfo_t<T>::writer_xruns_since_start

[private]

4.33.4.6 template<class T >unsigned mha_drifter_ﬁfo_t<T>::writer_xruns_in_succession

[private]

Reset to 0 every time a write succeeds without xrun.

4.33.4.7 template<class T >unsigned mha_drifter_ﬁfo_t<T>::reader_xruns_in_succession

[private]

Reset to 0 every time a read succeeds without xrun.

4.33.4.8 template<class T >unsigned mha_drifter_ﬁfo_t<T>::maximum_writer_xruns_in_←-

succession_before_stop [private]

4.33.4.9 template<class T >unsigned mha_drifter_ﬁfo_t<T>::maximum_reader_xruns_in_←-

succession_before_stop [private]

4.33.4.10 template<class T >unsigned mha_drifter_ﬁfo_t<T>::startup_zeros [private]

These null_data (p. 129) instances are not transmitted through the ﬁfo because ﬁlling the ﬁfo

with enough null_data (p. 129) might not be realtime safe and this ﬁlling has to be initiated by

starting (p. 131) or stop (p. 130) (this implementation: starting (p. 131)) which are be called

with realtime constraints.

132 CONTENTS

4.34 MHA_Error Class Reference

Error reporting exception class.

Inherits exception.

Public Member Functions

•MHA_Error (const char ∗ﬁle, int line, const char ∗fmt,...)

Create an instance of a MHA_Error (p. 132).

• const char ∗get_msg () const

Return the error message without source position.

• const char ∗get_longmsg () const

Return the error message with source position.

• const char ∗what () const throw ()

overwrite std::execption::what()

4.34.1 Detailed Description

This class is used for error handling in the openMHA. It is used by the openMHA kernel and by

the openMHA toolbox library. Please note that exceptions should not be used accross ANSI-C

interfaces. It is necessary to catch exceptions within the library.

The MHA_Error (p. 132) class holds source ﬁle name, line number and an error message.

4.34.2 Constructor & Destructor Documentation

4.34.2.1 MHA_Error::MHA_Error ( const char ∗s_ﬁle, int l, const char ∗fmt, ... )

Parameters

s_ﬁle source ﬁle name (FILE)

lsource line (LINE)

fmt format string for error message (as in printf)

4.35 mha_ﬁfo_lw_t<T>Class Template Reference

This FIFO uses locks to synchronize access.

4.35 mha_ﬁfo_lw_t<T>Class Template Reference 133

Inheritance diagram for mha_ﬁfo_lw_t<T>:

mha_fifo_lw_t< T >

mha_fifo_t< T >

Public Member Functions

• virtual void write (const T ∗data, unsigned count)

write speciﬁed ammount of data to the ﬁfo.

• virtual void read (T ∗buf, unsigned count)

read data from ﬁfo.

•mha_ﬁfo_lw_t (unsigned max_ﬁll_count)

Create FIFO with ﬁxed buffer size.

• virtual ∼mha_ﬁfo_lw_t ()

release synchronization object

• virtual void set_error (unsigned index, MHA_Error ∗error)

Process waiting for more data or space should bail out, throwing this error.

Private Attributes

•mha_ﬁfo_thread_platform_t ∗sync

platform speciﬁc thread synchronization

•MHA_Error ∗error [2]

If waiting for synchronization should be aborted then exception to be thrown by reader process

(index 0) or writer process (index 1) has to be placed here.

Additional Inherited Members

4.35.1 Detailed Description

template<class T>

class mha_ﬁfo_lw_t<T>

Reading and writing can block until the operation can be executed.

134 CONTENTS

4.35.2 Member Function Documentation

4.35.2.1 template<class T >void mha_ﬁfo_lw_t<T>::write ( const T ∗data, unsigned count )

[virtual]

If there is not enough space, then wait for more space.

Parameters

data Pointer to source data.

count Number of instances to copy.

Exceptions

MHA_Error (p. 132)when detecting a deadlock situation.

Reimplemented from mha_ﬁfo_t<T>(p. 136).

4.35.2.2 template<class T >void mha_ﬁfo_lw_t<T>::read ( T ∗buf, unsigned count )

[virtual]

If there is not enough data, then wait for more data.

Parameters

buf Pointer to the target buffer.

count Number of instances to copy.

Exceptions

MHA_Error (p. 132)when detecting a deadlock situation.

Reimplemented from mha_ﬁfo_t<T>(p. 137).

4.35.2.3 template<class T >void mha_ﬁfo_lw_t<T>::set_error ( unsigned index, MHA_Error ∗

error )[virtual]

Parameters

index Use 0 for terminating reader, 1 for terminating writer.

error MHA_Error (p. 132) to be thrown

4.35.3 Member Data Documentation

4.36 mha_ﬁfo_t<T>Class Template Reference 135

4.35.3.1 template<class T >MHA_Error∗mha_ﬁfo_lw_t<T>::error[2] [private]

4.36 mha_ﬁfo_t<T>Class Template Reference

A FIFO class for blocksize adaptation Synchronization: None.

Inheritance diagram for mha_ﬁfo_t<T>:

mha_fifo_t< T >

mha_drifter_fifo_t< T > mha_fifo_lw_t< T >

Public Types

• typedef T value_type

The data type exchanged by this ﬁfo.

Public Member Functions

• virtual void write (const T ∗data, unsigned count)

write speciﬁed ammount of data to the ﬁfo.

• virtual void read (T ∗buf, unsigned count)

read data from ﬁfo

• virtual unsigned get_ﬁll_count () const

Read-only access to ﬁll_count.

• virtual unsigned get_available_space () const

Read-only access to available_space.

• virtual unsigned get_max_ﬁll_count () const

The capacity of this ﬁfo.

•mha_ﬁfo_t (unsigned max_ﬁll_count)

Create FIFO with ﬁxed buffer size.

•mha_ﬁfo_t (unsigned max_ﬁll_count, const T &t)

Create FIFO with ﬁxed buffer size, where all (initially unused) copies of T are initialized as

copies of t.

•mha_ﬁfo_t (const mha_ﬁfo_t &src)

Copy constructor.

• virtual ∼mha_ﬁfo_t ()

Destroy FIFO.

•mha_ﬁfo_t<T>&operator= (const mha_ﬁfo_t<T>&)

Assignment operator.

136 CONTENTS

Protected Member Functions

• void clear ()

Empty the ﬁfo at once.

Private Attributes

• const unsigned max_ﬁll_count

The maximum ﬁll count of this FIFO.

• T ∗buf

The memory allocated to store the data.

• T ∗write_ptr

points to location where to write next

• const T ∗read_ptr

points to location where to read next

• bool buf_uses_placement_new

wether buf was allocated using placement new or array new.

4.36.1 Detailed Description

template<class T>

class mha_ﬁfo_t<T>

Use external synchronisation or synchronization in inheriting class.

4.36.2 Member Function Documentation

4.36.2.1 template<class T >void mha_ﬁfo_t<T>::write ( const T ∗data, unsigned count )

[virtual]

Parameters

data Pointer to source data.

count Number of instances to copy

Exceptions

MHA_Error (p. 132)when there is not enough space available.

Reimplemented in mha_ﬁfo_lw_t<T>(p. 134), and mha_drifter_ﬁfo_t<T>(p. 129).

4.37 mha_ﬁfo_thread_guard_t Class Reference 137

4.36.2.2 template<class T >void mha_ﬁfo_t<T>::read ( T ∗buf, unsigned count )[virtual]

Parameters

buf Pointer to the target buffer

count Number of instances to copy

Exceptions

MHA_Error (p. 132)when there is not enough data available.

Reimplemented in mha_ﬁfo_lw_t<T>(p. 134), and mha_drifter_ﬁfo_t<T>(p. 130).

4.36.2.3 template<class T >void mha_ﬁfo_t<T>::clear ( ) [inline],[protected]

Should be called by the reader, or when the reader is inactive.

4.36.3 Member Data Documentation

4.36.3.1 template<class T >const unsigned mha_ﬁfo_t<T>::max_ﬁll_count [private]

4.36.3.2 template<class T >T∗mha_ﬁfo_t<T>::buf [private]

max_ﬁll_count + 1 locations are allocated: At least one location is always unused, because we

have max_ﬁll_count + 1 possible ﬁllcounts [0:max_ﬁll_count] that we need to distinguish.

4.36.3.3 template<class T >bool mha_ﬁfo_t<T>::buf_uses_placement_new [private]

4.37 mha_ﬁfo_thread_guard_t Class Reference

Simple Mutex Guard Class.

4.38 mha_ﬁfo_thread_platform_t Class Reference

Abstract base class for synchronizing multithreaded (producer/consumer) ﬁfo operations.

Inherited by mha_ﬁfo_posix_threads_t.

138 CONTENTS

Public Member Functions

• virtual void aquire_mutex ()=0

Calling thread waits until it aquires the lock.

• virtual void release_mutex ()=0

Calling thread releases the lock.

• virtual void wait_for_decrease ()=0

Calling producer thread must own the lock.

• virtual void wait_for_increase ()=0

Calling consumer thread must own the lock.

• virtual void increment ()=0

To be called by producer thread after producing.

• virtual void decrement ()=0

To be called by consumer thread after consuming.

• virtual ∼mha_ﬁfo_thread_platform_t ()

Make destructor virtual.

•mha_ﬁfo_thread_platform_t ()

Make default constructor accessible.

4.38.1 Detailed Description

Works only with single producer and single consumer.

4.38.2 Member Function Documentation

4.38.2.1 virtual void mha_ﬁfo_thread_platform_t::aquire_mutex ( ) [pure virtual]

Must not be called when the lock is already aquired.

4.38.2.2 virtual void mha_ﬁfo_thread_platform_t::release_mutex ( ) [pure virtual]

May only be called when lock is owned.

4.38.2.3 virtual void mha_ﬁfo_thread_platform_t::wait_for_decrease ( ) [pure virtual]

Method releases lock, and waits for consumer thread to call decrease(). Then reaquires lock

and returns

4.38.2.4 virtual void mha_ﬁfo_thread_platform_t::wait_for_increase ( ) [pure virtual]

Method releases lock, and waits for producer thread to call increase(). Then reaquires lock and

returns

4.39 mha_rt_ﬁfo_element_t<T>Class Template Reference 139

4.38.2.5 virtual void mha_ﬁfo_thread_platform_t::increment ( ) [pure virtual]

Producer thread needs to own the lock to call this method.

4.38.2.6 virtual void mha_ﬁfo_thread_platform_t::decrement ( ) [pure virtual]

Consumer thread needs to own the lock to call this method.

4.39 mha_rt_ﬁfo_element_t<T>Class Template Reference

Object wrapper for mha_rt_ﬁfo_t (p. 140).

Public Member Functions

•mha_rt_ﬁfo_element_t (T ∗data)

Constructor.

Public Attributes

•mha_rt_ﬁfo_element_t<T>∗next

Pointer to next ﬁfo element. NULL for the last (newest) ﬁfo element.

• bool abandonned

Indicates that this element will no longer be used and may be deleted.

• T ∗data

Pointer to user data.

4.39.1 Constructor & Destructor Documentation

4.39.1.1 template<class T >mha_rt_ﬁfo_element_t<T>::mha_rt_ﬁfo_element_t ( T ∗data )

[inline]

This element assumes ownership of user data.

Parameters

data User data. Has to be allocated on the heap with standard operator new, because it

will be deleted in this element's destructor.

140 CONTENTS

4.40 mha_rt_ﬁfo_t<T>Class Template Reference

Template class for thread safe, half real time safe ﬁfo without explixit locks.

Public Member Functions

•mha_rt_ﬁfo_t ()

Construct empty ﬁfo.

•∼mha_rt_ﬁfo_t ()

Destructor will delete all data currently in the ﬁfo.

• T ∗poll ()

Retrieve the latest element in the Fifo.

• T ∗poll_1 ()

Retrieve the next element in the Fifo, if there is one, and mark the previous element as aban-

donned.

• void push (T ∗data)

Add element to the Fifo.

Private Member Functions

• void remove_abandonned ()

Deletes abandonned elements.

• void remove_all ()

Deletes all elements.

Private Attributes

•mha_rt_ﬁfo_element_t<T>∗root

The ﬁrst element in the ﬁfo. Deleting elements starts here.

•mha_rt_ﬁfo_element_t<T>∗current

The element most recently returned by poll (p. 141) or poll_1 (p. 141).

4.40.1 Detailed Description

template<class T>

class mha_rt_ﬁfo_t<T>

Reading from this ﬁfo is realtime safe, writing to it is not. This ﬁfo is designed for objects that

were constructed on the heap. It assumes ownership of these objects and calls delete on them

when they are no longer used. Objects remain inside the Fifo while being used by the reader.

A new ﬁfo element is inserted by using push (p. 141). The push operation is not real time safe,

it allocates and deallocates memory. The latest element is retrieved by calling poll (p. 141).

This operation will skip ﬁfo elements if more than one push (p. 141) has been occured since

the last poll. To avoid skipping, call the poll_1 (p. 141) operation instead.

4.41 mha_spec_t Struct Reference 141

4.40.2 Member Function Documentation

4.40.2.1 template<class T >T∗mha_rt_ﬁfo_t<T>::poll ( ) [inline]

Will skip ﬁfo elements if more than one element has been added since last poll invocation. Will

return the same element as on last call if no elements have been added in the mean time.

Marks former elements as abandonned.

Returns

The latest element in this Fifo. Returns NULL if the Fifo is empty.

4.40.2.2 template<class T >T∗mha_rt_ﬁfo_t<T>::poll_1 ( ) [inline]

Else, if there is no newer element, returns the same element as on last poll() (p. 141) or poll_1()

(p. 141) invocation.

Returns

The next element in this Fifo, if there is one, or the same as before. Returns NULL if the

Fifo is empty.

4.40.2.3 template<class T >void mha_rt_ﬁfo_t<T>::push ( T ∗data )[inline]

Deletes abandonned elements in the ﬁfo.

Parameters

data The new user data to place at the end of the ﬁfo. After this invocation, the ﬁfo is the

owner of this object and will delete it when it is no longer used. data must have been

allocated on the heap with standard operator new.

4.40.3 Member Data Documentation

4.40.3.1 template<class T >mha_rt_ﬁfo_element_t<T>∗mha_rt_ﬁfo_t<T>::current

[private]

Searching for new elements starts here.

4.41 mha_spec_t Struct Reference

Spectrum signal structure.

142 CONTENTS

Inheritance diagram for mha_spec_t:

mha_spec_t

MHASignal::spectrum_t

MHA_AC::spectrum_t

Public Attributes

•mha_complex_t ∗buf

signal buffer

• unsigned int num_channels

number of channels

• unsigned int num_frames

number of frames in each channel

•mha_channel_info_t ∗channel_info

detailed channel description

4.41.1 Detailed Description

This structure contains the short time fourier transform output of the windowed input signal.

The member num_frames describes the number of frequency bins in each channel. For an

even FFT length N, this is N/2+1. With odd FFT lengths, it is (N+ 1)/2. The imaginary part

of the ﬁrst bin is zero. For even FFT lengths, also the imaginary part at the Nyquist frequency

is zero.

buf[k].re Re(0) Re(1) Re(2) Re(3) Re(4) ... Re(n/2−1) Re(n/2)

buf[k].im Im(1) Im(2) Im(3) Im(4) ... Im(n/2−1)

k0 1 2 3 4 n/2−1 n/2

Figure 4 Data order of FFT spectrum.

4.42 MHA_TCP::Async_Notify Class Reference 143

4.42 MHA_TCP::Async_Notify Class Reference

Portable Multiplexable cross-thread notiﬁcation.

Inheritance diagram for MHA_TCP::Async_Notify:

MHA_TCP::Async_Notify

MHA_TCP::Wakeup_Event

Additional Inherited Members

4.43 MHA_TCP::Client Class Reference

A portable class for a tcp client connections.

Inheritance diagram for MHA_TCP::Client:

MHA_TCP::Client

MHA_TCP::Connection

Public Member Functions

•Client (const std::string &host, unsigned short port)

Constructor connects to host, port via TCP.

•Client (const std::string &host, unsigned short port, Timeout_Watcher &timeout_←-

watcher)

Constructor connects to host, port via TCP, using a timeout.

144 CONTENTS

Additional Inherited Members

4.43.1 Constructor & Destructor Documentation

4.43.1.1 Client::Client ( const std::string & host, unsigned short port )

Parameters

host The hostname of the TCP Server.

port The port or the TCP Server.

4.43.1.2 Client::Client ( const std::string & host, unsigned short port, Timeout_Watcher &

timeout_watcher )

Parameters

host The hostname of the TCP Server.

port The port or the TCP Server.

timeout_watcher an Event watcher that implements a timeout.

4.44 MHA_TCP::Connection Class Reference

Connection (p. 144) handles Communication between client and server, is used on both sides.

Inheritance diagram for MHA_TCP::Connection:

MHA_TCP::Connection

MHA_TCP::Client

Public Member Functions

• std::string get_peer_address ()

Get peer's IP Address.

• unsigned short get_peer_port ()

4.44 MHA_TCP::Connection Class Reference 145

Get peer's TCP port.

• SOCKET get_fd () const

Return the (protected) ﬁle descriptor of the connection.

• virtual ∼Connection ()

Destructor closes the underlying ﬁle descriptor.

• bool eof ()

Checks if the peer has closed the connection.

• bool can_read_line (char delim= '\n')

Checks if a full line of text has arrived by now.

• bool can_read_bytes (unsigned howmany)

Checks if the speciﬁed ammount of data can be read.

• std::string read_line (char delim= '\n')

Reads a single line of data from the socket.

• std::string read_bytes (unsigned howmany)

Reads the speciﬁed ammount of dat from the socket.

• void try_write (const std::string &data="")

Adds data to the internal "outgoing" buffer, and then tries to write as much data from that buffer

to the socket as possible without blocking.

• void write (const std::string &data="")

Adds data to the internal "outgoing" buffer, and then writes that that buffer to the socket, re-

gardless of blocking.

• bool needs_write ()

Checks if the internal "outgoing" buffer contains data.

• unsigned buffered_incoming_bytes () const

Returns the number of bytes in the internal "incoming" buffer.

• unsigned buffered_outgoing_bytes () const

Returns the number of bytes in the internal "outgoing" buffer.

Protected Member Functions

•Connection (SOCKET _fd)

Create a connection instance from a socket ﬁledescriptor.

Protected Attributes

• SOCKET fd

The ﬁle descriptor of the TCP Socket.

146 CONTENTS

Private Member Functions

• void init_peer_data ()

determine peer address and port

• bool can_sysread ()

Determine wether at least 1 byte can be read without blocking.

• bool can_syswrite ()

Determine wether at least 1 byte can be written without blocking.

• std::string sysread (unsigned bytes)

Call the system's read function and try to read bytes.

• std::string syswrite (const std::string &data)

Call the system's write function and try to write all characters in the string data.

4.44.1 Constructor & Destructor Documentation

4.44.1.1 MHA_TCP::Connection::Connection ( SOCKET _fd )[protected]

Parameters

←-

_←-

The ﬁle descriptor of the TCP Socket. This ﬁle descriptor is closed again in the

destructor.

Exceptions

MHA_Error (p. 132)If the ﬁle descriptor is <0.

4.44.2 Member Function Documentation

4.44.2.1 std::string Connection::sysread ( unsigned bytes )[private]

This will block in a situation where can_sysread returns false.

Parameters

bytes The desired number of characters.

Returns

The characters read from the socket. The result may have fewer characters than speciﬁed

by bytes. If the result is an empty string, then the socket has been closed by the peer.

4.44 MHA_TCP::Connection Class Reference 147

4.44.2.2 std::string Connection::syswrite ( const std::string & data )[private]

May write fewer characters, but will at least write one character.

Parameters

data A string of characters to write to the socket.

Returns

The rest of the characters that have not yet been written.

4.44.2.3 SOCKET MHA_TCP::Connection::get_fd ( ) const [inline]

Will be required for SSL.

4.44.2.4 bool Connection::eof ( )

As a side effect, this method ﬁlls the internal "incoming" buffer if it was empty and the socket is

readable and not eof.

4.44.2.5 bool Connection::can_read_line ( char delim = '\n' )

This method reads data from the socket into the internal "incoming" buffer if it can be done

without blocking.

Parameters

delim The line delimiter.

Returns

true if at least one full line of text is present in the internal buffer after this method call,

false otherwise.

4.44.2.6 bool Connection::can_read_bytes ( unsigned howmany )

This method reads data from the socket into an internal "incoming" buffer if it can be done

without blocking.

Parameters

howmany The number of bytes that the caller wants to have checked.

148 CONTENTS

Returns

true if at least the speciﬁed ammount of data is present in the internal buffer after this

method call, false otherwise

4.44.2.7 std::string Connection::read_line ( char delim = '\n' )

Blocks if necessary.

Parameters

delim The line delimiter.

Returns

The string of characters in this line, including the trailing delimiter. The delimiter may be

missing if the last line before EOF does not have a delimiter.

4.44.2.8 std::string Connection::read_bytes ( unsigned howmany )

Blocks if necessary.

Parameters

howmany The number of bytes to read.

Returns

The string of characters read. The string may be shorter if EOF is encountered.

4.44.2.9 void Connection::try_write ( const std::string & data = "" )

Parameters

data data to send over the socket.

4.44.2.10 void Connection::write ( const std::string & data = "" )

Parameters

data data to send over the socket.

4.45 MHA_TCP::Event_Watcher Class Reference 149

4.45 MHA_TCP::Event_Watcher Class Reference

OS-independent event watcher, uses select on Unix and WaitForMultipleObjects on Windows.

Inheritance diagram for MHA_TCP::Event_Watcher:

MHA_TCP::Event_Watcher

MHA_TCP::Timeout_Watcher

Public Member Functions

• void observe (Wakeup_Event ∗event)

Add an event to this observer.

• void ignore (Wakeup_Event ∗event)

Remove an event from this observer.

•std::set<Wakeup_Event ∗>wait ()

\ Wait for some event to occur.

Private Attributes

•std::set<Wakeup_Event ∗>events

The list of events to watch.

4.45.1 Member Function Documentation

4.45.1.1 void Event_Watcher::observe ( Wakeup_Event ∗event )

4.45.1.2 std::set<Wakeup_Event ∗>Event_Watcher::wait ( )

Return all events that are ready

150 CONTENTS

4.45.2 Member Data Documentation

4.45.2.1 std::set<Wakeup_Event∗>MHA_TCP::Event_Watcher::events [private]

4.46 MHA_TCP::Sockread_Event Class Reference

Watch socket for incoming data.

Inheritance diagram for MHA_TCP::Sockread_Event:

MHA_TCP::Sockread_Event

MHA_TCP::Wakeup_Event

Public Member Functions

•Sockread_Event (SOCKET s)

Set socket to watch for.

4.46.1 Constructor & Destructor Documentation

4.46.1.1 MHA_TCP::Sockread_Event::Sockread_Event ( SOCKET s)

Parameters

sThe socket to observe incoming data on.

4.47 MHA_TCP::Thread Class Reference

A very simple class for portable threads.

Public Types

• typedef void ∗(∗thr_f) (void ∗)

The thread function signature to use with this class.

4.47 MHA_TCP::Thread Class Reference 151

Public Member Functions

•Thread (thr_f func, void ∗arg=0)

Constructor starts a new thread.

• virtual ∼Thread ()

The destructor should only be called when the Thread (p. 150) is ﬁnished.

• virtual void run ()

The internal method that delegated the new thread to the registered Thread (p. 150) function.

Public Attributes

•Async_Notify thread_ﬁnish_event

Event will be triggered when the thread exits.

• enum MHA_TCP::Thread:: { ... } state

The current state of the thread.

•thr_f thread_func

The thread function that the client has registered.

• void ∗thread_arg

The argument that the client wants to be handed through to the thread function.

•MHA_Error ∗error

The MHA_Error (p. 132) that caused the thread to abort, if any.

Protected Member Functions

•Thread ()

Default constructor may only be used by derived classes that want to start the thread them-

selves.

Protected Attributes

• void ∗arg

The argument for the client's thread function.

• void ∗return_value

The return value from the client's thread function is stored here When that function returns.

Private Attributes

• pthread_t thread_handle

The posix thread handle.

• pthread_attr_t thread_attr

The posix thread attribute structure.

152 CONTENTS

4.47.1 Member Typedef Documentation

4.47.1.1 typedef void∗(∗MHA_TCP::Thread::thr_f) (void ∗)

Derive from this class and call protected standard constructor to start threads differently.

4.47.2 Constructor & Destructor Documentation

4.47.2.1 Thread::Thread ( Thread::thr_f func, void ∗arg = 0)

Parameters

func The function to be executed by the thread.

arg The argument given to pass to the thread function.

4.47.2.2 Thread::∼Thread ( ) [virtual]

There is preliminary support for forceful thread cancellation in the destructor, but probably not

very robust or portable..

4.47.3 Member Data Documentation

4.47.3.1 pthread_attr_t MHA_TCP::Thread::thread_attr [private]

Required for starting a thread in detached state. Detachment is required to eliminate the need

for joining this thread.

4.47.3.2 void∗MHA_TCP::Thread::arg [protected]

4.47.3.3 void∗MHA_TCP::Thread::thread_arg

4.48 MHA_TCP::Timeout_Watcher Class Reference

OS-independent event watcher with internal ﬁxed-end-time timeout.

4.49 MHA_TCP::Wakeup_Event Class Reference 153

Inheritance diagram for MHA_TCP::Timeout_Watcher:

MHA_TCP::Timeout_Watcher

MHA_TCP::Event_Watcher

Additional Inherited Members

4.49 MHA_TCP::Wakeup_Event Class Reference

A base class for asynchronous wakeup events.

Inheritance diagram for MHA_TCP::Wakeup_Event:

MHA_TCP::Wakeup_Event

MHA_TCP::Async_Notify MHA_TCP::Sockread_Event

Public Member Functions

•Wakeup_Event ()

Event Constructor.

• virtual void observed_by (Event_Watcher ∗observer)

Called by the Event_Watcher (p. 149) when this event is added to its list of observed events.

• virtual void ignored_by (Event_Watcher ∗observer)

Called by the Event_Watcher (p. 149) when this event is removed from its list of observed

events.

• virtual ∼Wakeup_Event ()

Destructor deregisters from observers.

154 CONTENTS

• virtual OS_EVENT_TYPE get_os_event ()

Get necessary information for the Event Watcher.

• virtual void reset ()

For pure notiﬁcation events, reset the "signalled" status.

• virtual bool status ()

Query wether the event is in signalled state now.

Private Attributes

•std::set<class Event_Watcher ∗>observers

A list of all Event_Watcher (p. 149) instances that this Wakeup_Event (p. 153) is observed by

(stored here for proper deregistering).

4.49.1 Constructor & Destructor Documentation

4.49.1.1 Wakeup_Event::Wakeup_Event ( )

The new event has invalid state.

4.50 mha_wave_t Struct Reference

Waveform signal structure.

Inheritance diagram for mha_wave_t:

mha_wave_t MHASignal::waveform_t

MHA_AC::waveform_t

MHAFilter::o1_ar_filter_t

MHAOvlFilter::fftfb_t

MHASignal::async_rmslevel_t

MHASignal::loop_wavefragment_t

MHASignal::ringbuffer_t

MHASignal::schroeder_t

MHAWindow::base_t

MHAFilter::o1flt_lowpass_t

MHAFilter::o1flt_maxtrack_t

MHAFilter::o1flt_mintrack_t

MHAOvlFilter::overlap

_save_filterbank_t

MHAWindow::fun_t

MHAWindow::user_t

MHAWindow::bartlett_t

MHAWindow::blackman_t

MHAWindow::hamming_t

MHAWindow::hanning_t

MHAWindow::rect_t

4.51 mhaconﬁg_t Struct Reference 155

Public Attributes

•mha_real_t ∗buf

signal buffer

• unsigned int num_channels

number of channels

• unsigned int num_frames

number of frames in each channel

•mha_channel_info_t ∗channel_info

detailed channel description

4.50.1 Detailed Description

This structure contains one fragment of a waveform signal. The member num_frames describes

the number of audio samples in each audio channel.

The ﬁeld channel_info must be an array of num_channels entries or NULL.

4.51 mhaconﬁg_t Struct Reference

MHA prepare conﬁguration structure.

Public Attributes

• unsigned int channels

Number of audio channels.

• unsigned int domain

Signal domain (MHA_WAVEFORM or MHA_SPECTRUM)

• unsigned int fragsize

Fragment size of waveform data.

• unsigned int wndlen

Window length of spectral data.

• unsigned int fftlen

FFT length of spectral data.

•mha_real_t srate

Sampling rate in Hz.

4.51.1 Detailed Description

This structure contains information about channel number and domain for input and output

signals of a openMHA Plugin. Each plugin can change any of these parameters, e.g. by

resampling of the signal. The only limitation is that the callback frequency is ﬁxed (except for

the plugins db and dbasync).

156 CONTENTS

4.52 MHAEvents::emitter_t Class Reference

Class for emitting openMHA events.

Public Member Functions

• void operator() ()

Emit an event without parameter.

• void operator() (const std::string &)

Emit an event with string parameter.

• void operator() (const std::string &, unsigned int, unsigned int)

Emit an event with string parameter and two unsigned int parameters.

4.52.1 Detailed Description

Use the template claas MHAEvents::patchbay_t (p. 156) for connecting to an emitter.

4.53 MHAEvents::patchbay_t<receiver_t >Class Template Reference

Patchbay which connects any event emitter with any member function of the parameter class.

Public Member Functions

• void connect (emitter_t ∗, receiver_t ∗, void(receiver_t::∗)())

Connect a receiver member function void (receiver_t::∗)() with an event emitter.

• void connect (emitter_t ∗, receiver_t ∗, void(receiver_t::∗)(const std::string &))

Connect a receiver member function void (receiver_t::∗)(const std::string&) with an event emit-

ter.

4.53.1 Detailed Description

template<class receiver_t>

class MHAEvents::patchbay_t<receiver_t >

The connections created by the connect() (p. 156) function are hold until the destructor is

called. To avoid access to invalid function pointers, it is required to destruct the patchbay

before the receiver, usually by declaring the patchbay as a member of the receiver.

The receiver can be any claas or structure; the event callback can be either a member function

without arguments or with const std::string& argument.

4.53.2 Member Function Documentation

4.53.2.1 template<class receiver_t>void MHAEvents::patchbay_t<receiver_t >::connect (

emitter_t ∗e, receiver_t ∗r, void(receiver_t::∗)() rfun )

Create a connection.

The connection is removed when the patchbay is destructed.

4.54 MHAFilter::adapt_ﬁlter_t Class Reference 157

Parameters

ePointer to an event emitter

rPointer to the receiver

rfun Pointer to a member function of the receiver class

4.53.2.2 template<class receiver_t>void MHAEvents::patchbay_t<receiver_t >::connect (

emitter_t ∗e, receiver_t ∗r, void(receiver_t::∗)(const std::string &) rfun )

Create a connection.

The connection is removed when the patchbay is destructed.

Parameters

ePointer to an event emitter

rPointer to the receiver

rfun Pointer to a member function of the receiver class

4.54 MHAFilter::adapt_ﬁlter_t Class Reference

Adaptive ﬁlter.

Inheritance diagram for MHAFilter::adapt_ﬁlter_t:

MHAFilter::adapt_filter_t

MHAParser::parser_tMHAParser::base_t

MHAPlugin::config_t

< adapt_filter_state_t >

MHAPlugin::config_t

< adapt_filter_param_t >

Additional Inherited Members

4.55 MHAFilter::blockprocessing_polyphase_resampling_t Class Reference

A class that does polyphase resampling and takes into account block processing.

158 CONTENTS

Public Member Functions

•blockprocessing_polyphase_resampling_t (ﬂoat source_srate, unsigned source_←-

fragsize, ﬂoat target_srate, unsigned target_fragsize, ﬂoat nyquist_ratio, ﬂoat irslen, un-

signed nchannels, bool add_delay)

Contructs a polyphase resampling ﬁlter that can be used for blockprocessing with the given

parameters.

• void write (mha_wave_t &signal)

Write signal to the ringbuffer.

• void read (mha_wave_t &signal)

Read resampled signal.

• bool can_read () const

Checks if the resampling ring buffer can produce another output signal block.

4.55.1 Detailed Description

4.55.2 Constructor & Destructor Documentation

4.55.2.1 MHAFilter::blockprocessing_polyphase_resampling_t::blockprocessing_polyphase_←-

resampling_t ( ﬂoat source_srate, unsigned source_fragsize, ﬂoat target_srate, unsigned

target_fragsize, ﬂoat nyquist_ratio, ﬂoat irslen, unsigned nchannels, bool add_delay )

Parameters

source_srate Source sampling rate / Hz

source_fragsize Fragment size of incoming audio blocks / frames at source_srate

target_srate Target sampling rate / Hz

target_fragsize Fragment size of produced audio blocks / frames at target_srate

nyquist_ratio Low pass ﬁlter cutoff frequency relative to the nyquist frequency of the

smaller of the two sampling rates. Example values: 0.8, 0.9

irslen Impulse response length used for low pass ﬁltering / s

nchannels Number of audio channels

add_delay To avoid underruns, a delay is generally necessary for round trip block

size adaptations. It is only necessary to add this delay to one of the two

resampling chains. Set this parameter to true for the ﬁrst resampling

object of a round trip pair. It will add the necessary delay, and calculate

the size of the ring buffer appropriately, When set to false, only the

ringbuffer size will be set sufﬁciently.

4.55.3 Member Function Documentation

4.55.3.1 void MHAFilter::blockprocessing_polyphase_resampling_t::write ( mha_wave_t &signal )

4.56 MHAFilter::complex_bandpass_t Class Reference 159

Parameters

signal input signal in original sampling rate

Exceptions

MHA_Error (p. 132)Raises exception if there is not enough room, if the number of

channels does not match, or if the number of frames is not equal to

the number speciﬁed in the constructor

4.55.3.2 void MHAFilter::blockprocessing_polyphase_resampling_t::read ( mha_wave_t &signal )

Will perform the resampling and remove no longer needed samples from the input buffer.

Parameters

signal buffer to write the resampled signal to.

Exceptions

MHA_Error (p. 132)Raises exception if there is not enough input signal, if the number of

channels of frames does not match.

4.56 MHAFilter::complex_bandpass_t Class Reference

Complex bandpass ﬁlter.

Public Member Functions

•complex_bandpass_t (std::vector<mha_complex_t >A, std::vector<mha_←-

complex_t >B)

Constructor with ﬁlter coefﬁcients (one per channel)

• void set_weights (std::vector<mha_complex_t >new_B)

Allow to modify the input weights at a later stage.

• void ﬁlter (const mha_wave_t &X, mha_spec_t &Y)

Filter method for real value input.

• void ﬁlter (const mha_wave_t &X, mha_wave_t &Yre, mha_wave_t &Yim)

Filter method for real value input.

• void ﬁlter (const mha_spec_t &X, mha_spec_t &Y)

Filter method for complex value input.

• void ﬁlter (const mha_wave_t &Xre, const mha_wave_t &Xim, mha_wave_t &Yre,

mha_wave_t &Yim)

Filter method for complex value input.

160 CONTENTS

4.56.1 Constructor & Destructor Documentation

4.56.1.1 MHAFilter::complex_bandpass_t::complex_bandpass_t ( std::vector<mha_complex_t >A,

std::vector<mha_complex_t >B)

Parameters

Acomplex ﬁlter coefﬁcients, one per band

Bcomplex weights

4.57 MHAFilter::diff_t Class Reference

Differentiator class (non-normalized)

Inheritance diagram for MHAFilter::diff_t:

MHAFilter::diff_t

MHAFilter::filter_t

Additional Inherited Members

4.58 MHAFilter::fftﬁlter_t Class Reference

FFT based FIR ﬁlter implementation.

Inheritance diagram for MHAFilter::fftﬁlter_t:

MHAFilter::fftfilter_t

MHAFilter::resampling

_filter_t

4.58 MHAFilter::fftﬁlter_t Class Reference 161

Public Member Functions

•fftﬁlter_t (unsigned int fragsize, unsigned int channels, unsigned int fftlen)

Constructor.

• void update_coeffs (const mha_wave_t ∗pwIRS)

Update the set of coefﬁcients.

• void ﬁlter (const mha_wave_t ∗pwIn, mha_wave_t ∗∗ppwOut, const mha_wave_t ∗pw←-

IRS)

Apply ﬁlter with changing coefﬁcients to a waveform fragment.

• void ﬁlter (const mha_wave_t ∗pwIn, mha_wave_t ∗∗ppwOut)

Apply ﬁlter to waveform fragment, without changing the coefﬁcients.

• void ﬁlter (const mha_wave_t ∗pwIn, mha_wave_t ∗∗ppwOut, const mha_spec_t ∗ps←-

Weights)

Apply ﬁlter with changing coefﬁcients to a waveform fragment.

4.58.1 Detailed Description

The maximal number of coefﬁcients can be FFT length - fragsize + 1.

4.58.2 Constructor & Destructor Documentation

4.58.2.1 MHAFilter::fftﬁlter_t::fftﬁlter_t ( unsigned int fragsize, unsigned int channels, unsigned int

fftlen )

Parameters

fragsize Number of frames expected in input signal (each cycle).

channels Number of channels expected in input signal.

fftlen FFT length of ﬁlter.

4.58.3 Member Function Documentation

4.58.3.1 void MHAFilter::fftﬁlter_t::update_coeffs ( const mha_wave_t ∗pwIRS )

Parameters

pwIRS Coefﬁcients structure

Note

The number of channels in h must match the number of channels given in the constructor.

The ﬁlter length is limited to fftlen-fragsize+1 (longer IRS will be shortened).

162 CONTENTS

4.58.3.2 void MHAFilter::fftﬁlter_t::ﬁlter ( const mha_wave_t ∗pwIn, mha_wave_t ∗∗ ppwOut,

const mha_wave_t ∗pwIRS )

Parameters

pw←-

Input signal pointer.

Return values

ppwOut Pointer to output signal pointer, will be set to a valid signal.

Parameters

pwIRS Pointer to FIR coefﬁcients structure.

4.58.3.3 void MHAFilter::fftﬁlter_t::ﬁlter ( const mha_wave_t ∗pwIn, mha_wave_t ∗∗ ppwOut )

Parameters

pw←-

Input signal pointer.

Return values

ppwOut Pointer to output signal pointer, will be set to a valid signal

4.58.3.4 void MHAFilter::fftﬁlter_t::ﬁlter ( const mha_wave_t ∗pwIn, mha_wave_t ∗∗ ppwOut,

const mha_spec_t ∗psWeights )

Parameters

pw←-

Input signal pointer.

Return values

ppwOut Pointer to output signal pointer, will be set to a valid signal.

Parameters

psWeights Pointer to ﬁlter weights structure.

4.59 MHAFilter::fftﬁlterbank_t Class Reference 163

4.59 MHAFilter::fftﬁlterbank_t Class Reference

FFT based FIR ﬁlterbank implementation.

Inheritance diagram for MHAFilter::fftﬁlterbank_t:

MHAFilter::fftfilterbank_t

MHAOvlFilter::overlap

_save_filterbank_t

Public Member Functions

•fftﬁlterbank_t (unsigned int fragsize, unsigned int inputchannels, unsigned int ﬁrchan-

nels, unsigned int fftlen)

Constructor.

• void update_coeffs (const mha_wave_t ∗h)

Update the set of coefﬁcients.

• void ﬁlter (const mha_wave_t ∗s_in, mha_wave_t ∗∗s_out, const mha_wave_t ∗h)

Apply ﬁlter with changing coefﬁcients to a waveform fragment.

• void ﬁlter (const mha_wave_t ∗s_in, mha_wave_t ∗∗s_out)

Apply ﬁlter to waveform fragment, without changing the coefﬁcients.

• const mha_wave_t ∗get_irs () const

Return the current IRS.

4.59.1 Detailed Description

This class convolves n input channels with m ﬁlter coefﬁcient sets and returns n∗m output

channels.

The maximal number of coefﬁcients can be FFT length - fragsize + 1.

4.59.2 Constructor & Destructor Documentation

4.59.2.1 MHAFilter::fftﬁlterbank_t::fftﬁlterbank_t ( unsigned int fragsize, unsigned int inputchannels,

unsigned int ﬁrchannels, unsigned int fftlen )

164 CONTENTS

Parameters

fragsize Number of frames expected in input signal (each cycle).

inputchannels Number of channels expected in input signal.

ﬁrchannels Number of channels expected in FIR ﬁlter coefﬁcients (= number of bands).

fftlen FFT length of ﬁlter.

The number of output channels is inputchannels∗ﬁrchannels.

4.59.3 Member Function Documentation

4.59.3.1 void MHAFilter::fftﬁlterbank_t::update_coeffs ( const mha_wave_t ∗h)

Parameters

hCoefﬁcients structure

Note

The number of channels in h must match the number of channels given in the constructor,

and the number of frames can not be more than fftlen-fragsize+1.

4.59.3.2 void MHAFilter::fftﬁlterbank_t::ﬁlter ( const mha_wave_t ∗s_in, mha_wave_t ∗∗ s_out,

const mha_wave_t ∗h)

Parameters

s←-

_in

Input signal pointer.

Return values

s_out Pointer to output signal pointer, will be set to a valid signal

Parameters

hFIR coefﬁcients

4.59.3.3 void MHAFilter::fftﬁlterbank_t::ﬁlter ( const mha_wave_t ∗s_in, mha_wave_t ∗∗ s_out )

4.60 MHAFilter::ﬁlter_t Class Reference 165

Parameters

s←-

_in

Input signal pointer.

Return values

s_out Pointer to output signal pointer, will be set to a valid signal

4.60 MHAFilter::ﬁlter_t Class Reference

Generic IIR ﬁlter class.

Inheritance diagram for MHAFilter::ﬁlter_t:

MHAFilter::filter_t

MHAFilter::diff_t

Public Member Functions

•ﬁlter_t (unsigned int ch, unsigned int lena, unsigned int lenb)

Constructor.

•ﬁlter_t (unsigned int ch, const std::vector<mha_real_t >&vA, const std::vector<mha←-

_real_t >&vB)

Constructor with initialization of coefﬁcients.

• void ﬁlter (mha_wave_t ∗out, const mha_wave_t ∗in)

Filter all channels in a waveform structure.

• void ﬁlter (mha_real_t ∗dest, const mha_real_t ∗src, unsigned int dframes, unsigned

int frame_dist, unsigned int channel_dist, unsigned int channel_begin, unsigned int

channel_end)

Filter parts of a waveform structure.

•mha_real_t ﬁlter (mha_real_t x, unsigned int ch)

Filter one sample.

• unsigned int get_len_A () const

Return length of recursive coefﬁcients.

• unsigned int get_len_B () const

Return length of non-recursive coefﬁcients.

166 CONTENTS

Public Attributes

• double ∗A

Pointer to recursive coefﬁcients.

• double ∗B

Pointer to non-recursive coefﬁcients.

4.60.1 Detailed Description

This class implements a generic multichannel IIR ﬁlter. It is realized as direct form II. It can

work on any ﬂoat array or on mha_wave_t (p. 154) structs. The ﬁlter coefﬁcients can be directly

accessed.

4.60.2 Constructor & Destructor Documentation

4.60.2.1 MHAFilter::ﬁlter_t::ﬁlter_t ( unsigned int ch, unsigned int lena, unsigned int lenb )

Parameters

ch Number of channels

lena Number of recursive coefﬁcients

lenb Number of non-recursive coefﬁcients

4.60.2.2 MHAFilter::ﬁlter_t::ﬁlter_t ( unsigned int ch, const std::vector<mha_real_t >&vA, const

std::vector<mha_real_t >&vB )

Parameters

ch Number of channels.

vA Recursive coefﬁcients.

vB Non-recursive coefﬁcients.

4.60.3 Member Function Documentation

4.60.3.1 void MHAFilter::ﬁlter_t::ﬁlter ( mha_wave_t ∗out, const mha_wave_t ∗in )

Parameters

out Output signal

in Input signal

4.61 MHAFilter::gamma_ﬂt_t Class Reference 167

4.60.3.2 void MHAFilter::ﬁlter_t::ﬁlter ( mha_real_t ∗dest, const mha_real_t ∗src, unsigned int

dframes, unsigned int frame_dist, unsigned int channel_dist, unsigned int channel_begin,

unsigned int channel_end )

Parameters

dest Output signal.

src Input signal.

dframes Number of frames to be ﬁltered.

frame_dist Index distance between frames of one channel

channel_dist Index distance between audio channels

channel_begin Number of ﬁrst channel to be processed

channel_end Number of last channel to be processed

4.60.3.3 mha_real_t MHAFilter::ﬁlter_t::ﬁlter ( mha_real_t x, unsigned int ch )

Parameters

xInput value

ch Channel number to use in ﬁlter state

4.61 MHAFilter::gamma_ﬂt_t Class Reference

Class for gammatone ﬁlter.

Public Member Functions

•gamma_ﬂt_t (std::vector<mha_real_t >cf, std::vector<mha_real_t >bw, mha_real←-

_t srate, unsigned int order)

Constructor.

• void operator() (mha_wave_t &X, mha_spec_t &Y)

Filter method.

• void operator() (mha_wave_t &X, mha_wave_t &Yre, mha_wave_t &Yim)

Filter method.

• void operator() (mha_wave_t &Yre, mha_wave_t &Yim, unsigned int stage)

Filter method for speciﬁc stage.

4.61.1 Constructor & Destructor Documentation

4.61.1.1 MHAFilter::gamma_ﬂt_t::gamma_ﬂt_t ( std::vector<mha_real_t >cf, std::vector<

mha_real_t >bw, mha_real_t srate, unsigned int order )

168 CONTENTS

Parameters

cf Center frequency in Hz.

bw Bandwidth in Hz (same number of entries as in cf).

srate Sampling frequency in Hz.

order Filter order.

4.62 MHAFilter::iir_ﬁlter_t Class Reference

IIR ﬁlter class wrapper for integration into parser structure.

Inheritance diagram for MHAFilter::iir_ﬁlter_t:

MHAFilter::iir_filter_t

MHAParser::parser_t

MHAParser::base_t

MHAPlugin::config_t

< iir_filter_state_t >

Public Member Functions

•iir_ﬁlter_t (std::string help="IIR ﬁlter structure", std::string def_A="[1]", std::string def_←-

B="[1]", unsigned int channels=1)

Constructor of the IIR ﬁlter.

• void ﬁlter (mha_wave_t ∗y, const mha_wave_t ∗x)

The ﬁlter processes the audio signal.

•mha_real_t ﬁlter (mha_real_t x, unsigned int ch)

Filter a single audio sample.

• void resize (unsigned int channels)

Change the number of channels after object creation.

4.62 MHAFilter::iir_ﬁlter_t Class Reference 169

Additional Inherited Members

4.62.1 Detailed Description

This class implements an inﬁnite impulse response ﬁlter. Since it inherits from MHAParser←-

::parser_t (p. 220), it can easily be integrated in the openMHA conﬁguration tree. It provides

the conﬁguration language variables "A" (vector of recursive ﬁlter coefﬁcients) and "B" (vector

of non-recursive ﬁlter coefﬁcients).

The ﬁlter instance reacts to changes in ﬁlter coefﬁcients through the openMHA conﬁguration

language, and uses the updated coefﬁcients in the next invocation of the ﬁlter method.

Update of the coefﬁcients is thread-safe and non-blocking. Simply add this subparser to your

parser items and use the "ﬁlter" member function. Filter states are reset to all 0 on update.

4.62.2 Constructor & Destructor Documentation

4.62.2.1 MHAFilter::iir_ﬁlter_t::iir_ﬁlter_t ( std::string help = "IIR ﬁlter structure",std::string

def_A = "[1]",std::string def_B = "[1]",unsigned int channels = 1)

Initialises the sub-parser structure and the memory for holding the ﬁlter's state.

Parameters

help The help string for the parser that groups the conﬁguration variables of this ﬁlter.

Could be used to describe the purpose of this IIR ﬁlter.

def_A The initial value of the vector of the recursive ﬁlter coefﬁcients, represented as

string.

def_B The initial value of the vector of the non-recursive ﬁlter coefﬁcients, represented

as string.

channels The number of indipendent audio channels to process with this ﬁlter. Needed to

allocate a state vector for each audio channel.

4.62.3 Member Function Documentation

4.62.3.1 void MHAFilter::iir_ﬁlter_t::ﬁlter ( mha_wave_t ∗y, const mha_wave_t ∗x)

All channels in the audio signal are processed using the same ﬁlter coefﬁcients. Indipendent

state is stored between calls for each audio channel.

170 CONTENTS

Parameters

yPointer to output signal holder. The output signal is stored here. Has to have the same

signal dimensions as the input signal x. In-place processing (y and x pointing to the

same signal holder) is possible.

xPointer to input signal holder. Number of channels has to be the same as given to the

constructor, or to the resize (p. 170) method.

4.62.3.2 mha_real_t MHAFilter::iir_ﬁlter_t::ﬁlter ( mha_real_t x, unsigned int ch )

Parameters

xThe single audio sample

ch Zero-based channel index. Use and change the state of channel ch. ch has to be less

than the number of channels given to the constructor or the resize (p. 170) method.

Returns

the ﬁltered result sample.

4.62.3.3 void MHAFilter::iir_ﬁlter_t::resize ( unsigned int channels )

Parameters

channels The new number of channels. Old ﬁlter states are lost.

4.63 MHAFilter::iir_ord1_real_t Class Reference

First order recursive ﬁlter.

Public Member Functions

•iir_ord1_real_t (std::vector<mha_real_t >A, std::vector<mha_real_t >B)

Constructor with ﬁlter coefﬁcients (one per channel)

•iir_ord1_real_t (std::vector<mha_real_t >tau, mha_real_t srate)

Constructor for low pass ﬁlter (one time constant per channel)

•mha_real_t operator() (unsigned int ch, mha_real_t x)

Filter method for real value input, one element.

•mha_complex_t operator() (unsigned int ch, mha_complex_t x)

Filter method for complex input, one element.

• void operator() (const mha_wave_t &X, mha_wave_t &Y)

Filter method for real value input.

• void operator() (const mha_spec_t &X, mha_spec_t &Y)

4.64 MHAFilter::o1_ar_ﬁlter_t Class Reference 171

Filter method for complex value input.

• void operator() (const mha_wave_t &Xre, const mha_wave_t &Xim, mha_wave_t &Yre,

mha_wave_t &Yim)

Filter method for complex value input.

4.64 MHAFilter::o1_ar_ﬁlter_t Class Reference

First order attack-release lowpass ﬁlter.

Inheritance diagram for MHAFilter::o1_ar_ﬁlter_t:

MHAFilter::o1_ar_filter_t

MHAFilter::o1flt_lowpass_t

MHASignal::waveform_t

mha_wave_t

MHAFilter::o1flt_maxtrack_t MHAFilter::o1flt_mintrack_t

Public Member Functions

•o1_ar_ﬁlter_t (unsigned int channels,mha_real_t fs=1.0f, std::vector<mha_real_t >

tau_a=std::vector<ﬂoat >(1, 0.0f), std::vector<mha_real_t >tau_r=std::vector<ﬂoat

>(1, 0.0f))

Constructor, setting all taus to zero.

• void set_tau_attack (unsigned int ch, mha_real_t tau)

Set the attack time constant.

• void set_tau_release (unsigned int ch, mha_real_t tau)

Set the release time constant.

•mha_real_t operator() (unsigned int ch, mha_real_t x)

Apply ﬁlter to value x, using state channel ch.

• void operator() (const mha_wave_t &in, mha_wave_t &out)

Apply ﬁlter to a mha_wave_t (p. 154) data.

172 CONTENTS

Additional Inherited Members

4.64.1 Detailed Description

This ﬁlter is the base of ﬁrst order lowpass ﬁlter, maximum tracker and minimum tracker.

4.64.2 Constructor & Destructor Documentation

4.64.2.1 MHAFilter::o1_ar_ﬁlter_t::o1_ar_ﬁlter_t ( unsigned int channels, mha_real_t fs = 1.0f,

std::vector<mha_real_t >tau_a = std::vector<float>(1,0.0f),std::vector<

mha_real_t >tau_r = std::vector<float>(1,0.0f) )

The ﬁlter state can be accessed through the member functions of MHASignal::waveform_t

(p. 268).

Parameters

channels Number of independent ﬁlters

fs Sampling rate (optional, default = 1)

tau_a Attack time constants (optional, default = 0)

tau_r Release time constants (optional, default = 0)

4.64.3 Member Function Documentation

4.64.3.1 void MHAFilter::o1_ar_ﬁlter_t::set_tau_attack ( unsigned int ch, mha_real_t tau )

Parameters

ch Channel number

tau Time constant

4.64.3.2 void MHAFilter::o1_ar_ﬁlter_t::set_tau_release ( unsigned int ch, mha_real_t tau )

Parameters

ch Channel number

tau Time constant

4.64.3.3 mha_real_t MHAFilter::o1_ar_ﬁlter_t::operator() ( unsigned int ch, mha_real_t x)

[inline]

4.65 MHAFilter::o1ﬂt_lowpass_t Class Reference 173

Parameters

ch Cannel number

xInput value

Returns

Output value

4.64.3.4 void MHAFilter::o1_ar_ﬁlter_t::operator() ( const mha_wave_t &in, mha_wave_t &out )

[inline]

Parameters

in Input signal

out Output signal

The number of channels must match the number of ﬁlter bands.

4.65 MHAFilter::o1ﬂt_lowpass_t Class Reference

First order low pass ﬁlter.

Inheritance diagram for MHAFilter::o1ﬂt_lowpass_t:

MHAFilter::o1flt_lowpass_t

MHAFilter::o1flt_maxtrack_t MHAFilter::o1flt_mintrack_t

MHAFilter::o1_ar_filter_t

MHASignal::waveform_t

mha_wave_t

174 CONTENTS

Public Member Functions

•o1ﬂt_lowpass_t (const std::vector<mha_real_t >&, mha_real_t,mha_real_t=0)

Constructor of low pass ﬁlter, sets sampling rate and time constants.

• void set_tau (unsigned int ch, mha_real_t tau)

change the time constant in one channel

• void set_tau (mha_real_t tau)

set time constant in all channels to tau

Additional Inherited Members

4.65.1 Constructor & Destructor Documentation

4.65.1.1 MHAFilter::o1ﬂt_lowpass_t::o1ﬂt_lowpass_t ( const std::vector<mha_real_t >&tau,

mha_real_t fs, mha_real_t startval = 0)

Parameters

tau Vector of time constants

fs Sampling rate

startval Initial internal state value

4.66 MHAFilter::o1ﬂt_maxtrack_t Class Reference

First order maximum tracker.

4.66 MHAFilter::o1ﬂt_maxtrack_t Class Reference 175

Inheritance diagram for MHAFilter::o1ﬂt_maxtrack_t:

MHAFilter::o1flt_maxtrack_t

MHAFilter::o1flt_lowpass_t

MHAFilter::o1_ar_filter_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

•o1ﬂt_maxtrack_t (const std::vector<mha_real_t >&, mha_real_t,mha_real_t=0)

Constructor of low pass ﬁlter, sets sampling rate and time constants.

• void set_tau (unsigned int ch, mha_real_t tau)

change the time constant in one channel

• void set_tau (mha_real_t tau)

set time constant in all channels to tau

Additional Inherited Members

4.66.1 Detailed Description

4.66.2 Constructor & Destructor Documentation

4.66.2.1 MHAFilter::o1ﬂt_maxtrack_t::o1ﬂt_maxtrack_t ( const std::vector<mha_real_t >&tau,

mha_real_t fs, mha_real_t startval = 0)

176 CONTENTS

Parameters

tau Vector of time constants

fs Sampling rate

startval Initial internal state value

4.67 MHAFilter::o1ﬂt_mintrack_t Class Reference

First order minimum tracker.

Inheritance diagram for MHAFilter::o1ﬂt_mintrack_t:

MHAFilter::o1flt_mintrack_t

MHAFilter::o1flt_lowpass_t

MHAFilter::o1_ar_filter_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

• void set_tau (unsigned int ch, mha_real_t tau)

change the time constant in one channel

• void set_tau (mha_real_t tau)

set time constant in all channels to tau

4.68 MHAFilter::partitioned_convolution_t Class Reference 177

Additional Inherited Members

4.67.1 Detailed Description

4.68 MHAFilter::partitioned_convolution_t Class Reference

A ﬁlter class for partitioned convolution.

Classes

• struct index_t

Bookkeeping class.

Public Member Functions

•partitioned_convolution_t (unsigned int fragsize, unsigned int nchannels_in, unsigned

int nchannels_out, const transfer_matrix_t &transfer)

Create a new partitioned convolver.

•∼partitioned_convolution_t ()

Free fftw resource allocated in constructor.

•mha_wave_t ∗process (const mha_wave_t ∗s_in)

processing

Public Attributes

• unsigned int fragsize

Audio fragment size, always equal to partition size.

• unsigned int nchannels_in

Number of audio input channels.

• unsigned int nchannels_out

Number of audio output channels.

• unsigned int output_partitions

The maximum number of partitions in any of the impulse responses.

• unsigned int ﬁlter_partitions

The total number of non-zero impulse response partitions.

•MHASignal::waveform_t input_signal_wave

Buffer for input signal.

• unsigned int current_input_signal_buffer_half_index

A counter modulo 2.

•MHASignal::spectrum_t input_signal_spec

Buffer for FFT transformed input signal.

•MHASignal::spectrum_t frequency_response

Buffers for frequency response spectra of impulse response partitions.

178 CONTENTS

• std::vector<index_t >bookkeeping

Keeps track of input channels, output channels, impulse response partition, and delay.

• std::vector<MHASignal::spectrum_t >output_signal_spec

Buffers for FFT transformed output signal.

• unsigned int current_output_partition_index

A counter modulo output_partitions, indexing the "current" output partition.

•MHASignal::waveform_t output_signal_wave

Buffer for the wave output signal.

•mha_fft_t fft

The FFT transformer.

4.68.1 Detailed Description

Impulse responses are partitioned into sections of fragment size. Audio signal is convolved with

every partition and delayed as needed. Convolution is done according to overlap-save. FFT

length used is 2 times fragment size.

4.68.2 Constructor & Destructor Documentation

4.68.2.1 MHAFilter::partitioned_convolution_t::partitioned_convolution_t ( unsigned int fragsize,

unsigned int nchannels_in, unsigned int nchannels_out, const transfer_matrix_t &transfer

)

Parameters

fragsize Audio fragment size, equal to partition size.

nchannels_in Number of input audio channels.

nchannels_out Number of output audio channels.

transfer A sparse matrix of impulse responses.

4.68.3 Member Data Documentation

4.68.3.1 unsigned int MHAFilter::partitioned_convolution_t::fragsize

4.68.3.2 unsigned int MHAFilter::partitioned_convolution_t::nchannels_in

4.68.3.3 unsigned int MHAFilter::partitioned_convolution_t::nchannels_out

4.68.3.4 unsigned int MHAFilter::partitioned_convolution_t::output_partitions

Determines the size if the delay line.

4.69 MHAFilter::partitioned_convolution_t::index_t Struct

Reference 179

4.68.3.5 unsigned int MHAFilter::partitioned_convolution_t::ﬁlter_partitions

4.68.3.6 MHASignal::waveform_t MHAFilter::partitioned_convolution_t::input_signal_wave

Has nchannels_in channels and fragsize∗2 frames

4.68.3.7 unsigned int MHAFilter::partitioned_convolution_t::current_input_signal_buffer_half_index

Indicates the buffer half in input signal wave into which to copy the current input signal.

4.68.3.8 MHASignal::spectrum_t MHAFilter::partitioned_convolution_t::input_signal_spec

Has nchannels_in channels and fragsize+1 frames (fft bins).

4.68.3.9 MHASignal::spectrum_t MHAFilter::partitioned_convolution_t::frequency_response

Each "channel" contains another partition of some impulse response. The bookkeeping array

is used to keep track what to do with these frequency responses. This container has ﬁlter_←-

partitions channels and fragsize+1 frames (fft bins).

4.68.3.10 std::vector<index_t>MHAFilter::partitioned_convolution_t::bookkeeping

The index into this array is the same as the "channel" index into the frequency_response array.

Array has ﬁlter_partitions entries.

4.68.3.11 std::vector<MHASignal::spectrum_t>MHAFilter::partitioned_convolution_t::output_←-

signal_spec

For each array member, Number of channels is equal to nchannels_out, number of frames (fft

bins) is equal to fragsize+1. Array size is equal to output_partitions.

4.68.3.12 unsigned int MHAFilter::partitioned_convolution_t::current_output_partition_index

4.68.3.13 MHASignal::waveform_t MHAFilter::partitioned_convolution_t::output_signal_wave

Number of channels is equal to nchannels_out, number of frames is equal to fragsize

4.69 MHAFilter::partitioned_convolution_t::index_t Struct Reference

Bookkeeping class.

180 CONTENTS

Public Member Functions

•index_t (unsigned int src, unsigned int tgt, unsigned int dly)

Data constructor.

•index_t ()

Default constructor for STL compatibility.

Public Attributes

• unsigned int source_channel_index

The input channel index to apply the current partition to.

• unsigned int target_channel_index

The index of the output channel to which the ﬁlter result should go.

• unsigned int delay

The delay (in blocks) of this partition.

4.69.1 Detailed Description

For each impulse response partition, keeps track of which input to ﬁlter, which output channel

to ﬁlter to, and the delay in blocks. Objects of class Index should be kept in an array with the

same indices as the corresponding inpulse response partitions.

4.69.2 Constructor & Destructor Documentation

4.69.2.1 MHAFilter::partitioned_convolution_t::index_t::index_t ( unsigned int src, unsigned int tgt,

unsigned int dly )[inline]

Parameters

src The input channel index to apply the current partition to.

tgt The index of the output channel to which the ﬁlter result should go.

dly The delay (in blocks) of this partition

4.69.3 Member Data Documentation

4.69.3.1 unsigned int MHAFilter::partitioned_convolution_t::index_t::source_channel_index

4.69.3.2 unsigned int MHAFilter::partitioned_convolution_t::index_t::target_channel_index

4.70 MHAFilter::polyphase_resampling_t Class Reference

A class that does polyphase resampling.

4.70 MHAFilter::polyphase_resampling_t Class Reference 181

Public Member Functions

•polyphase_resampling_t (unsigned n_up, unsigned n_down, mha_real_t nyquist_ratio,

unsigned n_irs, unsigned n_ringbuffer, unsigned n_channels, unsigned n_preﬁll)

Initialize a polyphase resampler.

• void write (mha_wave_t &signal)

Write signal to the ringbuffer.

• void read (mha_wave_t &signal)

Read resampled signal.

• unsigned readable_frames () const

Number of frames at target sampling rate that can be produced.

Private Attributes

• unsigned upsampling_factor

Interpolation rate / source rate.

• unsigned downsampling_factor

Interpolation rate / target rate.

• unsigned now_index

points to "now" in the interpolated sampling rate

• bool underﬂow

indicates if an underﬂow has occurred. Object cannot be used then.

•MHAWindow::hanning_t impulse_response

contains the lowpass impulse response at interpolation rate

•MHASignal::ringbuffer_t ringbuffer

storage of input signal

4.70.1 Detailed Description

4.70.2 Constructor & Destructor Documentation

4.70.2.1 MHAFilter::polyphase_resampling_t::polyphase_resampling_t ( unsigned n_up, unsigned

n_down, mha_real_t nyquist_ratio, unsigned n_irs, unsigned n_ringbuffer, unsigned

n_channels, unsigned n_preﬁll )

Parameters

n_up upsampling factor

n_down downsampling factor

nyquist_ratio low pass ﬁlter cutoff frequency relative to the nyquist frequency of the

smaller of the two sampling rates. Example values: 0.8, 0.9

n_irs length of impulse response (in samples at interpolation rate)

n_ringbuffer length of ringbuffer, in samples at source sampling rate

n_channels audio channels count

n_preﬁll Preﬁll the ringbuffer with this many zero frames in samples at source

sampling rate

182 CONTENTS

4.70.3 Member Function Documentation

4.70.3.1 void MHAFilter::polyphase_resampling_t::write ( mha_wave_t &signal )

Parameters

signal input signal in original sampling rate

Exceptions

MHA_Error (p. 132)Raises exception if there is not enough room or if the number of

channels does not match.

4.70.3.2 void MHAFilter::polyphase_resampling_t::read ( mha_wave_t &signal )

Will perform the resampling and remove no longer needed samples from the input buffer.

Parameters

signal buffer to write the resampled signal to.

Exceptions

MHA_Error (p. 132)Raises exception if there is not enough input signal or if the number

of channels is too high.

4.70.3.3 unsigned MHAFilter::polyphase_resampling_t::readable_frames ( ) const [inline]

Warning: This method only checks for enough future samples present, therefore, this number

can be positive and a read operation can still fail if there are not enough past samples present

to perform the ﬁltering for the ﬁrst output sample.

4.71 MHAFilter::resampling_ﬁlter_t Class Reference

Hann shaped low pass ﬁlter for resampling.

4.72 MHAFilter::smoothspec_t Class Reference 183

Inheritance diagram for MHAFilter::resampling_ﬁlter_t:

MHAFilter::resampling

_filter_t

MHAFilter::fftfilter_t

Public Member Functions

•resampling_ﬁlter_t (unsigned int fftlen, unsigned int irslen, unsigned int channels, un-

signed int Nup, unsigned int Ndown, double fCutOff)

Constructor.

4.71.1 Detailed Description

This class uses FFT ﬁlter at upsampled rate.

4.71.2 Constructor & Destructor Documentation

4.71.2.1 MHAFilter::resampling_ﬁlter_t::resampling_ﬁlter_t ( unsigned int fftlen, unsigned int irslen,

unsigned int channels, unsigned int Nup, unsigned int Ndown, double fCutOff )

Parameters

fftlen FFT length.

irslen Length of ﬁlter.

channels Number of channels to be ﬁltered.

Nup Upsampling ratio.

Ndown Downsampling ratio.

fCutOff Cut off frequency (relative to lower Nyquist Frequency)

4.72 MHAFilter::smoothspec_t Class Reference

Smooth spectral gains, create a windowed impulse response.

184 CONTENTS

Public Member Functions

•smoothspec_t (unsigned int fftlen, unsigned int nchannels, const MHAWindow::base←-

_t &window, bool minphase, bool linphase_asym=false)

Constructor.

• void smoothspec (const mha_spec_t &s_in, mha_spec_t &s_out)

Create a smoothed spectrum.

• void smoothspec (mha_spec_t &spec)

Create a smoothed spectrum (in place)

• void spec2ﬁr (const mha_spec_t &spec, mha_wave_t &ﬁr)

Return FIR coefﬁcients.

4.72.1 Detailed Description

Spectral gains are smoothed by multiplicating the impulse response with a window function.

If a minimal phase is used, then the original phase is discarded and replaced by the minimal

phase function. In this case, the window is applied to the beginning of the inverse Fourier

transform of the input spectrum, and the remaining signal set to zero. If the original phase

is kept, the window is applied symmetrical arround zero, i.e. to the ﬁrst and last samples of

the inverse Fourier transform of the input spectrum. The spec2ﬁr() (p. 185) function creates a

causal impulse response by circular shifting the impulse response by half of the window length.

The signal dimensions of the arguments of smoothspec() (p. 184) must correspond to the FFT

length and number of channels provided in the constructor. The function spec2ﬁr() (p. 185)

can ﬁll signal structures with more than window length frames.

4.72.2 Constructor & Destructor Documentation

4.72.2.1 MHAFilter::smoothspec_t::smoothspec_t ( unsigned int fftlen, unsigned int nchannels, const

MHAWindow::base_t &window, bool minphase, bool linphase_asym = false )

Parameters

fftlen FFT length of input spectrum (fftlen/2+1 bins)

nchannels Number of channels in input spectrum

window Window used for smoothing

minphase Use minimal phase (true) or original phase (false)

linphase_asym Keep phase, but apply full window at beginning of IRS

4.72.3 Member Function Documentation

4.72.3.1 void MHAFilter::smoothspec_t::smoothspec ( const mha_spec_t &s_in, mha_spec_t &

s_out )

4.73 MHAFilter::transfer_function_t Struct Reference 185

Parameters

s←-

_in

Input spectrum

Return values

s_out Output spectrum

4.72.3.2 void MHAFilter::smoothspec_t::smoothspec ( mha_spec_t &spec )[inline]

Parameters

spec Spectrum to be smoothed.

4.72.3.3 void MHAFilter::smoothspec_t::spec2ﬁr ( const mha_spec_t &spec, mha_wave_t &ﬁr )

Parameters

spec Input spectrum

Return values

ﬁr FIR coefﬁcients, minimum length is window length

4.73 MHAFilter::transfer_function_t Struct Reference

a structure containing a source channel number, a target channel number, and an impulse

response.

Public Member Functions

•transfer_function_t ()

Default constructor for STL conformity.

•transfer_function_t (unsigned int source_channel_index, unsigned int target_←-

channel_index, const std::vector<ﬂoat >&impulse_response)

Data constructor.

• unsigned int partitions (unsigned int fragsize) const

for the given partition size, return the number of partitions of the impulse response.

• unsigned int non_empty_partitions (unsigned int fragsize) const

for the given partition size, return the number of non-empty partitions of the impulse response.

• bool isempty (unsigned int fragsize, unsigned int index) const

checks if the partition contains only zeros

186 CONTENTS

Public Attributes

• unsigned int source_channel_index

Source audio channel index for this transfer function.

• unsigned int target_channel_index

Target audio channel index for this transfer function.

• std::vector<ﬂoat >impulse_response

Impulse response of transfer from source to target channel.

4.73.1 Constructor & Destructor Documentation

4.73.1.1 MHAFilter::transfer_function_t::transfer_function_t ( ) [inline]

Not used.

4.73.1.2 MHAFilter::transfer_function_t::transfer_function_t ( unsigned int source_channel_index,

unsigned int target_channel_index, const std::vector<ﬂoat >&impulse_response )

Parameters

source_channel_index Source audio channel index for this transfer function

target_channel_index Target audio channel index for this transfer function

impulse_response Impulse response of transfer from source to target channel

4.73.2 Member Function Documentation

4.73.2.1 unsigned int MHAFilter::transfer_function_t::partitions ( unsigned int fragsize ) const

[inline]

Parameters

fragsize partition size

Returns

number of partitions occupied by the impulse response

4.73.2.2 unsigned int MHAFilter::transfer_function_t::non_empty_partitions ( unsigned int fragsize )

const [inline]

Parameters

fragsize partition size

4.74 MHAFilter::transfer_matrix_t Struct Reference 187

Returns

the number of non-empty partitions of the impulse response, i.e. partitions containing

only zeros are not counted.

4.73.2.3 bool MHAFilter::transfer_function_t::isempty ( unsigned int fragsize, unsigned int index ) const

[inline]

Parameters

fragsize partition size

index partition index

Returns

true when this partition of the impulse response contains only zeros.

4.74 MHAFilter::transfer_matrix_t Struct Reference

A sparse matrix of transfer function partitionss.

Inherits vector<transfer_function_t >.

Public Member Functions

• std::valarray<unsigned int >partitions (unsigned fragsize) const

Returns an array of the results of calling the partitions() (p. 187) method on every matrix

member.

• std::valarray<unsigned int >non_empty_partitions (unsigned int fragsize) const

Returns an array of the results of calling the non_empty_partitions() (p. 187) method on every

matrix member.

4.74.1 Detailed Description

Each matrix element knows its position in the matrix, so they can be stored as a vector.

188 CONTENTS

4.75 MHAIOJack::io_jack_t Class Reference

Main class for JACK IO.

Inheritance diagram for MHAIOJack::io_jack_t:

MHAIOJack::io_jack_t

MHAParser::parser_t

MHAParser::base_t

MHAJack::client_t

Public Member Functions

• void prepare (int, int)

Allocate buffers, activate JACK client and install internal ports.

Private Member Functions

• void reconnect_inports ()

Connect the input ports when connection variable is accessed.

• void reconnect_outports ()

Connect the output ports when connection variable is accessed.

Additional Inherited Members

4.75.1 Detailed Description

This class registers a JACK client. JACK and framework states are managed by this class.

4.76 MHAJack::client_avg_t Class Reference 189

4.75.2 Member Function Documentation

4.75.2.1 void io_jack_t::prepare ( int nch_in, int nch_out )

4.75.2.2 void io_jack_t::reconnect_inports ( ) [private]

4.75.2.3 void io_jack_t::reconnect_outports ( ) [private]

4.76 MHAJack::client_avg_t Class Reference

Generic JACK client for averaging a system response across time.

Inheritance diagram for MHAJack::client_avg_t:

MHAJack::client_avg_t

MHAJack::client_t

Public Member Functions

•client_avg_t (const std::string &name, const unsigned int &nrep_)

Constructor for averaging client.

• void io (mha_wave_t ∗s_out, mha_wave_t ∗s_in, const std::vector<std::string >&p_←-

out, const std::vector<std::string >&p_in, ﬂoat ∗srate=NULL, unsigned int ∗fragsize=N←-

ULL)

Recording function.

4.76.1 Detailed Description

4.76.2 Constructor & Destructor Documentation

4.76.2.1 MHAJack::client_avg_t::client_avg_t ( const std::string & name_, const unsigned int & nrep_ )

190 CONTENTS

Parameters

name←-

Name of JACK client

nrep←-

Number of repetitions

4.76.3 Member Function Documentation

4.76.3.1 void MHAJack::client_avg_t::io ( mha_wave_t ∗is_out, mha_wave_t ∗is_in, const

std::vector<std::string >&p_out, const std::vector<std::string >&p_in, ﬂoat ∗srate =

NULL,unsigned int ∗fragsize = NULL )

long-description

Parameters

is_out Input (test) signal, which will be repeated

is_in System response (averaged, same length as input required)

p_out Ports to play back the test signal

p_in Ports to record from the system response

srate Pointer to sampling rate variable, will be ﬁlled with server sampling rate

fragsize Pointer to fragment size variable, will be ﬁlled with server fragment size

4.77 MHAJack::client_noncont_t Class Reference

Generic client for synchronous playback and recording of waveform fragments.

Inheritance diagram for MHAJack::client_noncont_t:

MHAJack::client_noncont_t

MHAJack::client_t

4.78 MHAJack::client_t Class Reference 191

Additional Inherited Members

4.78 MHAJack::client_t Class Reference

Generic asynchronous JACK client.

Inheritance diagram for MHAJack::client_t:

MHAJack::client_t

MHAIOJack::io_jack_t MHAJack::client_avg_t MHAJack::client_noncont_t

Public Member Functions

• void prepare (const std::string &client_name, const unsigned int &nchannels_in, const

unsigned int &nchannels_out)

Allocate buffers, activate JACK client and install internal ports.

• void prepare (const std::string &server_name, const std::string &client_name, const un-

signed int &nchannels_in, const unsigned int &nchannels_out)

Allocate buffers, ports, and activates JACK client.

• void release ()

Remove JACK client and deallocate internal ports and buffers.

• void connect_input (const std::vector<std::string >&)

Connect the input ports when connection variable is accessed.

• void connect_output (const std::vector<std::string >&)

Connect the output ports when connection variable is accessed.

• void get_ports (std::vector<std::string >&, unsigned long jack_ﬂags)

Get a list of Jack ports.

Private Member Functions

• void prepare_impl (const char ∗server_name, const char ∗client_name, const unsigned

int &nchannels_in, const unsigned int &nchannels_out)

Allocate buffers, activate JACK client and allocates jack ports Registers the jack client with the

given server and activates it.

• int jack_proc_cb (jack_nframes_t)

This is the main processing callback.

192 CONTENTS

4.78.1 Detailed Description

4.78.2 Member Function Documentation

4.78.2.1 void MHAJack::client_t::prepare ( const std::string & client_name, const unsigned int & nch_in,

const unsigned int & nch_out )

Registers the jack client with the default jack server and activates it.

Parameters

client_name Name of this jack client

nch_in Input ports to register

nch_out Output ports to register

4.78.2.2 void MHAJack::client_t::prepare ( const std::string & server_name, const std::string &

client_name, const unsigned int & nch_in, const unsigned int & nch_out )

Registers the jack client with speciﬁed jack server and activates it.

Parameters

server_name Name of the jack server to register with

client_name Name of this jack client

nch_in Input ports to register

nch_out Output ports to register

4.78.2.3 void MHAJack::client_t::release ( )

4.78.2.4 void MHAJack::client_t::connect_input ( const std::vector<std::string >&con )

4.78.2.5 void MHAJack::client_t::connect_output ( const std::vector<std::string >&con )

4.78.2.6 void MHAJack::client_t::get_ports ( std::vector<std::string >&res, unsigned long jack_ﬂags

)

Parameters

res Result string vector

jack_ﬂags Jack port ﬂags (JackPortInput etc.)

4.79 MHAJack::port_t Class Reference 193

4.78.2.7 void MHAJack::client_t::prepare_impl ( const char ∗server_name, const char ∗client_name,

const unsigned int & nch_in, const unsigned int & nch_out )[private]

Parameters

server_name Name of the jack server to register with

client_name Name of this jack client

nch_in Input ports to register

nch_out Output ports to register

4.78.2.8 int MHAJack::client_t::jack_proc_cb ( jack_nframes_t n)[private]

Here happens double buffering and downsampling.

4.79 MHAJack::port_t Class Reference

Class for one channel/port.

Public Member Functions

•port_t (jack_client_t ∗jc, dir_t dir, int id)

•port_t (jack_client_t ∗jc, dir_t dir, const std::string &id)

Constructor to create port with speciﬁc name.

• void read (mha_wave_t ∗s, unsigned int ch)

• void write (mha_wave_t ∗s, unsigned int ch)

• void mute (unsigned int n)

• void connect_to (const char ∗pn)

• const char ∗get_short_name ()

Return the port name.

4.79.1 Detailed Description

This class represents one JACK port. Double buffering for asynchronous process callbacks is

managed by this class.

4.79.2 Constructor & Destructor Documentation

4.79.2.1 MHAJack::port_t::port_t ( jack_client_t ∗jc, dir_t dir, int id )

194 CONTENTS

Parameters

jc JACK client.

dir Direction (input/output).

id Number in port name (starting with 1).

4.79.2.2 MHAJack::port_t::port_t ( jack_client_t ∗jc, dir_t dir, const std::string & id )

Parameters

jc JACK client.

dir Direction (input/output).

id Port name.

4.79.3 Member Function Documentation

4.79.3.1 void MHAJack::port_t::read ( mha_wave_t ∗s, unsigned int ch )

Parameters

sSignal structure to store the audio data.

ch Channel number in audio data structure to be used.

4.79.3.2 void MHAJack::port_t::write ( mha_wave_t ∗s, unsigned int ch )

Parameters

sSignal structure from which the audio data is read.

ch Channel number in audio data structure to be used.

4.79.3.3 void MHAJack::port_t::mute ( unsigned int n)

Parameters

nNumber of samples to be muted (must be the same as reported by Jack processing

callback).

4.79.3.4 void MHAJack::port_t::connect_to ( const char ∗pn )

Parameters

pn Port name to connect to

4.80 MHAMultiSrc::base_t Class Reference 195

4.80 MHAMultiSrc::base_t Class Reference

Base class for source selection.

Inheritance diagram for MHAMultiSrc::base_t:

MHAMultiSrc::base_t

MHAPlugin::config_t

< MHAMultiSrc::channels_t >

Public Member Functions

• void select_source (const std::vector<std::string >&src, int in_channels)

Change the selection of input sources.

4.80.1 Detailed Description

See also

MHAMultiSrc::channel_t

MHAMultiSrc::channels_t

4.80.2 Member Function Documentation

4.80.2.1 void MHAMultiSrc::base_t::select_source ( const std::vector<std::string >&src, int

in_channels )

This function is real-time and thread safe.

Parameters

src List of input sources

in_channels Number of input channels in direct input (the processed signal)

196 CONTENTS

4.81 MHAOvlFilter::fftfb_t Class Reference

FFT based overlapping ﬁlter bank.

Inheritance diagram for MHAOvlFilter::fftfb_t:

MHAOvlFilter::fftfb_t

MHAOvlFilter::overlap

_save_filterbank_t

MHAOvlFilter::fspacing_t MHASignal::waveform_t

mha_wave_t

Public Member Functions

•fftfb_t (MHAOvlFilter::fftfb_vars_t &par, unsigned int nfft, mha_real_t fs)

Constructor for a FFT-based overlapping ﬁlter bank.

• unsigned int bin1 (unsigned int band) const

Return index of ﬁrst non-zero ﬁlter shape window.

• unsigned int bin2 (unsigned int band) const

Return index of ﬁrst zero ﬁlter shape window above center frequency.

• unsigned int get_fftlen () const

Return fft length.

•mha_real_t w (unsigned int k, unsigned int b) const

Return ﬁlter shape window at index k in band b.

Additional Inherited Members

4.81.1 Constructor & Destructor Documentation

4.81.1.1 MHAOvlFilter::fftfb_t::fftfb_t ( MHAOvlFilter::fftfb_vars_t &par, unsigned int nfft,

mha_real_t fs )

4.82 MHAOvlFilter::fftfb_vars_t Class Reference 197

Parameters

par Parameters for the FFT ﬁlterbank that can not be deduced from the signal dimensions

are taken from this set of conﬁguration variables.

nfft FFT length

fs Sampling rate / Hz

4.81.2 Member Function Documentation

4.81.2.1 mha_real_t MHAOvlFilter::fftfb_t::w ( unsigned int k, unsigned int b) const [inline]

Parameters

kFrequency index

bBand index

4.82 MHAOvlFilter::fftfb_vars_t Class Reference

Set of conﬁguration variables for FFT-based overlapping ﬁlters.

Inherited by MHAOvlFilter::overlap_save_ﬁlterbank_t::vars_t.

Public Member Functions

•fftfb_vars_t (MHAParser::parser_t &p)

construct a set of openMHA conﬁguration language variables suitable for conﬁguring the FFT-

based overlapping ﬁlterbank.

Public Attributes

• scale_var_t fscale

Frequency scale type (lin/bark/log/erb).

• scale_var_t ovltype

Filter shape (rect/lin/hann).

•MHAParser::ﬂoat_t plateau

relative plateau width.

•MHAParser::kw_t ftype

Flag to decide wether edge or center frequencies are used.

• fscale_t f

Frequency.

•MHAParser::bool_t normalize

Normalize sum of channels.

•MHAParser::bool_t fail_on_nonmonotonic

198 CONTENTS

Fail if frequency entries are non-monotonic (otherwise sort)

•MHAParser::bool_t fail_on_unique_bins

Fail if center frequencies share the same FFT bin.

•MHAParser::vﬂoat_mon_t cf

Final center frequencies in Hz.

•MHAParser::vﬂoat_mon_t ef

Final edge frequencies in Hz.

•MHAParser::vﬂoat_mon_t cLTASS

Bandwidth correction for LTASS noise (level of 0 dB RMS LTASS noise)

4.82.1 Detailed Description

This class enables easy conﬁguration of the FFT-based overlapping ﬁlterbank. An instance of

fftfb_vars_t (p. 197) creates openMHA conﬁguration language variables needed for conﬁguring

the ﬁlterbank, and inserts these variables in the openMHA conﬁguration tree.

This way, the variables are visible to the user and can be conﬁgured using the openMHA con-

ﬁguration language.

4.82.2 Constructor & Destructor Documentation

4.82.2.1 MHAOvlFilter::fftfb_vars_t::fftfb_vars_t ( MHAParser::parser_t &p)

Parameters

pThe node of the conﬁguration tree where the variables created by this instance are

inserted.

4.83 MHAOvlFilter::fspacing_t Class Reference

Class for frequency spacing, used by ﬁlterbank shape generator class.

4.84 MHAOvlFilter::overlap_save_ﬁlterbank_t Class Reference 199

Inheritance diagram for MHAOvlFilter::fspacing_t:

MHAOvlFilter::fspacing_t

MHAOvlFilter::fftfb_t

MHAOvlFilter::overlap

_save_filterbank_t

Public Member Functions

• unsigned int nbands () const

Return number of bands in ﬁlter bank.

4.84 MHAOvlFilter::overlap_save_ﬁlterbank_t Class Reference

A time-domain minimal phase ﬁlter bank with frequency shapes from MHAOvlFilter::fftfb_t

(p. 196).

Inheritance diagram for MHAOvlFilter::overlap_save_ﬁlterbank_t:

MHAOvlFilter::overlap

_save_filterbank_t

MHAOvlFilter::fftfb_t

MHAOvlFilter::fspacing_t MHASignal::waveform_t

mha_wave_t

MHAFilter::fftfilterbank_t

200 CONTENTS

Additional Inherited Members

4.85 MHAParser::base_t Class Reference

Base class for all parser items.

Inheritance diagram for MHAParser::base_t:

MHAParser::base_t

MHAParser::monitor_t

MHAParser::parser_t

MHAParser::bool_mon_t

MHAParser::complex

_mon_t

MHAParser::float_mon_t

MHAParser::int_mon_t

MHAParser::mcomplex

_mon_t

MHAParser::mfloat_mon_t

MHAParser::string_mon_t

MHAParser::variable_t

MHAParser::vcomplex

_mon_t

MHAParser::vfloat_mon_t

MHAParser::vint_mon_t

MHAParser::vstring

_mon_t

MHAParser::bool_t

MHAParser::kw_t

MHAParser::range_var_t

MHAParser::string_t

MHAParser::vstring_t

MHAParser::commit_t

< receiver_t >

MHAParser::complex_t

MHAParser::float_t

MHAParser::int_t

MHAParser::mcomplex_t

MHAParser::mfloat_t

MHAParser::vcomplex_t

MHAParser::vfloat_t

MHAParser::vint_t

AuditoryProfile::parser_t

DynComp::dc_afterburn

_vars_t

io_file_t

io_parser_t

io_tcp_parser_t

MHAFilter::adapt_filter_t

MHAFilter::iir_filter_t

MHAIOJack::io_jack_t

MHAParser::window_t

MHAPlugin::plugin_t

< runtime_cfg_t >

DynComp::dc_afterburn_t

Public Member Functions

•base_t (const std::string &)

Constructor for base class of all parser nodes.

• virtual std::string parse (const std::string &)

Causes this node to process a command in the openMHA conﬁguration language.

• virtual void parse (const char ∗, char ∗, unsigned int)

This function parses a command and writes the parsing result into a C character array.

4.85 MHAParser::base_t Class Reference 201

• void set_node_id (const std::string &)

Set the identiﬁcation string of this parser node.

• void set_help (const std::string &)

Set the help comment of a variable or parser.

• const std::string & fullname () const

Return the full dot-separated path name of this parser node in the openMHA conﬁguration tree.

Public Attributes

•MHAEvents::emitter_t writeaccess

Event emitted on write access.

•MHAEvents::emitter_t valuechanged

Event emitted if the value has changed.

•MHAEvents::emitter_t readaccess

Event emitted on read access.

•MHAEvents::emitter_t prereadaccess

Event emitted on read access, before the data ﬁeld is accessed.

4.85.1 Detailed Description

The key method of the parser base class is the std::string parse(const std::string&) (p. 201)

method. Parser proxy derivatives which overwrite any of the other parse() (p. 201) methods to

be the key method must make sure that the original parse() (p. 201) method utilizes the new

key method.

4.85.2 Constructor & Destructor Documentation

4.85.2.1 MHAParser::base_t::base_t ( const std::string & h)

Parameters

hHelp text describing this parser node. This help text is accessible to the conﬁguration

language through the "?help" query command.

4.85.3 Member Function Documentation

4.85.3.1 std::string MHAParser::base_t::parse ( const std::string & cs )[virtual]

Parameters

cs The command to parse

202 CONTENTS

Returns

The response to the command, if successful

Exceptions

MHA_Error (p. 132)If the command cannot be executed successfully. The reason for

failure is given in the message string of the exception.

4.85.3.2 void MHAParser::base_t::parse ( const char ∗cmd, char ∗retv, unsigned int len )

[virtual]

This base class implementation delegates to parse(const std::string &) (p. 201).

Parameters

cmd Command to be parsed

retv Buffer for the result

len Length of buffer

4.85.3.3 void MHAParser::base_t::set_node_id ( const std::string & s)

The id can be queried from the conﬁguration language using the ?id query command. Nodes

can be found by id using the ?listid query command on a containing parser node.

Parameters

sThe new identiﬁcation string.

4.85.3.4 void MHAParser::base_t::set_help ( const std::string & s)

Parameters

sNew help comment.

4.85.3.5 const std::string & MHAParser::base_t::fullname ( ) const

4.85.4 Member Data Documentation

4.85.4.1 MHAEvents::emitter_t MHAParser::base_t::writeaccess

To connect a callback that is invoked on write access to this parser variable, use MHAEvents←-

::patchbay_t<receiver_t>method connect(&writeaccess,&receiver_t::callback) where callback

is a method that expects no parameters and returns void.

4.86 MHAParser::bool_mon_t Class Reference 203

4.85.4.2 MHAEvents::emitter_t MHAParser::base_t::valuechanged

To connect a callback that is invoked when write access to this parser variable actually changes

its value, use MHAEvents::patchbay_t<receiver_t>method connect(&valuechanged,&receiver←-

_t::callback) where callback is a method that expects no parameters and returns void.

4.85.4.3 MHAEvents::emitter_t MHAParser::base_t::readaccess

To connect a callback that is invoked after the value of this variable has been read

through the conﬁguration interface, use MHAEvents::patchbay_t<receiver_t>method

connect(&readaccess,&receiver_t::callback) where callback is a method that expects no pa-

rameters and returns void.

4.85.4.4 MHAEvents::emitter_t MHAParser::base_t::prereadaccess

To connect a callback that is invoked when the value of this variable is about to be read through

the conﬁguration interface, so that the callback can inﬂuence the value that is reported, use

MHAEvents::patchbay_t<receiver_t>method connect(&prereadaccess,&receiver_t::callback)

where callback is a method that expects no parameters and returns void.

4.86 MHAParser::bool_mon_t Class Reference

Monitor with string value.

Inheritance diagram for MHAParser::bool_mon_t:

MHAParser::bool_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•bool_mon_t (const std::string &hlp)

Create a monitor variable for string values.

204 CONTENTS

Public Attributes

• bool data

Data ﬁeld.

4.86.1 Constructor & Destructor Documentation

4.86.1.1 MHAParser::bool_mon_t::bool_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.87 MHAParser::bool_t Class Reference

Variable with a boolean value ("yes"/"no")

Inheritance diagram for MHAParser::bool_t:

MHAParser::bool_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•bool_t (const std::string &help_text, const std::string &initial_value)

Constructor for a conﬁguration language variable for boolean values.

4.88 MHAParser::commit_t<receiver_t >Class Template Reference205

Public Attributes

• bool data

Data ﬁeld.

4.87.1 Constructor & Destructor Documentation

4.87.1.1 MHAParser::bool_t::bool_t ( const std::string & help_text, const std::string & initial_value )

Parameters

help_text A human-readable text describing the purpose of this conﬁguration variable.

initial_value The initial value for this variable as a string. The string representation of 'true'

is either "yes" or "1". The string representation of 'false' is either "no" or "0".

4.88 MHAParser::commit_t<receiver_t >Class Template Reference

Parser variable with event-emission functionality.

Inheritance diagram for MHAParser::commit_t<receiver_t >:

MHAParser::commit_t

< receiver_t >

MHAParser::kw_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

206 CONTENTS

Additional Inherited Members

4.88.1 Detailed Description

template<class receiver_t>

class MHAParser::commit_t<receiver_t >

The commit_t (p. 205) variable can register an event receiver in its constructor, which is called

whenever the variable is set to "commit".

4.89 MHAParser::complex_mon_t Class Reference

Monitor with complex value.

Inheritance diagram for MHAParser::complex_mon_t:

MHAParser::complex

_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•complex_mon_t (const std::string &hlp)

Create a complex monitor variable.

Public Attributes

•mha_complex_t data

Data ﬁeld.

4.89.1 Constructor & Destructor Documentation

4.89.1.1 MHAParser::complex_mon_t::complex_mon_t ( const std::string & hlp )

4.90 MHAParser::complex_t Class Reference 207

Parameters

hlp A help text describing this monitor variable.

4.90 MHAParser::complex_t Class Reference

Variable with complex value.

Inheritance diagram for MHAParser::complex_t:

MHAParser::complex_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Attributes

•mha_complex_t data

Data ﬁeld.

Additional Inherited Members

4.91 MHAParser::ﬂoat_mon_t Class Reference

Monitor with ﬂoat value.

208 CONTENTS

Inheritance diagram for MHAParser::ﬂoat_mon_t:

MHAParser::float_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•ﬂoat_mon_t (const std::string &hlp)

Initialize a ﬂoating point (32 bits) monitor variable.

Public Attributes

• ﬂoat data

Data ﬁeld.

4.91.1 Constructor & Destructor Documentation

4.91.1.1 MHAParser::ﬂoat_mon_t::ﬂoat_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.92 MHAParser::ﬂoat_t Class Reference

Variable with ﬂoat value.

4.92 MHAParser::ﬂoat_t Class Reference 209

Inheritance diagram for MHAParser::ﬂoat_t:

MHAParser::float_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•ﬂoat_t (const std::string &help_text, const std::string &initial_value, const std::string

&range="")

Constructor for a conﬁguration language variable for 32bit ieee ﬂoating-point values.

Public Attributes

• ﬂoat data

Data ﬁeld.

Additional Inherited Members

4.92.1 Constructor & Destructor Documentation

4.92.1.1 MHAParser::ﬂoat_t::ﬂoat_t ( const std::string & help_text, const std::string & initial_value,

const std::string & range = "" )

210 CONTENTS

Parameters

help_text A human-readable text describing the purpose of this conﬁguration variable.

initial_value The initial value for this variable as a string (decimal representation of the

ﬂoating-point variable). If a range is given in the third parameter, then the

initial value has to be within the range. A human-readable text describing the

purpose of this conﬁguration variable.

range The range of values that this variable can hold can be restricted. A range is a

string of the form "[a,b]", where a and b are decimal representations of the

inclusive boundaries of the range. a<=b. In a range of the form "]a,b[", both

boundaries are excluded. Mixed forms are permitted. a or b can also be

omitted if there is no lower or upper limit. The range of values is always

restricted by the representable range of the underlying C data type.

4.93 MHAParser::int_mon_t Class Reference

Monitor variable with int value.

Inheritance diagram for MHAParser::int_mon_t:

MHAParser::int_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•int_mon_t (const std::string &hlp)

Create a monitor variable for integral values.

Public Attributes

• int data

Data ﬁeld.

4.94 MHAParser::int_t Class Reference 211

4.93.1 Detailed Description

Monitor variables can be of many types. These variables can be queried through the parser.

The public data element contains the monitored state. Write access is only possible from the

C++ code by direct access to the data ﬁeld.

4.93.2 Constructor & Destructor Documentation

4.93.2.1 MHAParser::int_mon_t::int_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.94 MHAParser::int_t Class Reference

Variable with integer value.

Inheritance diagram for MHAParser::int_t:

MHAParser::int_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

212 CONTENTS

Public Member Functions

•int_t (const std::string &help_text, const std::string &initial_value, const std::string

&range="")

Constructor for a conﬁguration language variable for integral values.

Public Attributes

• int data

Data ﬁeld.

Additional Inherited Members

4.94.1 Constructor & Destructor Documentation

4.94.1.1 MHAParser::int_t::int_t ( const std::string & help_text, const std::string & initial_value, const

std::string & range = "" )

Parameters

help_text A human-readable text describing the purpose of this conﬁguration variable.

initial_value The initial value for this variable as a string (decimal representation of the

integer variable). If a range is given in the third parameter, then the initial

value has to be within the range.

range The range of values that this variable can hold can be restricted. A range is a

string of the form "[a,b]", where a and b are decimal representations of the

integral inclusive boundaries of the range. a<=b. In a range of the form

"]a,b[", both boundaries are excluded. Mixed forms are permitted. a or b can

also be omitted if there is no lower or upper limit. The range of values is

always restricted by the representable range of the underlying C data type

(usually 32 bits, [-2147483648,2147483647]).

4.95 MHAParser::keyword_list_t Class Reference

Keyword list class.

Public Member Functions

• void set_value (const std::string &)

Select a value from keyword list.

• void set_entries (const std::string &)

Set keyword list entries.

4.95 MHAParser::keyword_list_t Class Reference 213

• const std::string & get_value () const

Return selected value.

• const std::vector<std::string >&get_entries () const

Return keyword list.

• const size_t & get_index () const

Return index of selected value.

• void validate () const

Check if index of selected value is valid.

•keyword_list_t ()

Constructor.

Private Attributes

• size_t index

Index into list.

• std::vector<std::string >entries

List of valid entries.

4.95.1 Detailed Description

The stucture keyword_list_t (p. 212) deﬁnes a keyword list (vector of strings) with an index into

the list. Used as MHAParser::kw_t (p. 214), it can be used to access a set of valid keywords

through the parser (i.e. one of "pear apple banana").

4.95.2 Member Function Documentation

4.95.2.1 void MHAParser::keyword_list_t::set_value ( const std::string & s)

This function selects a value from the keyword list. The index is set to the last matching entry.

Parameters

sValue to be selected.

4.95.2.2 void MHAParser::keyword_list_t::set_entries ( const std::string & s)

With this function, the keyword list can be set from a space separated string list.

Parameters

sSpace separated entry list.

214 CONTENTS

4.96 MHAParser::kw_t Class Reference

Variable with keyword list value.

Inheritance diagram for MHAParser::kw_t:

MHAParser::kw_t

MHAParser::commit_t

< receiver_t >

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•kw_t (const std::string &, const std::string &, const std::string &)

Constructor of a keyword list openMHA conﬁguration variable.

•kw_t (const kw_t &)

Copy constructor.

• void set_range (const std::string &)

Set/change the list of valid entries.

• bool isval (const std::string &) const

Test if the given value is selected.

Public Attributes

•keyword_list_t data

Variable data in its native type.

4.97 MHAParser::mcomplex_mon_t Class Reference 215

4.96.1 Constructor & Destructor Documentation

4.96.1.1 MHAParser::kw_t::kw_t ( const std::string & h, const std::string & v, const std::string & rg )

Parameters

hA help string describing the purpose of this variable.

vThe initial value, has to be a value from the list of possible values given in the last

parameter.

rg A string containing the list of valid entries. The entries have to be separated by spaces.

The list of entries has to be delimited by brackets "[", "]".

4.96.2 Member Function Documentation

4.96.2.1 void MHAParser::kw_t::set_range ( const std::string & r)

Parameters

rA string containing the list of valid entries. The entries have to be separated by spaces.

The list of entries has to be delimited by brackets "[", "]".

4.97 MHAParser::mcomplex_mon_t Class Reference

Matrix of complex numbers monitor.

Inheritance diagram for MHAParser::mcomplex_mon_t:

MHAParser::mcomplex

_mon_t

MHAParser::monitor_t

MHAParser::base_t

216 CONTENTS

Public Member Functions

•mcomplex_mon_t (const std::string &hlp)

Create a matrix of complex ﬂoating point monitor values.

Public Attributes

• std::vector<std::vector<mha_complex_t > > data

Data ﬁeld.

4.97.1 Constructor & Destructor Documentation

4.97.1.1 MHAParser::mcomplex_mon_t::mcomplex_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.98 MHAParser::mcomplex_t Class Reference

Matrix variable with complex value.

4.99 MHAParser::mﬂoat_mon_t Class Reference 217

Inheritance diagram for MHAParser::mcomplex_t:

MHAParser::mcomplex_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Attributes

• std::vector<std::vector<mha_complex_t > > data

Data ﬁeld.

Additional Inherited Members

4.99 MHAParser::mﬂoat_mon_t Class Reference

Matrix of ﬂoats monitor.

218 CONTENTS

Inheritance diagram for MHAParser::mﬂoat_mon_t:

MHAParser::mfloat_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•mﬂoat_mon_t (const std::string &hlp)

Create a matrix of ﬂoating point monitor values.

Public Attributes

• std::vector<std::vector<ﬂoat > > data

Data ﬁeld.

4.99.1 Constructor & Destructor Documentation

4.99.1.1 MHAParser::mﬂoat_mon_t::mﬂoat_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.100 MHAParser::mﬂoat_t Class Reference

Matrix variable with ﬂoat value.

4.100 MHAParser::mﬂoat_t Class Reference 219

Inheritance diagram for MHAParser::mﬂoat_t:

MHAParser::mfloat_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•mﬂoat_t (const std::string &, const std::string &, const std::string &="")

Create a ﬂoat matrix parser variable.

Public Attributes

• std::vector<std::vector<ﬂoat > > data

Data ﬁeld.

Additional Inherited Members

4.100.1 Constructor & Destructor Documentation

4.100.1.1 MHAParser::mﬂoat_t::mﬂoat_t ( const std::string & h, const std::string & v, const std::string

&rg = "" )

220 CONTENTS

Parameters

hA human-readable text describing the purpose of this conﬁguration variable.

vThe initial value of the variable, as a string, in openMHA conﬁguration language: (e.g. "[[0

1]; [2 3]]" for a matrix), described in the "Multidimensional Variables" s2.1.3 section of the

openMHA User Manual.

rg The numeric range to enforce on all members of the matrix.

4.101 MHAParser::mhapluginloader_t Class Reference

Class to create a plugin loader in a parser, including the load logic.

4.102 MHAParser::monitor_t Class Reference

Base class for monitors and variable nodes.

Inheritance diagram for MHAParser::monitor_t:

MHAParser::monitor_t

MHAParser::bool_mon_t

MHAParser::complex

_mon_t

MHAParser::float_mon_t

MHAParser::int_mon_t

MHAParser::mcomplex

_mon_t

MHAParser::mfloat_mon_t

MHAParser::string_mon_t

MHAParser::variable_t

MHAParser::vcomplex

_mon_t

MHAParser::vfloat_mon_t

MHAParser::vint_mon_t

MHAParser::vstring

_mon_t

MHAParser::base_t

MHAParser::bool_t

MHAParser::kw_t

MHAParser::range_var_t

MHAParser::string_t

MHAParser::vstring_t

MHAParser::commit_t

< receiver_t >

MHAParser::complex_t

MHAParser::float_t

MHAParser::int_t

MHAParser::mcomplex_t

MHAParser::mfloat_t

MHAParser::vcomplex_t

MHAParser::vfloat_t

MHAParser::vint_t

Additional Inherited Members

4.103 MHAParser::parser_t Class Reference

Parser node class.

4.103 MHAParser::parser_t Class Reference 221

Inheritance diagram for MHAParser::parser_t:

MHAParser::parser_t

AuditoryProfile::parser_t

DynComp::dc_afterburn

_vars_t

io_file_t

io_parser_t

io_tcp_parser_t

MHAFilter::adapt_filter_t

MHAFilter::iir_filter_t

MHAIOJack::io_jack_t

MHAParser::window_t

MHAPlugin::plugin_t

< runtime_cfg_t >

MHAParser::base_t

DynComp::dc_afterburn_t

Public Member Functions

•parser_t (const std::string &help_text="")

Construct detached node to be used in the conﬁguration tree.

• void insert_item (const std::string &, base_t ∗)

• void remove_item (const std::string &)

Remove an item by name.

• void force_remove_item (const std::string &)

Remove an item by name.

• void remove_item (const base_t ∗)

Remove an item by address.

Private Attributes

• std::string id_string

identiﬁcation string

Additional Inherited Members

4.103.1 Detailed Description

Aparser_t (p. 220) instance is a node in the conﬁguration tree. A parser node can contain any

number of other parser_t (p. 220) instances or conﬁguration language variables. These items

are inserted into a parser node using the parser_t::insert_item (p. 222) method.

222 CONTENTS

4.103.2 Constructor & Destructor Documentation

4.103.2.1 MHAParser::parser_t::parser_t ( const std::string & help_text = "" )

Parameters

help_text A text describing this node. E.g. if this node lives at the root of some openMHA

plugin, then the help text should describe the functionality of the plugin.

4.103.3 Member Function Documentation

4.103.3.1 void MHAParser::parser_t::insert_item ( const std::string & n, MHAParser::base_t ∗e)

This function registers an item under a given name into this sub-parser and makes it accessible

to the parser interface.

Parameters

nName of the item in the conﬁguration tree

eC++ pointer to the item instance. e can either point to a variable, to a monitor, or to

another sub-parser.

4.103.3.2 void MHAParser::parser_t::remove_item ( const std::string & n)

If the item does not exist, an error is being reported.

Parameters

nName of parser item to be removed from list.

4.103.3.3 void MHAParser::parser_t::force_remove_item ( const std::string & n)

Non-existing items are ignored.

Parameters

nName of parser item to be removed from list.

4.103.3.4 void MHAParser::parser_t::remove_item ( const base_t ∗addr )

The item belonging to an address is being removed from the list of items.

4.104 MHAParser::range_var_t Class Reference 223

Parameters

addr Address of parser item to be removed.

4.104 MHAParser::range_var_t Class Reference

Base class for all variables with a numeric value range.

Inheritance diagram for MHAParser::range_var_t:

MHAParser::range_var_t

MHAParser::complex_t

MHAParser::float_t

MHAParser::int_t

MHAParser::mcomplex_t

MHAParser::mfloat_t

MHAParser::vcomplex_t

MHAParser::vfloat_t

MHAParser::vint_t

MHAParser::variable_tMHAParser::monitor_tMHAParser::base_t

Public Member Functions

• void set_range (const std::string &r)

Change the valid range of a variable.

Protected Attributes

• ﬂoat low_limit

Lower limit of range.

• ﬂoat up_limit

Upper limit of range.

• bool low_incl

Lower limit is included (or excluded) in range.

• bool up_incl

Upper limit is included (or excluded) in range.

• bool check_low

Check lower limit.

• bool check_up

Check upper limit.

• bool check_range

Range checking is active.

224 CONTENTS

Additional Inherited Members

4.104.1 Member Function Documentation

4.104.1.1 void MHAParser::range_var_t::set_range ( const std::string & r)

Parameters

rNew range of the variable (string representation)

4.105 MHAParser::string_mon_t Class Reference

Monitor with string value.

Inheritance diagram for MHAParser::string_mon_t:

MHAParser::string_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•string_mon_t (const std::string &hlp)

Create a monitor variable for string values.

Public Attributes

• std::string data

Data ﬁeld.

4.105.1 Constructor & Destructor Documentation

4.105.1.1 MHAParser::string_mon_t::string_mon_t ( const std::string & hlp )

4.106 MHAParser::string_t Class Reference 225

Parameters

hlp A help text describing this monitor variable.

4.106 MHAParser::string_t Class Reference

Variable with a string value.

Inheritance diagram for MHAParser::string_t:

MHAParser::string_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•string_t (const std::string &, const std::string &)

Constructor of a openMHA conﬁguration variable for string values.

Public Attributes

• std::string data

Data ﬁeld.

4.106.1 Constructor & Destructor Documentation

4.106.1.1 MHAParser::string_t::string_t ( const std::string & h, const std::string & v)

226 CONTENTS

Parameters

hA help string describing the purpose of this variable.

vThe initial string value

4.107 MHAParser::variable_t Class Reference

Base class for variable nodes.

Inheritance diagram for MHAParser::variable_t:

MHAParser::variable_t

MHAParser::bool_t

MHAParser::kw_t

MHAParser::range_var_t

MHAParser::string_t

MHAParser::vstring_t

MHAParser::monitor_tMHAParser::base_t

MHAParser::commit_t

< receiver_t >

MHAParser::complex_t

MHAParser::float_t

MHAParser::int_t

MHAParser::mcomplex_t

MHAParser::mfloat_t

MHAParser::vcomplex_t

MHAParser::vfloat_t

MHAParser::vint_t

Public Member Functions

• void setlock (const bool &)

Lock a variable against write access.

Additional Inherited Members

4.107.1 Member Function Documentation

4.107.1.1 void MHAParser::variable_t::setlock ( const bool & b)

Parameters

bLock state

4.108 MHAParser::vcomplex_mon_t Class Reference

Monitor with vector of complex values.

4.109 MHAParser::vcomplex_t Class Reference 227

Inheritance diagram for MHAParser::vcomplex_mon_t:

MHAParser::vcomplex

_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vcomplex_mon_t (const std::string &hlp)

Create a vector of complex monitor values.

Public Attributes

• std::vector<mha_complex_t >data

Data ﬁeld.

4.108.1 Constructor & Destructor Documentation

4.108.1.1 MHAParser::vcomplex_mon_t::vcomplex_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.109 MHAParser::vcomplex_t Class Reference

Vector variable with complex value.

228 CONTENTS

Inheritance diagram for MHAParser::vcomplex_t:

MHAParser::vcomplex_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Attributes

• std::vector<mha_complex_t >data

Data ﬁeld.

Additional Inherited Members

4.110 MHAParser::vﬂoat_mon_t Class Reference

Vector of ﬂoats monitor.

4.111 MHAParser::vﬂoat_t Class Reference 229

Inheritance diagram for MHAParser::vﬂoat_mon_t:

MHAParser::vfloat_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vﬂoat_mon_t (const std::string &hlp)

Create a vector of ﬂoating point monitor values.

Public Attributes

• std::vector<ﬂoat >data

Data ﬁeld.

4.110.1 Constructor & Destructor Documentation

4.110.1.1 MHAParser::vﬂoat_mon_t::vﬂoat_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.111 MHAParser::vﬂoat_t Class Reference

Vector variable with ﬂoat value.

230 CONTENTS

Inheritance diagram for MHAParser::vﬂoat_t:

MHAParser::vfloat_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vﬂoat_t (const std::string &, const std::string &, const std::string &="")

Create a ﬂoat vector parser variable.

Public Attributes

• std::vector<ﬂoat >data

Data ﬁeld.

Additional Inherited Members

4.111.1 Constructor & Destructor Documentation

4.111.1.1 MHAParser::vﬂoat_t::vﬂoat_t ( const std::string & h, const std::string & v, const std::string &

rg = "" )

4.112 MHAParser::vint_mon_t Class Reference 231

Parameters

hA human-readable text describing the purpose of this conﬁguration variable.

vThe initial value of the variable, as a string, in openMHA conﬁguration language:

(e.g. "[0 1 2.1 3]" for a vector), described in the "Multidimensional Variables"

s2.1.3 section of the openMHA User Manual.

rg The numeric range to enforce on all members of the vector.

•

4.112 MHAParser::vint_mon_t Class Reference

Vector of ints monitor.

Inheritance diagram for MHAParser::vint_mon_t:

MHAParser::vint_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vint_mon_t (const std::string &hlp)

Create a vector of integer monitor values.

Public Attributes

• std::vector<int >data

Data ﬁeld.

4.112.1 Constructor & Destructor Documentation

4.112.1.1 MHAParser::vint_mon_t::vint_mon_t ( const std::string & hlp )

232 CONTENTS

Parameters

hlp A help text describing this monitor variable.

4.113 MHAParser::vint_t Class Reference

Variable with vector<int>value.

Inheritance diagram for MHAParser::vint_t:

MHAParser::vint_t

MHAParser::range_var_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vint_t (const std::string &, const std::string &, const std::string &="")

Constructor.

Public Attributes

• std::vector<int >data

Data ﬁeld.

4.114 MHAParser::vstring_mon_t Class Reference 233

Additional Inherited Members

4.113.1 Constructor & Destructor Documentation

4.113.1.1 MHAParser::vint_t::vint_t ( const std::string & h, const std::string & v, const std::string & rg =

"" )

Parameters

hhelp string

vinitial value

rg optional: range constraint for all elements

4.114 MHAParser::vstring_mon_t Class Reference

Vector of monitors with string value.

Inheritance diagram for MHAParser::vstring_mon_t:

MHAParser::vstring

_mon_t

MHAParser::monitor_t

MHAParser::base_t

Public Member Functions

•vstring_mon_t (const std::string &hlp)

Create a vector of string monitor values.

Public Attributes

• std::vector<std::string >data

Data ﬁeld.

234 CONTENTS

4.114.1 Constructor & Destructor Documentation

4.114.1.1 MHAParser::vstring_mon_t::vstring_mon_t ( const std::string & hlp )

Parameters

hlp A help text describing this monitor variable.

4.115 MHAParser::vstring_t Class Reference

Vector variable with string values.

Inheritance diagram for MHAParser::vstring_t:

MHAParser::vstring_t

MHAParser::variable_t

MHAParser::monitor_t

MHAParser::base_t

Public Attributes

• std::vector<std::string >data

Data ﬁeld.

Additional Inherited Members

4.116 MHAParser::window_t Class Reference

MHA conﬁguration interface for a window function generator.

4.116 MHAParser::window_t Class Reference 235

Inheritance diagram for MHAParser::window_t:

MHAParser::window_t

MHAParser::parser_t

MHAParser::base_t

Public Member Functions

•window_t (const std::string &help="Window type conﬁguration.")

Constructor to create parser class.

•MHAWindow::base_t get_window (unsigned int len) const

Create a window instance, use default parameters.

•MHAWindow::base_t get_window (unsigned int len, ﬂoat xmin) const

Create a window instance.

•MHAWindow::base_t get_window (unsigned int len, ﬂoat xmin, ﬂoat xmax) const

Create a window instance.

•MHAWindow::base_t get_window (unsigned int len, ﬂoat xmin, ﬂoat xmax, bool minin-

cluded) const

Create a window instance.

•MHAWindow::base_t get_window (unsigned int len, ﬂoat xmin, ﬂoat xmax, bool minin-

cluded, bool maxincluded) const

Create a window instance.

• MHAParser::window_t::wtype_t get_type () const

Return currently selected window type.

Additional Inherited Members

4.116.1 Detailed Description

This class implements a conﬁguration interface (sub-parser) for window type selection and

user-deﬁned window type. It provides member functions to generate an instance of MHA←-

Window::base_t (p. 278) based on the values provided by the conﬁguration interface.

236 CONTENTS

The conﬁguration interface is derived from MHAParser::parser_t (p. 220) and can thus be

inserted into the conﬁguration tree using the insert_item() (p. 222) method of the parent parser.

If one of the pre-deﬁned window types is used, then the window is generated using the MHA←-

Window::fun_t (p. 280) class constructor; for the user-deﬁned type the values from the "user"

variable are copied.

4.117 MHAPlugin::conﬁg_t<runtime_cfg_t >Class Template Reference

Template class for thread safe conﬁguration.

Inheritance diagram for MHAPlugin::conﬁg_t<runtime_cfg_t >:

MHAPlugin::config_t

< runtime_cfg_t >

MHAPlugin::plugin_t

< runtime_cfg_t >

Protected Member Functions

• runtime_cfg_t ∗poll_conﬁg ()

Receive the latest run time conﬁguration.

• runtime_cfg_t ∗last_conﬁg ()

Receive the latest run time conﬁguration.

• void push_conﬁg (runtime_cfg_t ∗ncfg)

Push a new run time conﬁguration into the conﬁguration ﬁfo.

4.117.1 Detailed Description

template<class runtime_cfg_t>

class MHAPlugin::conﬁg_t<runtime_cfg_t >

This template class provides a mechanism for the handling of thread safe conﬁguration which

is required for run time conﬁguration changes of the openMHA plugins.

4.117 MHAPlugin::conﬁg_t<runtime_cfg_t >Class Template

Reference 237

The template parameter runtime_cfg_t is the run time conﬁguration class of the openMHA

plugin. The constructor of that class should transform the MHAParser (p. 76) variables into

derived runtime conﬁguration. The constructor should fail if the conﬁguration is invalid by any

reason.

A new runtime conﬁguration is provided by the function push_conﬁg() (p. 238). In the process-

ing thread, the actual conﬁguration can be received by a call of poll_conﬁg() (p. 237).

in_use next

configuration

current configuration

in_use next

configuration

Figure 5 Schematic drawing of runtime conﬁguration update: conﬁguration updated,

but not used yet.

in_use next

configuration

current configuration

in_use next

configuration

Figure 6 Schematic drawing of runtime conﬁguration update: conﬁguration in use.

4.117.2 Member Function Documentation

4.117.2.1 template<class runtime_cfg_t >runtime_cfg_t ∗MHAPlugin::conﬁg_t<runtime_cfg_t

>::poll_conﬁg ( ) [protected]

This function stores the latest run time conﬁguration into the protected class member variable

‘cfg'. If no conﬁguration exists, then an exception will be thrown. If no changes occured, then

the value of ‘cfg' will be untouched. This function should be called before any access to the

‘cfg' variable, typically once in each signal processing call.

This function should be only called from the processing thread.

238 CONTENTS

Exceptions

MHA_Error (p. 132)if the resulting runtime conﬁguration is NULL. This usually means

that no push_conﬁg has occured.

4.117.2.2 template<class runtime_cfg_t >runtime_cfg_t ∗MHAPlugin::conﬁg_t<runtime_cfg_t

>::last_conﬁg ( ) [protected]

This function stores the latest run time conﬁguration into the protected class member variable

‘cfg'. If no conﬁguration exists, then an exception will be thrown. If no changes occured, then

the value of ‘cfg' will be untouched. This function may be called instead of poll_conﬁg.

The difference between poll_conﬁg and last_conﬁg is that poll_conﬁg marks previous conﬁg-

urations as ready for deletion, while this function does not. Therefore, memory usage of all

runtime conﬁgurations will accumulate if only this function is called, but it enables safe access

to previous runtime conﬁgurations.

Also, last_conﬁg does not raise an Exception when the latest run time conﬁguration is NULL.

4.117.2.3 template<class runtime_cfg_t>void MHAPlugin::conﬁg_t<runtime_cfg_t

>::push_conﬁg ( runtime_cfg_t ∗ncfg )[protected]

This function adds a new run time conﬁguration. The next time poll_conﬁg (p. 237) is called,

this conﬁguration will be available. Conﬁgurations which are not in use or are outdated will be

removed.

This function should be only called from the conﬁguration thread.

Parameters

ncfg pointer on a new conﬁguration

Warning

The runtime conﬁguration passed to this function will be removed by the internal garbage

collector. Do not free manually.

4.118 MHAPlugin::plugin_t<runtime_cfg_t >Class Template Reference

The template class for C++ openMHA plugins.

4.118 MHAPlugin::plugin_t<runtime_cfg_t >Class Template

Reference 239

Inheritance diagram for MHAPlugin::plugin_t<runtime_cfg_t >:

MHAPlugin::plugin_t

< runtime_cfg_t >

MHAParser::parser_t

MHAParser::base_t

MHAPlugin::config_t

< runtime_cfg_t >

Public Member Functions

•plugin_t (const std::string &, const algo_comm_t &)

Constructor of plugin template.

• bool is_prepared () const

Flag, if the prepare method is successfully called (or currently evaluated)

•mhaconﬁg_t input_cfg () const

Current input channel conﬁguration.

•mhaconﬁg_t output_cfg () const

Current output channel conﬁguration.

Protected Attributes

•mhaconﬁg_t tftype

Member for storage of plugin interface conﬁguration.

•algo_comm_t ac

AC handle of the chain.

Additional Inherited Members

4.118.1 Detailed Description

template<class runtime_cfg_t>

class MHAPlugin::plugin_t<runtime_cfg_t >

240 CONTENTS

Template Parameters

runtime_←-

cfg_t

run-time conﬁguration.

This template class provides thread safe conﬁguration handling and standard methods to be

compatible to the C++ openMHA plugin wrapper macro MHAPLUGIN_CALLBACKS (p. 8).

The template parameter runtime_cfg_t should be the runtime conﬁguration of the plugin.

See MHAPlugin::conﬁg_t (p. 236) for details on the thread safe communication update mech-

anism.

4.118.2 Constructor & Destructor Documentation

4.118.2.1 template<class runtime_cfg_t >MHAPlugin::plugin_t<runtime_cfg_t >::plugin_t (

const std::string & help, const algo_comm_t &iac )

Parameters

help Help comment to provide some general information about the plugin.

iac AC space handle (will be stored into the member variable ac).

4.118.3 Member Data Documentation

4.118.3.1 template<class runtime_cfg_t >mhaconﬁg_t MHAPlugin::plugin_t<runtime_cfg_t

>::tftype [protected]

This member is deﬁned for convenience of the developer. Typically, the actual contents of

mhaconﬁg_t (p. 155) are stored in this member in the prepare() method.

Note

This member is likely to be removed in later versions, use input_cfg() (p. 239) and

output_cfg() (p. 239) instead.

4.118.3.2 template<class runtime_cfg_t >algo_comm_t MHAPlugin::plugin_t<runtime_cfg_t

>::ac [protected]

This variable is initialized in the constructor and can be used by derived plugins to access the

AC space. Its contents should not be modiﬁed.

4.119 mhaserver_t Class Reference 241

4.119 mhaserver_t Class Reference

MHA Framework listening on TCP port for commands.

Inherits fw_t.

Public Member Functions

•mhaserver_t (const std::string &ao, const std::string &af, const std::string &lf)

• virtual std::string received_group (const std::string &line)

A line of text was received from network client.

• virtual void acceptor_started (int status)

Notiﬁcation: "TCP port is open".

• virtual void set_announce_port (unsigned short announce_port)

If set to nonzero, the spawning process has asked to be notiﬁed of the TCP port used by this

process.

• void logstring (const std::string &)

Log a message to log ﬁle.

• int run (unsigned short port, const std::string &_interface)

Accept network connections and act on commands.

4.119.1 Constructor & Destructor Documentation

4.119.1.1 mhaserver_t::mhaserver_t ( const std::string & ao, const std::string & af, const std::string &

lf )

Parameters

ao Acknowledgement string at end of successful command responses

af Achknoledgement string at end of failed command responses

lf File system path of ﬁle to use as log ﬁle. MHA appends.

4.119.2 Member Function Documentation

4.119.2.1 void mhaserver_t::set_announce_port ( unsigned short announce_port )[virtual]

4.119.2.2 int mhaserver_t::run ( unsigned short port, const std::string & _interface )

Calls acceptor_started() (p. 241) when the TCP port is opened. Calls received_group for every

line received.

Returns

exit code that can be used as process exit code

242 CONTENTS

4.120 MHASignal::async_rmslevel_t Class Reference

Class for asynchronous level metering.

Inheritance diagram for MHASignal::async_rmslevel_t:

MHASignal::async_rmslevel_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

•async_rmslevel_t (unsigned int frames, unsigned int channels)

Constructor for level metering class.

• std::vector<ﬂoat >rmslevel () const

Read-only function for querying the current RMS level.

• std::vector<ﬂoat >peaklevel () const

Read-only function for querying the current peak level.

• void process (mha_wave_t ∗s)

Function to store a chunk of audio in the level meter.

Additional Inherited Members

4.120.1 Detailed Description

4.120.2 Constructor & Destructor Documentation

4.120.2.1 MHASignal::async_rmslevel_t::async_rmslevel_t ( unsigned int frames, unsigned int channels

)

Allocate memory for metering. The RMS integration time corresponds to the number of frames

in the buffer.

4.121 MHASignal::delay_t Class Reference 243

Parameters

frames Number of frames to integrate.

channels Number of channels used for level-metering.

4.120.3 Member Function Documentation

4.120.3.1 std::vector<ﬂoat >MHASignal::async_rmslevel_t::rmslevel ( ) const

Returns

Vector of ﬂoats, one value for each channel, containing the RMS level in dB (SPL if cali-

brated properly).

4.120.3.2 std::vector<ﬂoat >MHASignal::async_rmslevel_t::peaklevel ( ) const

Returns

Vector of ﬂoats, one value for each channel, containing the peak level in dB (SPL if cali-

brated properly).

4.120.3.3 void MHASignal::async_rmslevel_t::process ( mha_wave_t ∗s)

Parameters

sAudio chunk (same number of channels required as given in the constructor).

4.121 MHASignal::delay_t Class Reference

Class to realize a simple delay of waveform streams.

Public Member Functions

•delay_t (std::vector<int >delays, unsigned int channels)

Constructor.

•mha_wave_t ∗process (mha_wave_t ∗s)

Processing method.

4.121.1 Constructor & Destructor Documentation

4.121.1.1 MHASignal::delay_t::delay_t ( std::vector<int >delays, unsigned int channels )

244 CONTENTS

Parameters

delays Vector of delays, one entry for each channel.

channels Number of channels expected.

4.121.2 Member Function Documentation

4.121.2.1 mha_wave_t ∗MHASignal::delay_t::process ( mha_wave_t ∗s)

Parameters

sInput waveform fragment, with number of channels provided in constructor.

Returns

Output waveform fragment.

4.122 MHASignal::delay_wave_t Class Reference

Delayline containing wave fragments.

4.122.1 Detailed Description

The delayline contains waveform fragments. The delay can be conﬁgured in integer fragments

(sample delay or sub-sample delay is not possible).

4.123 MHASignal::doublebuffer_t Class Reference

Double-buffering class.

Public Member Functions

•doublebuffer_t (unsigned int nchannels_in, unsigned int nchannels_out, unsigned int

outer_fragsize, unsigned int inner_fragsize)

Constructor of double buffer.

•mha_wave_t ∗outer_process (mha_wave_t ∗s)

Method to pass audio fragments into the inner layer.

Protected Member Functions

• virtual mha_wave_t ∗inner_process (mha_wave_t ∗s)=0

Method to realize inner processing callback.

4.123 MHASignal::doublebuffer_t Class Reference 245

4.123.1 Detailed Description

This class has two layers: The outer layer, with an outer fragment size, and an inner layer,

with its own fragment size. Data is passed into the inner layer through the doublebuffer_t←-

::outr_process() callback. The pure virtual method doublebuffer_t::inner_process() (p. 245)

is called whenever enough data is available.

4.123.2 Constructor & Destructor Documentation

4.123.2.1 MHASignal::doublebuffer_t::doublebuffer_t ( unsigned int nchannels_in, unsigned int

nchannels_out, unsigned int outer_fragsize, unsigned int inner_fragsize )

Parameters

nchannels_in Number of channels at the input (both layers).

nchannels_out Number of channels at the output (both layers).

outer_fragsize Fragment size of the outer layer (e.g., hardware fragment size)

inner_fragsize Fragment size of the inner layer (e.g., software fragment size)

4.123.3 Member Function Documentation

4.123.3.1 mha_wave_t ∗MHASignal::doublebuffer_t::outer_process ( mha_wave_t ∗s)

Parameters

sPointer to input waveform fragment.

Returns

Pointer to output waveform fragment.

4.123.3.2 virtual mha_wave_t∗MHASignal::doublebuffer_t::inner_process ( mha_wave_t ∗s)

[protected],[pure virtual]

To be overwritten by derived classes.

Parameters

sPointer to input waveform fragment.

246 CONTENTS

Returns

Pointer to output waveform fragment.

4.124 MHASignal::hilbert_t Class Reference

Hilbert transformation of a waveform segment.

Inheritance diagram for MHASignal::hilbert_t:

MHASignal::hilbert_t

MHASignal::minphase_t

Public Member Functions

•hilbert_t (unsigned int len)

• void operator() (const mha_wave_t ∗,mha_wave_t ∗)

Apply Hilbert transformation on a waveform segment.

4.124.1 Detailed Description

Returns the imaginary part of the inverse Fourier transformation of the Fourier transformed

input signal with negative frequencies set to zero.

4.124.2 Constructor & Destructor Documentation

4.124.2.1 MHASignal::hilbert_t::hilbert_t ( unsigned int len )

Parameters

len Length of waveform segment

4.125 MHASignal::loop_wavefragment_t Class Reference 247

4.125 MHASignal::loop_wavefragment_t Class Reference

Copy a ﬁxed waveform fragment to a series of waveform fragments of other size.

Inheritance diagram for MHASignal::loop_wavefragment_t:

MHASignal::loop_wavefragment_t

MHASignal::waveform_t

mha_wave_t

Public Types

Public Member Functions

•loop_wavefragment_t (const mha_wave_t &src, bool loop, level_mode_t level_mode,

std::vector<int >channels, unsigned int startpos=0)

Constructor to create an instance of loop_wavefragment_t (p. 247) based on an existing

waveform block.

• void playback (mha_wave_t ∗s, playback_mode_t pmode, mha_wave_t ∗level_pa,

const std::vector<int >&channels)

Add source waveform block to an output block.

• void playback (mha_wave_t ∗s, playback_mode_t pmode, mha_wave_t ∗level_pa)

Add source waveform block to an output block.

• void playback (mha_wave_t ∗s, playback_mode_t pmode)

Add source waveform block to an output block.

Additional Inherited Members

4.125.1 Detailed Description

This class is designed to continously play back a waveform to an output stream, with variable

output block size.

248 CONTENTS

4.125.2 Member Enumeration Documentation

4.125.2.1 enum MHASignal::loop_wavefragment_t::level_mode_t

Enumerator

relative The nominal level is applied as a gain to the source signal.

peak The nominal level is the peak level of source signal in Pascal.

rms The nominal level is the RMS level of the source signal in Pascal.

4.125.2.2 enum MHASignal::loop_wavefragment_t::playback_mode_t

Enumerator

add Add source signal to output stream.

replace Replace output stream by source signal.

input Do nothing, keep output stream (source position is unchanged).

mute Mute output stream (source position is unchanged).

4.125.3 Constructor & Destructor Documentation

4.125.3.1 MHASignal::loop_wavefragment_t::loop_wavefragment_t ( const mha_wave_t &src, bool

loop, level_mode_t level_mode, std::vector<int >channels, unsigned int startpos = 0)

Parameters

src Waveform block to copy data from.

loop Flag whether the block should be looped or played once.

level_mode Conﬁguration of playback level (see

MHASignal::loop_wavefragment_t::level_mode_t (p. 248) for details)

channels Mapping of input to output channels.

startpos Starting position

4.125.4 Member Function Documentation

4.125.4.1 void MHASignal::loop_wavefragment_t::playback ( mha_wave_t ∗s, playback_mode_t

pmode, mha_wave_t ∗level_pa, const std::vector<int >&channels )

Parameters

sOutput block (streamed signal).

pmode Playback mode (add, replace, input, mute).

level_pa Linear output level/gain (depending on level_mode parameter in constructor);

one value for each sample in output block.

4.126 MHASignal::matrix_t Class Reference 249

Parameters

channels Output channels

4.125.4.2 void MHASignal::loop_wavefragment_t::playback ( mha_wave_t ∗s, playback_mode_t

pmode, mha_wave_t ∗level_pa )

Parameters

sOutput block (streamed signal).

pmode Playback mode (add, replace, input, mute).

level_pa Linear output level/gain (depending on level_mode parameter in constructor);

one value for each sample in output block.

4.125.4.3 void MHASignal::loop_wavefragment_t::playback ( mha_wave_t ∗s, playback_mode_t

pmode )

Parameters

sOutput block (streamed signal).

pmode Playback mode (add, replace, input, mute).

4.126 MHASignal::matrix_t Class Reference

n-dimensional matrix with real or complex ﬂoating point values.

Inheritance diagram for MHASignal::matrix_t:

MHASignal::matrix_t

MHA_AC::ac2matrix_t

MHASignal::uint_vector_t

250 CONTENTS

Public Member Functions

•matrix_t (unsigned int nrows, unsigned int ncols, bool b_is_complex=true)

Create a two-dimensional matrix.

•matrix_t (const mha_spec_t &spec)

Create a two-dimensional matrix from a spectrum, copy values.

•matrix_t (const MHASignal::uint_vector_t &size, bool b_is_complex=true)

Create n-dimensional matrix, descriped by size argument.

•matrix_t (const uint8_t ∗buf, unsigned int len)

Construct from memory area.

•MHASignal::matrix_t &operator= (const comm_var_t &v)

Fill matrix with data of an AC variable object.

•comm_var_t get_comm_var ()

Return a AC communication variable pointing to the data of the current matrix.

• unsigned int dimension () const

Return the dimension of the matrix.

• unsigned int size (unsigned int k) const

Return the size of the matrix.

• unsigned int get_nelements () const

Return total number of elements.

• bool is_same_size (const MHASignal::matrix_t &)

Test if matrix has same size as other.

• bool iscomplex () const

Return information about complexity.

•mha_real_t &real (const MHASignal::uint_vector_t &index)

Access real part of an element in a n-dimensional matrix.

•mha_real_t &imag (const MHASignal::uint_vector_t &index)

Access imaginary part of an element in a n-dimensional matrix.

•mha_complex_t &operator() (const MHASignal::uint_vector_t &index)

Access complex value of an element in a n-dimensional matrix.

• const mha_real_t &real (const MHASignal::uint_vector_t &index) const

Access real part of an element in a n-dimensional matrix.

• const mha_real_t &imag (const MHASignal::uint_vector_t &index) const

Access imaginary part of an element in a n-dimensional matrix.

• const mha_complex_t &operator() (const MHASignal::uint_vector_t &index) const

Access complex value of an element in a n-dimensional matrix.

•mha_real_t &real (unsigned int row, unsigned int col)

Access real part of an element in a two-dimensional matrix.

•mha_real_t &imag (unsigned int row, unsigned int col)

Access imaginary part of an element in a two-dimensional matrix.

•mha_complex_t &operator() (unsigned int row, unsigned int col)

Access complex value of an element in a two-dimensional matrix.

• const mha_real_t &real (unsigned int row, unsigned int col) const

Access real part of an element in a two-dimensional matrix.

• const mha_real_t &imag (unsigned int row, unsigned int col) const

4.126 MHASignal::matrix_t Class Reference 251

Access imaginary part of an element in a two-dimensional matrix.

• const mha_complex_t &operator() (unsigned int row, unsigned int col) const

Access complex value of an element in a two-dimensional matrix.

• unsigned int numbytes () const

Return number of bytes needed to store into memory.

• unsigned int write (uint8_t ∗buf, unsigned int len) const

Copy to memory area.

• const mha_real_t ∗get_rdata () const

Return pointer of real data.

• const mha_complex_t ∗get_cdata () const

Return pointer of complex data.

Additional Inherited Members

4.126.1 Detailed Description

Warning

The member functions imag() (p. 253) and operator() should only be called if the matrix

is deﬁned to hold complex values.

4.126.2 Constructor & Destructor Documentation

4.126.2.1 MHASignal::matrix_t::matrix_t ( unsigned int nrows, unsigned int ncols, bool b_is_complex =

true )

Parameters

nrows Number of rows

ncols Number of columns

b_is_complex Add space for complex values

4.126.2.2 MHASignal::matrix_t::matrix_t ( const mha_spec_t &spec )

Parameters

spec Source spectrum structure

4.126.2.3 MHASignal::matrix_t::matrix_t ( const MHASignal::uint_vector_t &size, bool

b_is_complex = true )

252 CONTENTS

Parameters

size Size vector

b_is_complex Add space for complex values

4.126.2.4 MHASignal::matrix_t::matrix_t ( const uint8_t ∗buf, unsigned int len )

Warning

This constructor is not real time safe

4.126.3 Member Function Documentation

4.126.3.1 MHASignal::matrix_t & MHASignal::matrix_t::operator= ( const comm_var_t &v)

Parameters

vSource AC variable (comm_var_t (p. 95))

Note

The type and dimension of the AC variable must match the type and dimension of the

matrix.

4.126.3.2 comm_var_t MHASignal::matrix_t::get_comm_var ( )

Returns

AC variable object (comm_var_t (p. 95)), valid for the life time of the matrix.

4.126.3.3 unsigned int MHASignal::matrix_t::dimension ( ) const [inline]

Returns

Dimension of the matrix

4.126.3.4 unsigned int MHASignal::matrix_t::size ( unsigned int k) const [inline]

Parameters

kDimension

4.126 MHASignal::matrix_t Class Reference 253

Returns

Size of the matrix in dimension k

4.126.3.5 mha_real_t& MHASignal::matrix_t::real ( const MHASignal::uint_vector_t &index )

[inline]

Parameters

index Index vector

4.126.3.6 mha_real_t& MHASignal::matrix_t::imag ( const MHASignal::uint_vector_t &index )

[inline]

Parameters

index Index vector

4.126.3.7 mha_complex_t& MHASignal::matrix_t::operator() ( const MHASignal::uint_vector_t

&index )[inline]

Parameters

index Index vector

4.126.3.8 const mha_real_t& MHASignal::matrix_t::real ( const MHASignal::uint_vector_t &

index ) const [inline]

Parameters

index Index vector

4.126.3.9 const mha_real_t& MHASignal::matrix_t::imag ( const MHASignal::uint_vector_t &

index ) const [inline]

Parameters

index Index vector

4.126.3.10 const mha_complex_t& MHASignal::matrix_t::operator() ( const

MHASignal::uint_vector_t &index ) const [inline]

254 CONTENTS

Parameters

index Index vector

4.126.3.11 mha_real_t& MHASignal::matrix_t::real ( unsigned int row, unsigned int col )[inline]

Parameters

row Row number of element

col Column number of element

4.126.3.12 mha_real_t& MHASignal::matrix_t::imag ( unsigned int row, unsigned int col )

[inline]

Parameters

row Row number of element

col Column number of element

4.126.3.13 mha_complex_t& MHASignal::matrix_t::operator() ( unsigned int row, unsigned int col )

[inline]

Parameters

row Row number of element

col Column number of element

4.126.3.14 const mha_real_t& MHASignal::matrix_t::real ( unsigned int row, unsigned int col ) const

[inline]

Parameters

row Row number of element

col Column number of element

4.126.3.15 const mha_real_t& MHASignal::matrix_t::imag ( unsigned int row, unsigned int col ) const

[inline]

Parameters

row Row number of element

col Column number of element

4.127 MHASignal::minphase_t Class Reference 255

4.126.3.16 const mha_complex_t& MHASignal::matrix_t::operator() ( unsigned int row, unsigned int

col ) const [inline]

Parameters

row Row number of element

col Column number of element

4.127 MHASignal::minphase_t Class Reference

Minimal phase function.

Inheritance diagram for MHASignal::minphase_t:

MHASignal::minphase_t

MHASignal::hilbert_t

Public Member Functions

•minphase_t (unsigned int fftlen, unsigned int ch)

Constructor.

• void operator() (mha_spec_t ∗s)

Transform input spectrum to a minimal-phase spectrum, discarding the original phase.

Additional Inherited Members

4.127.1 Detailed Description

The output spectrum Y(f)is

Y(f) = |X(f)|eiH{log |X(f)|},

with the input spectrum X(f)and the Hilbert transformation H{· · ·}.

4.127.2 Constructor & Destructor Documentation

4.127.2.1 MHASignal::minphase_t::minphase_t ( unsigned int fftlen, unsigned int ch )

256 CONTENTS

Parameters

fftlen FFT length

ch Number of channels

4.127.3 Member Function Documentation

4.127.3.1 void MHASignal::minphase_t::operator() ( mha_spec_t ∗s)

Parameters

sSpectrum to operate on.

4.128 MHASignal::quantizer_t Class Reference

Simple simulation of ﬁxpoint quantization.

Public Member Functions

•quantizer_t (unsigned int num_bits)

Constructor.

• void operator() (mha_wave_t &s)

Quantization of a waveform fragment.

4.128.1 Detailed Description

4.128.2 Constructor & Destructor Documentation

4.128.2.1 MHASignal::quantizer_t::quantizer_t ( unsigned int num_bits )

Parameters

num_bits Number of bits to simulate, or zero for limiting to [-1,1] only.

4.128.3 Member Function Documentation

4.128.3.1 void MHASignal::quantizer_t::operator() ( mha_wave_t &s)

4.129 MHASignal::ringbuffer_t Class Reference 257

Parameters

sWaveform fragment to be quantized.

4.129 MHASignal::ringbuffer_t Class Reference

A ringbuffer class for time domain audio signal, which makes no assumptions with respect to

fragment size.

Inheritance diagram for MHASignal::ringbuffer_t:

MHASignal::ringbuffer_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

•ringbuffer_t (unsigned frames, unsigned channels, unsigned preﬁlled_frames)

Creates new ringbuffer.

• unsigned contained_frames () const

number of currently contained frames

•mha_real_t &value (unsigned frame, unsigned channel)

Access to value stored in ringbuffer.

• void discard (unsigned frames)

Discards the oldest frames Makes room for new write (p. 259), alters base frame index for

value (p. 258).

• void write (mha_wave_t &signal)

Copies the contents of the signal into the ringbuffer if there is space.

Private Attributes

• unsigned next_read_frame_index

identiﬁes place with oldest frame in ringbuffer

• unsigned next_write_frame_index

identiﬁes place to store next frame in ringbuffer

258 CONTENTS

Additional Inherited Members

4.129.1 Detailed Description

4.129.2 Constructor & Destructor Documentation

4.129.2.1 ringbuffer_t::ringbuffer_t ( unsigned frames, unsigned channels, unsigned preﬁlled_frames )

Parameters

frames Size of ringbuffer. Maximum frames can be stored in ringbuffer.

channels Number of audio channels.

preﬁlled_frames Number of frames to be preﬁlled with zero values

Exceptions

MHA_Error (p. 132)if preﬁlled_frames >frames

4.129.3 Member Function Documentation

4.129.3.1 mha_real_t& MHASignal::ringbuffer_t::value ( unsigned frame, unsigned channel )

[inline]

Parameters

frame frame index, 0 corresponds to oldest frame stored.

channel audio channel

Returns

reference to contained sample value

Exceptions

MHA_Error (p. 132)if channel or frame out of bounds.

4.129.3.2 void MHASignal::ringbuffer_t::discard ( unsigned frames )[inline]

Parameters

frames how many frames to discard.

4.130 MHASignal::schroeder_t Class Reference 259

Exceptions

MHA_Error (p. 132)if frames >contained_frames (p. 257)

4.129.3.3 void MHASignal::ringbuffer_t::write ( mha_wave_t &signal )[inline]

Parameters

signal New signal to be appended to the signal already present in the ringbuffer

Exceptions

MHA_Error (p. 132)if there is not enough space or if the channel count mismatches.

4.130 MHASignal::schroeder_t Class Reference

Schroeder tone complex class.

Inheritance diagram for MHASignal::schroeder_t:

MHASignal::schroeder_t

MHASignal::waveform_t

mha_wave_t

Public Types

• typedef ﬂoat(∗groupdelay_t) (ﬂoat f, ﬂoat fmin, ﬂoat fmax)

Function type for group delay deﬁnition.

260 CONTENTS

Public Member Functions

•schroeder_t (unsigned int len, unsigned int channels=1, schroeder_t::sign_t sign=up,

mha_real_t speed=1)

Constructor.

•schroeder_t (unsigned int len, unsigned int channels=1, schroeder_t::groupdelay_t

freqfun=MHASignal::schroeder_t::identity, ﬂoat fmin=0, ﬂoat fmax=1, ﬂoat eps=1e-10)

Construct create Schroeder tone complex from a given frequency function.

Additional Inherited Members

4.130.1 Detailed Description

The Schroeder tone complex is a sweep deﬁned in the sampled spectrum:

Φ(f) = σ2πτ(2f/fs)2α, S(f) = eiΦ(f)

fis the sampled frequency in Hz, σis the sign of the sweep (-1 for up sweep, +1 for down

sweep), τis the sweep duration in samples, fsis the sampling rate in Hz and αis the relative

sweep speed.

4.130.2 Member Typedef Documentation

4.130.2.1 typedef ﬂoat(∗MHASignal::schroeder_t::groupdelay_t) (ﬂoat f, ﬂoat fmin, ﬂoat fmax)

Parameters

fFrequency relative to Nyquist frequency.

fmin Minimum frequency relative to Nyquist frequency.

fmax Maximum frequency relative to Nyquist frequency.

4.130.3 Member Enumeration Documentation

4.130.3.1 enum MHASignal::schroeder_t::sign_t

Enumerator

up Sweep from zero to Nyquist frequency ( σ=−1)

down Sweep from Nyquist frequency to zero ( σ= +1)

4.131 MHASignal::spectrum_t Class Reference 261

4.130.4 Constructor & Destructor Documentation

4.130.4.1 MHASignal::schroeder_t::schroeder_t ( unsigned int len, unsigned int channels = 1,

schroeder_t::sign_t sign = up,mha_real_t speed = 1)

Parameters of the Schroeder tone complex are conﬁgured in the constructor.

Parameters

len Length τof the Schroeder tone complex in samples

channels Number of channels

sign Sign σof Schroeder sweep

speed Relative speed α(curvature of phase function)

4.130.4.2 MHASignal::schroeder_t::schroeder_t ( unsigned int len, unsigned int channels = 1,

schroeder_t::groupdelay_t freqfun = MHASignal::schroeder_t::identity,

ﬂoat fmin = 0,ﬂoat fmax = 1,ﬂoat eps = 1e-10 )

The frequency function g(f)deﬁnes the sweep speed and sign (based on the group delay). It

must be deﬁned in the interval [0,1) and should return values in the interval [0,1].

Φ(f) = −4πτ

g(f)df, S(f) = eiΦ(f)

Parameters

len Length τof the Schroeder tone complex in samples.

channels Number of channels.

freqfun Frequency function g(f).

fmin Start frequency (relative to Nyquist frequency).

fmax End frequency (relative to Nyquist frequency).

eps Stability constant for frequency ranges not covered by Schroeder tone complex.

4.131 MHASignal::spectrum_t Class Reference

a signal processing class for spectral data (based on mha_spec_t (p. 141))

262 CONTENTS

Inheritance diagram for MHASignal::spectrum_t:

MHASignal::spectrum_t

MHA_AC::spectrum_t

mha_spec_t

Public Member Functions

•spectrum_t (const unsigned int &frames, const unsigned int &channels)

constructor of spectrum class

•spectrum_t (const mha_spec_t &)

Copy constructor.

•spectrum_t (const MHASignal::spectrum_t &)

Copy constructor.

•mha_complex_t &operator() (unsigned int f, unsigned int ch)

Access to element.

•mha_complex_t &operator[ ] (unsigned int k)

Access to a single element, direct index into data buffer.

•mha_complex_t &value (unsigned int f, unsigned int ch)

Access to element.

• void copy (const mha_spec_t &)

copy all elements from a spectrum

• void copy_channel (const mha_spec_t &s, unsigned sch, unsigned dch)

Copy one channel of a given spectrum signal to a target channel.

• void export_to (mha_spec_t &)

copy elements to spectrum structure

• void scale (const unsigned int &, const unsigned int &, const unsigned int &, const mha←-

_real_t &)

scale section [a,b) in channel "ch" by "val"

• void scale_channel (const unsigned int &, const mha_real_t &)

scale all elements in one channel

4.131 MHASignal::spectrum_t Class Reference 263

Additional Inherited Members

4.131.1 Constructor & Destructor Documentation

4.131.1.1 spectrum_t::spectrum_t ( const unsigned int & frames, const unsigned int & channels )

Allocates buffers and initializes memory to zeros.

Parameters

frames number of frames (fft bins) in one channel. Number of Frames is usually fftlen /

2+1

channels number of channels

4.131.2 Member Function Documentation

4.131.2.1 mha_complex_t& MHASignal::spectrum_t::operator() ( unsigned int f, unsigned int ch )

[inline]

Parameters

fBin number

ch Channel number

Returns

Reference to element

4.131.2.2 mha_complex_t& MHASignal::spectrum_t::operator[ ] ( unsigned int k)[inline]

Parameters

kBuffer index

Returns

Reference to element

4.131.2.3 mha_complex_t& MHASignal::spectrum_t::value ( unsigned int f, unsigned int ch )

[inline]

264 CONTENTS

Parameters

fBin number

ch Channel number

Returns

Reference to element

4.131.2.4 void spectrum_t::copy ( const mha_spec_t &src )

Parameters

src input spectrum

4.131.2.5 void spectrum_t::copy_channel ( const mha_spec_t &s, unsigned sch, unsigned dch )

Parameters

sInput spectrum signal

sch Channel index in source signal

dch Channel index in destination (this) signal

4.131.2.6 void spectrum_t::export_to ( mha_spec_t &dest )

Parameters

dest destination spectrum structure

4.131.2.7 void spectrum_t::scale ( const unsigned int & a, const unsigned int & b, const unsigned int &

ch, const mha_real_t &val )

Parameters

astarting frame

bend frame (excluded)

ch channel number

val scale factor

4.131.2.8 void spectrum_t::scale_channel ( const unsigned int & ch, const mha_real_t &src )

4.132 MHASignal::subsample_delay_t Class Reference 265

Parameters

ch channel number

src scale factor

4.132 MHASignal::subsample_delay_t Class Reference

implements subsample delay in spectral domain.

Public Member Functions

•subsample_delay_t (const std::vector<ﬂoat >&subsample_delay, unsigned fftlen)

Constructor computes complex phase factors to apply to achieve subsample delay.

• void process (mha_spec_t ∗s)

Apply the phase_gains to s to achieve the subsample delay.

• void process (mha_spec_t ∗s, unsigned idx)

Apply the pase gains to channel idx in s to achieve the subsample delay in channel idx.

Public Attributes

•spectrum_t phase_gains

The complex factors to apply to achieve the necessary phase shift.

Private Attributes

• unsigned last_complex_bin

index of the last complex fft bin for the used fft length.

4.132.1 Detailed Description

When transformed back to the time domain, the signal is delayed by the conﬁgured fraction of

a sample. This operation must not be used in a smoothgains bracket.

4.132.2 Constructor & Destructor Documentation

4.132.2.1 MHASignal::subsample_delay_t::subsample_delay_t ( const std::vector<ﬂoat >&

subsample_delay, unsigned fftlen )

266 CONTENTS

Parameters

subsample_delay The subsample delay to apply. -0.5 <= subsample_delay <= 0.5

fftlen FFT length

Exceptions

MHA_Error (p. 132)if the parameters are out of range

4.132.3 Member Function Documentation

4.132.3.1 void MHASignal::subsample_delay_t::process ( mha_spec_t ∗s, unsigned idx )

Parameters

ssignal

idx channel index, 0-based

Exceptions

MHA_Error (p. 132)if idx >= s->num_channels

4.132.4 Member Data Documentation

4.132.4.1 unsigned MHASignal::subsample_delay_t::last_complex_bin [private]

4.133 MHASignal::uint_vector_t Class Reference

Vector of unsigned values, used for size and index description of n-dimensional matrixes.

4.133 MHASignal::uint_vector_t Class Reference 267

Inheritance diagram for MHASignal::uint_vector_t:

MHASignal::uint_vector_t

MHASignal::matrix_t

MHA_AC::ac2matrix_t

Public Member Functions

•uint_vector_t (unsigned int len)

Constructor, initializes all elements to zero.

•uint_vector_t (const uint8_t ∗buf, unsigned int len)

Construct from memory area.

• bool operator== (const uint_vector_t &) const

Check for equality.

•uint_vector_t &operator= (const uint_vector_t &)

Assign from other uint_vector_t (p. 266).

• unsigned int get_length () const

Return the length of the vector.

• const uint32_t & operator[ ] (unsigned int k) const

Read-only access to elements.

• uint32_t & operator[ ] (unsigned int k)

Access to elements.

• unsigned int numbytes () const

Return number of bytes needed to store into memory.

• unsigned int write (uint8_t ∗buf, unsigned int len) const

Copy to memory area.

• const uint32_t ∗getdata () const

Return pointer to the data ﬁeld.

4.133.1 Constructor & Destructor Documentation

4.133.1.1 MHASignal::uint_vector_t::uint_vector_t ( unsigned int len )

268 CONTENTS

Parameters

len Length of vector.

4.133.1.2 MHASignal::uint_vector_t::uint_vector_t ( const uint8_t ∗buf, unsigned int len )

Warning

This constructor is not real time safe

4.133.2 Member Function Documentation

4.133.2.1 uint_vector_t & MHASignal::uint_vector_t::operator= ( const uint_vector_t &src )

Warning

This assignment will fail if the lengths mismatch.

4.133.2.2 unsigned int MHASignal::uint_vector_t::numbytes ( ) const

4.133.2.3 unsigned int MHASignal::uint_vector_t::write ( uint8_t ∗buf, unsigned int len ) const

4.133.2.4 const uint32_t∗MHASignal::uint_vector_t::getdata ( ) const [inline]

4.134 MHASignal::waveform_t Class Reference

signal processing class for waveform data (based on mha_wave_t (p. 154))

Inheritance diagram for MHASignal::waveform_t:

MHASignal::waveform_t

MHA_AC::waveform_t

MHAFilter::o1_ar_filter_t

MHAOvlFilter::fftfb_t

MHASignal::async_rmslevel_t

MHASignal::loop_wavefragment_t

MHASignal::ringbuffer_t

MHASignal::schroeder_t

MHAWindow::base_t

mha_wave_t

MHAFilter::o1flt_lowpass_t

MHAFilter::o1flt_maxtrack_t

MHAFilter::o1flt_mintrack_t

MHAOvlFilter::overlap

_save_filterbank_t

MHAWindow::fun_t

MHAWindow::user_t

MHAWindow::bartlett_t

MHAWindow::blackman_t

MHAWindow::hamming_t

MHAWindow::hanning_t

MHAWindow::rect_t

4.134 MHASignal::waveform_t Class Reference 269

Public Member Functions

•waveform_t (const unsigned int &frames, const unsigned int &channels)

constructor of waveform_t (p. 268)

•waveform_t (const mhaconﬁg_t &cf)

Constructor to create a waveform from plugin conﬁguration.

•waveform_t (const mha_wave_t &src)

Copy contructor for mha_wave_t (p. 154) source.

•waveform_t (const MHASignal::waveform_t &src)

Copy contructor.

•waveform_t (const std::vector<mha_real_t >&src)

Copy contructor for std::vector<mha_real_t>source.

•mha_real_t &value (unsigned int t, unsigned int ch)

Element accessor.

•mha_real_t &operator() (unsigned int t, unsigned int ch)

Element accessor.

• const mha_real_t &value (unsigned int t, unsigned int ch) const

Constant element accessor.

• const mha_real_t &operator() (unsigned int t, unsigned int ch) const

Constant element accessor.

•mha_real_t sum (const unsigned int &a, const unsigned int &b)

sum of all elements between [a,b) in all channels

•mha_real_t sum (const unsigned int &a, const unsigned int &b, const unsigned int &ch)

sum of all elements between [a,b) in channel ch

•mha_real_t sum ()

sum of all elements

•mha_real_t sumsqr ()

sum of square of all elements

•mha_real_t sum_channel (const unsigned int &)

return sum of all elements in one channel

• void assign (const unsigned int &k, const unsigned int &ch, const mha_real_t &val)

set frame "k" in channel "ch" to value "val"

• void assign (const mha_real_t &)

set all elements to value

• void assign_frame (const unsigned int &k, const mha_real_t &val)

assign value "val" to frame k in all channels

• void assign_channel (const unsigned int &c, const mha_real_t &val)

assign value "val" to channel ch in all frames

• void copy (const mha_wave_t &)

copy data from source into current waveform

• void copy_channel (const mha_wave_t &, unsigned int, unsigned int)

Copy one channel of a given waveform signal to a target channel.

• void copy_from_at (unsigned int, unsigned int, const mha_wave_t &, unsigned int)

Copy part of the source signal into part of this waveform object.

• void export_to (mha_wave_t &)

270 CONTENTS

copy data into allocated mha_wave_t (p. 154) structure

• void limit (const mha_real_t &min, const mha_real_t &max)

limit target to range [min,max]

• void power (const waveform_t &)

transform waveform signal (in Pa) to squared signal (in W/m∧2)

• void powspec (const mha_spec_t &)

get the power spectrum (in W/m∧2) from a complex spectrum

• void scale (const unsigned int &a, const unsigned int &b, const unsigned int &ch, const

mha_real_t &val)

scale section [a,b) in channel "ch" by "val"

• void scale (const unsigned int &k, const unsigned int &ch, const mha_real_t &val)

scale one element

• void scale_channel (const unsigned int &, const mha_real_t &)

scale one channel of target with a scalar

Additional Inherited Members

4.134.1 Constructor & Destructor Documentation

4.134.1.1 waveform_t::waveform_t ( const unsigned int & frames, const unsigned int & channels )

Allocates buffer memory and initializes values to zero.

Parameters

frames number of frames in each channel

channels number of channels

4.134.1.2 waveform_t::waveform_t ( const mhaconﬁg_t &cf )[explicit]

Parameters

cf Plugin conﬁguration

4.134.1.3 waveform_t::waveform_t ( const std::vector<mha_real_t >&src )

A waveform structure with a single channel is created, the length is equal to the number of

elements in the source vector.

4.134.2 Member Function Documentation

4.134.2.1 mha_real_t& MHASignal::waveform_t::value ( unsigned int t, unsigned int ch )

[inline]

4.134 MHASignal::waveform_t Class Reference 271

Parameters

tFrame number

ch Channel number

Returns

Reference to element

4.134.2.2 mha_real_t& MHASignal::waveform_t::operator() ( unsigned int t, unsigned int ch )

[inline]

Parameters

tFrame number

ch Channel number

Returns

Reference to element

4.134.2.3 const mha_real_t& MHASignal::waveform_t::value ( unsigned int t, unsigned int ch ) const

[inline]

Parameters

tFrame number

ch Channel number

Returns

Reference to element

4.134.2.4 const mha_real_t& MHASignal::waveform_t::operator() ( unsigned int t, unsigned int ch )

const [inline]

Parameters

tFrame number

ch Channel number

Returns

Reference to element

272 CONTENTS

4.134.2.5 mha_real_t waveform_t::sum ( const unsigned int & a, const unsigned int & b)

Parameters

astarting frame

bend frame (excluded)

Returns

sum

4.134.2.6 mha_real_t waveform_t::sum ( const unsigned int & a, const unsigned int & b, const

unsigned int & ch )

Parameters

astarting frame

bend frame (exluded)

ch channel number

Returns

sum

4.134.2.7 mha_real_t waveform_t::sum ( )

Returns

sum of all elements

4.134.2.8 mha_real_t waveform_t::sumsqr ( )

Returns

sum of square of all elements

4.134.2.9 mha_real_t waveform_t::sum_channel ( const unsigned int & ch )

Parameters

ch channel number

4.134 MHASignal::waveform_t Class Reference 273

Returns

sum

4.134.2.10 void waveform_t::assign ( const unsigned int & k, const unsigned int & ch, const

mha_real_t &val )

Parameters

kframe number

ch channel number

val new value

4.134.2.11 void waveform_t::assign ( const mha_real_t &val )

Parameters

val new value

4.134.2.12 void waveform_t::assign_frame ( const unsigned int & k, const mha_real_t &val )

Parameters

kframe number

val new value

4.134.2.13 void waveform_t::assign_channel ( const unsigned int & ch, const mha_real_t &val )

Parameters

ch channel number

val new value

4.134.2.14 void waveform_t::copy ( const mha_wave_t &src )

Parameters

src input data (need to be same size as target)

4.134.2.15 void waveform_t::copy_channel ( const mha_wave_t &src, unsigned int src_channel,

unsigned int dest_channel )

274 CONTENTS

Parameters

src Input waveform signal

src_channel Channel in source signal

dest_channel Channel number in destination signal

4.134.2.16 void waveform_t::copy_from_at ( unsigned int to_pos, unsigned int len, const

mha_wave_t &src, unsigned int from_pos )

Source and target have to have the same number of channels.

Parameters

to_pos Offset in target

len Number of frames copied

src Source

from_pos Offset in source

4.134.2.17 void waveform_t::export_to ( mha_wave_t &dest )

Parameters

dest destination structure

4.134.2.18 void waveform_t::limit ( const mha_real_t &min, const mha_real_t &max )

Parameters

min lower limit

max upper limit

4.134.2.19 void waveform_t::power ( const waveform_t &src )

Parameters

src linear waveform signal (in Pa)

4.134.2.20 void waveform_t::powspec ( const mha_spec_t &src )

Parameters

src complex spectrum (normalized to Pa)

4.135 MHATableLookup::xy_table_t Class Reference 275

4.134.2.21 void waveform_t::scale ( const unsigned int & a, const unsigned int & b, const unsigned int

&ch, const mha_real_t &val )

Parameters

astarting frame

bend frame (excluded)

ch channel number

val scale factor

4.134.2.22 void waveform_t::scale ( const unsigned int & k, const unsigned int & ch, const

mha_real_t &val )

Parameters

kframe number

ch channel number

val scale factor

4.134.2.23 void waveform_t::scale_channel ( const unsigned int & ch, const mha_real_t &src )

Parameters

ch channel number

src factor

4.135 MHATableLookup::xy_table_t Class Reference

Class for interpolation with non-equidistant x values.

Inherits MHATableLookup::table_t.

Inherited by MHAOvlFilter::barkscale::bark2hz_t, and MHAOvlFilter::barkscale::hz2bark_t.

Public Member Functions

•mha_real_t lookup (mha_real_t x) const

Return the y-value at the position of the nearest x value below input.

•mha_real_t interp (mha_real_t x) const

Linear interpolation function.

• void add_entry (mha_real_t x, mha_real_t y)

Add a single x-y pair entry.

• void add_entry (mha_real_t ∗pVX, mha_real_t ∗pVY, unsigned int len)

276 CONTENTS

Add multiple entries at once.

• void clear ()

Clear the table and transformation functions.

• void set_xfun (ﬂoat(∗pXFun)(ﬂoat))

Set transformation function for x values.

• void set_yfun (ﬂoat(∗pYFun)(ﬂoat))

Set transformation function for y values during insertion.

• void set_xyfun (ﬂoat(∗pYFun)(ﬂoat, ﬂoat))

Set transformation function for y values during insertion, based on x and y values.

4.135.1 Detailed Description

Linear interpolation of the x-y table is performed. A transformation of x and y-values is possible;

if a transformation function is provided for the x-values, the same function is applied to the ar-

gument of xy_table_t::interp() (p. 276) and xy_table_t::lookup() (p. 276). The transformation

of y values is applied only during insertion into the table. Two functions for y-transformation

can be provided: a simple transformation which depends only on the y values, or a transfor-

mation which takes both (non-transformed) x and y value as an argument. The two-argument

transformation is applied before the one-argument transformation.

4.135.2 Member Function Documentation

4.135.2.1 mha_real_t xy_table_t::lookup ( mha_real_t x) const

Parameters

xInput value

Returns

y value at nearest x value below input.

4.135.2.2 mha_real_t xy_table_t::interp ( mha_real_t x) const

Parameters

xx value

Returns

interpolated y value

4.135.2.3 void xy_table_t::add_entry ( mha_real_t x, mha_real_t y)

4.136 MHAWindow::bartlett_t Class Reference 277

Parameters

xx value

ycorresponding y value

4.135.2.4 void xy_table_t::add_entry ( mha_real_t ∗pVX, mha_real_t ∗pVY, unsigned int uLength )

Parameters

pVX array of x values

pVY array of y values

uLength Length of x and y arrays

4.135.2.5 void xy_table_t::set_xfun ( ﬂoat(∗)(ﬂoat) fun )

Parameters

fun Transformation function.

4.135.2.6 void xy_table_t::set_yfun ( ﬂoat(∗)(ﬂoat) fun )

Parameters

fun Transformation function.

4.135.2.7 void xy_table_t::set_xyfun ( ﬂoat(∗)(ﬂoat, ﬂoat) fun )

Parameters

fun Transformation function.

4.136 MHAWindow::bartlett_t Class Reference

Bartlett window.

278 CONTENTS

Inheritance diagram for MHAWindow::bartlett_t:

MHAWindow::bartlett_t

MHAWindow::fun_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Additional Inherited Members

4.137 MHAWindow::base_t Class Reference

Common base for window types.

Inheritance diagram for MHAWindow::base_t:

MHAWindow::base_t

MHAWindow::fun_t

MHAWindow::user_t

MHASignal::waveform_tmha_wave_t

MHAWindow::bartlett_t

MHAWindow::blackman_t

MHAWindow::hamming_t

MHAWindow::hanning_t

MHAWindow::rect_t

4.138 MHAWindow::blackman_t Class Reference 279

Public Member Functions

•base_t (unsigned int len)

Constructor.

•base_t (const MHAWindow::base_t &src)

Copy constructor.

• void operator() (mha_wave_t &) const

Apply window to waveform segment (reference)

• void operator() (mha_wave_t ∗) const

Apply window to waveform segment (pointer)

• void ramp_begin (mha_wave_t &) const

Apply a ramp at the begining.

• void ramp_end (mha_wave_t &) const

Apply a ramp at the end.

Additional Inherited Members

4.137.1 Constructor & Destructor Documentation

4.137.1.1 MHAWindow::base_t::base_t ( unsigned int len )

Parameters

len Window length in samples.

4.137.1.2 MHAWindow::base_t::base_t ( const MHAWindow::base_t &src )

Parameters

src Source to be copied

4.138 MHAWindow::blackman_t Class Reference

Blackman window.

280 CONTENTS

Inheritance diagram for MHAWindow::blackman_t:

MHAWindow::blackman_t

MHAWindow::fun_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Additional Inherited Members

4.139 MHAWindow::fun_t Class Reference

Generic window based on a generator function.

Inheritance diagram for MHAWindow::fun_t:

MHAWindow::fun_t

MHAWindow::bartlett_t

MHAWindow::blackman_t

MHAWindow::hamming_t

MHAWindow::hanning_t

MHAWindow::rect_t

MHAWindow::base_tMHASignal::waveform_tmha_wave_t

Public Member Functions

•fun_t (unsigned int n, ﬂoat(∗fun)(ﬂoat), ﬂoat xmin=-1, ﬂoat xmax=1, bool min_←-

included=true, bool max_included=false)

Constructor.

4.140 MHAWindow::hamming_t Class Reference 281

Additional Inherited Members

4.139.1 Detailed Description

The generator function should return a valid window function in the interval [-1,1[.

4.139.2 Constructor & Destructor Documentation

4.139.2.1 MHAWindow::fun_t::fun_t ( unsigned int n, ﬂoat(∗)(ﬂoat) fun, ﬂoat xmin = -1,ﬂoat xmax = 1,

bool min_included = true,bool max_included = false )

Parameters

nWindow length

fun Generator function, i.e. MHAWindow::hanning() (p. 305)

xmin Start value of window, i.e. -1 for full window or 0 for fade-out ramp.

xmax Last value of window, i.e. 1 for full window

min_included Flag if minimum value is included

max_included Flag if maximum value is included

4.140 MHAWindow::hamming_t Class Reference

Hamming window.

282 CONTENTS

Inheritance diagram for MHAWindow::hamming_t:

MHAWindow::hamming_t

MHAWindow::fun_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Additional Inherited Members

4.141 MHAWindow::hanning_t Class Reference

von-Hann window

4.142 MHAWindow::rect_t Class Reference 283

Inheritance diagram for MHAWindow::hanning_t:

MHAWindow::hanning_t

MHAWindow::fun_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Additional Inherited Members

4.142 MHAWindow::rect_t Class Reference

Rectangular window.

284 CONTENTS

Inheritance diagram for MHAWindow::rect_t:

MHAWindow::rect_t

MHAWindow::fun_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Additional Inherited Members

4.143 MHAWindow::user_t Class Reference

User deﬁned window.

4.144 PluginLoader::fourway_processor_t Class Reference 285

Inheritance diagram for MHAWindow::user_t:

MHAWindow::user_t

MHAWindow::base_t

MHASignal::waveform_t

mha_wave_t

Public Member Functions

•user_t (const std::vector<mha_real_t >&wnd)

Constructor.

Additional Inherited Members

4.143.1 Constructor & Destructor Documentation

4.143.1.1 MHAWindow::user_t::user_t ( const std::vector<mha_real_t >&wnd )

Parameters

wnd User deﬁned window

4.144 PluginLoader::fourway_processor_t Class Reference

This abstract class deﬁnes the interface for classes that implement all types of signal domain

processing supported by the MHA: wave2wave, spec2spec, wave2spec, and spec2wave.

Inherited by PluginLoader::mhapluginloader_t.

286 CONTENTS

Public Member Functions

• virtual void process (mha_wave_t ∗s_in, mha_wave_t ∗∗s_out)=0

Pure waveform processing.

• virtual void process (mha_spec_t ∗s_in, mha_spec_t ∗∗s_out)=0

Pure spectrum processing.

• virtual void process (mha_wave_t ∗s_in, mha_spec_t ∗∗s_out)=0

Signal processing with domain transformation from waveform to spectrum.

• virtual void process (mha_spec_t ∗s_in, mha_wave_t ∗∗s_out)=0

Signal processing with domain transformation from spectrum to waveform.

• virtual void prepare (mhaconﬁg_t &settings)=0

Prepares the processor for signal processing.

• virtual void release ()=0

Resources allocated for signal processing in fourway_processor_t::prepare (p. 287) are re-

leased here in fourway_processor_t::release (p. 287).

• virtual std::string parse (const std::string &query)=0

Parser interface.

• virtual ∼fourway_processor_t ()

Classes with virtual methods need virtual destructor.

4.144.1 Detailed Description

For supporting different output domains for the same input domain, the processing methods

are overloaded with respect to input domain and output domain.

4.144.2 Constructor & Destructor Documentation

4.144.2.1 virtual PluginLoader::fourway_processor_t::∼fourway_processor_t ( ) [inline],

[virtual]

This destructor is empty.

4.144.3 Member Function Documentation

4.144.3.1 virtual void PluginLoader::fourway_processor_t::process ( mha_wave_t ∗s_in,

mha_wave_t ∗∗ s_out )[pure virtual]

Parameters

s_in input waveform signal

s_out output waveform signal

5 File Documentation 287

4.144.3.2 virtual void PluginLoader::fourway_processor_t::process ( mha_spec_t ∗s_in,

mha_spec_t ∗∗ s_out )[pure virtual]

Parameters

s_in input spectrum signal

s_out output spectrum signal

4.144.3.3 virtual void PluginLoader::fourway_processor_t::process ( mha_wave_t ∗s_in,

mha_spec_t ∗∗ s_out )[pure virtual]

Parameters

s_in input waveform signal

s_out output spectrum signal

4.144.3.4 virtual void PluginLoader::fourway_processor_t::process ( mha_spec_t ∗s_in,

mha_wave_t ∗∗ s_out )[pure virtual]

Parameters

s_in input spectrum signal

s_out output waveform signal

4.144.3.5 virtual void PluginLoader::fourway_processor_t::prepare ( mhaconﬁg_t &settings )

[pure virtual]

Parameters

settings domain and dimensions of the signal. The contents of settings may be modiﬁed

by the prepare implementation. Upon calling fourway_processor_t::prepare

(p. 287), settings reﬂects domain and dimensions of the input signal. When

fourway_processor_t::prepare (p. 287) returns, settings reﬂects domain and

dimensions of the output signal.

4.144.3.6 virtual void PluginLoader::fourway_processor_t::release ( ) [pure virtual]

5 File Documentation

5.1 mha.h File Reference

common types for MHA kernel, MHA framework applications and external plugins

288 CONTENTS

Classes

• struct mha_complex_t

Type for complex ﬂoating point values.

• struct mha_direction_t

Channel source direction structure.

• struct mha_channel_info_t

Channel information structure.

• struct mha_wave_t

Waveform signal structure.

• struct mha_spec_t

Spectrum signal structure.

• struct mha_audio_descriptor_t

Description of an audio fragment (planned as a replacement of mhaconﬁg_t (p. 155)).

• struct mha_audio_t

An audio fragment in the openMHA (planned as a replacement of mha_wave_t (p. 154) and

mha_spec_t (p. 141)).

• struct mhaconﬁg_t

MHA prepare conﬁguration structure.

• struct comm_var_t

Algorithm communication variable structure.

• struct algo_comm_t

A reference handle for algorithm communication variables.

Macros

• #deﬁne MHA_CALLBACK_TEST(x)

Test macro to compare function type deﬁnition and declaration.

• #deﬁne MHA_VERSION_MAJOR 4

Major version number of MHA.

• #deﬁne MHA_VERSION_MINOR 5

Minor version number of MHA.

• #deﬁne MHA_VERSION_RELEASE 5

Release number of MHA.

• #deﬁne MHA_VERSION_BUILD 0

Build number of MHA (currently unused)

• #deﬁne MHA_STRUCT_SIZEMATCH (unsigned int)((sizeof(mha_real_t)==4)+2∗(sizeof(mha←-

_complex_t)==8)+4∗(sizeof(mha_wave_t)==8+2∗sizeof(void∗))+8∗(sizeof(mha_spec←-

_t)==8+2∗sizeof(void∗))+16∗(sizeof(mhaconﬁg_t)==24))

Test number for structure sizes.

• #deﬁne MHA_VERSION (unsigned int)((MHA_STRUCT_SIZEMATCH |(MHA_VERS←-

ION_RELEASE << 8) |(MHA_VERSION_MINOR << 16) |(MHA_VERSION_MAJOR

<< 24)))

Full version number of MHA kernel.

5.2 mha_algo_comm.h File Reference 289

• #deﬁne MHA_VERSION_STRING MHA_XSTRF(MHA_VERSION_MAJOR) "." MHA_←-

XSTRF(MHA_VERSION_MINOR)

Version string of MHA kernel (major.minor)

• #deﬁne MHA_RELEASE_VERSION_STRING MHA_XSTRF(MHA_VERSION_MAJOR)

"." MHA_XSTRF(MHA_VERSION_MINOR) "." MHA_XSTRF(MHA_VERSION_RELE←-

ASE)

Version string of MHA kernel (major.minor.release)

Typedefs

• typedef ﬂoat mha_real_t

openMHA type for real numbers

• typedef void ∗mha_fft_t

Handle for an FFT object.

5.2 mha_algo_comm.h File Reference

Header ﬁle for Algorithm Communication.

Classes

• class MHA_AC::spectrum_t

Insert a MHASignal::spectrum_t (p. 261) class into the AC space.

• class MHA_AC::waveform_t

Insert a MHASignal::waveform_t (p. 268) class into the AC space.

• class MHA_AC::int_t

Insert a integer variable into the AC space.

• class MHA_AC::ﬂoat_t

Insert a ﬂoat point variable into the AC space.

• class MHA_AC::double_t

Insert a double precision ﬂoating point variable into the AC space.

• class MHA_AC::ac2matrix_t

Copy AC variable to a matrix.

• class MHA_AC::acspace2matrix_t

Copy all or a subset of all numeric AC variables into an array of matrixes.

Namespaces

•MHA_AC

Functions and classes for Algorithm Communication (AC) support.

290 CONTENTS

Functions

•mha_spec_t MHA_AC::get_var_spectrum (algo_comm_t ac, const std::string &name)

Convert an AC variable into a spectrum.

•mha_wave_t MHA_AC::get_var_waveform (algo_comm_t ac, const std::string

&name)

Convert an AC variable into a waveform.

• int MHA_AC::get_var_int (algo_comm_t ac, const std::string &name)

Return value of an integer scalar AC variable.

• ﬂoat MHA_AC::get_var_ﬂoat (algo_comm_t ac, const std::string &name)

Return value of an ﬂoating point scalar AC variable.

• std::vector<ﬂoat >MHA_AC::get_var_vﬂoat (algo_comm_t ac, const std::string

&name)

Return value of an ﬂoating point vector AC variable as standard vector of ﬂoats.

5.3 mha_defs.h File Reference

Preprocessor deﬁnitions common to all MHA components.

Macros

• #deﬁne M_PI 3.14159265358979323846

Deﬁne pi if it is not deﬁned yet.

• #deﬁne MIN(a, b) (((a)<(b))?(a):(b))

Macro for minimum function.

• #deﬁne MAX(a, b) (((a)>(b))?(a):(b))

Macro for maximum function.

5.3.1 Detailed Description

This ﬁle contains all preprocessor and type deﬁnitions which are common to all Master Hearing

Aid components.

5.4 mha_error.cpp File Reference

Implementation of openMHA error handling.

5.5 mha_ﬁlter.hh File Reference 291

Functions

• unsigned mha_error_helpers::digits (unsigned n)

Compute number of decimal digits required to represent an unsigned integer.

• unsigned mha_error_helpers::snprintf_required_length (const char ∗formatstring,...)

snprintf_required_length Compute the number of bytes (excluding the terminating nul) required

to store the result of an snprintf.

• void mha_debug (const char ∗fmt,...)

Print an info message (stderr on Linux, OutputDebugString in Windows).

5.4.1 Detailed Description

This ﬁle forms a seperate library.

5.4.2 Function Documentation

5.4.2.1 unsigned mha_error_helpers::digits ( unsigned n)

Parameters

nThe unsigned integer that we want to know the number of required decimal digits for.

return The number of decimal digits in n.

5.4.2.2 unsigned mha_error_helpers::snprintf_required_length ( const char ∗formatstring, ... )

Parameters

formatstring The format string with standard printf formatstring

Returns

the number of bytes required by printf without the terminating nul

5.5 mha_ﬁlter.hh File Reference

Header ﬁle for IIR ﬁlter classes.

Classes

• class MHAFilter::ﬁlter_t

Generic IIR ﬁlter class.

292 CONTENTS

• class MHAFilter::diff_t

Differentiator class (non-normalized)

• class MHAFilter::o1_ar_ﬁlter_t

First order attack-release lowpass ﬁlter.

• class MHAFilter::o1ﬂt_lowpass_t

First order low pass ﬁlter.

• class MHAFilter::o1ﬂt_maxtrack_t

First order maximum tracker.

• class MHAFilter::o1ﬂt_mintrack_t

First order minimum tracker.

• class MHAFilter::iir_ﬁlter_t

IIR ﬁlter class wrapper for integration into parser structure.

• class MHAFilter::adapt_ﬁlter_t

Adaptive ﬁlter.

• class MHAFilter::fftﬁlter_t

FFT based FIR ﬁlter implementation.

• class MHAFilter::fftﬁlterbank_t

FFT based FIR ﬁlterbank implementation.

• struct MHAFilter::transfer_function_t

a structure containing a source channel number, a target channel number, and an impulse

response.

• struct MHAFilter::transfer_matrix_t

A sparse matrix of transfer function partitionss.

• class MHAFilter::partitioned_convolution_t

A ﬁlter class for partitioned convolution.

• struct MHAFilter::partitioned_convolution_t::index_t

Bookkeeping class.

• class MHAFilter::smoothspec_t

Smooth spectral gains, create a windowed impulse response.

• class MHAFilter::resampling_ﬁlter_t

Hann shaped low pass ﬁlter for resampling.

• class MHAFilter::polyphase_resampling_t

A class that does polyphase resampling.

• class MHAFilter::blockprocessing_polyphase_resampling_t

A class that does polyphase resampling and takes into account block processing.

• class MHAFilter::iir_ord1_real_t

First order recursive ﬁlter.

Namespaces

•MHAFilter

Namespace for IIR and FIR ﬁlter classes.

5.6 mha_parser.hh File Reference 293

Functions

• void MHAFilter::o1_lp_coeffs (const mha_real_t tau, const mha_real_t fs, mha_real←-

_t &c1, mha_real_t &c2)

Set ﬁrst order ﬁlter coefﬁcients from time constant and sampling rate.

• void MHAFilter::butter_stop_ord1 (double ∗A, double ∗B, double f1, double f2, double

fs)

Setup a ﬁrst order butterworth band stop ﬁlter.

•MHASignal::waveform_t ∗MHAFilter::spec2ﬁr (const mha_spec_t ∗spec, const un-

signed int fftlen, const MHAWindow::base_t &window, const bool minphase)

Create a windowed impulse response/FIR ﬁlter coefﬁcients from a spectrum.

• unsigned MHAFilter::gcd (unsigned a, unsigned b)

greatest common divisor

• double MHAFilter::sinc (double x)

sin(x)/x function, coping with x=0.

• std::pair<unsigned, unsigned >MHAFilter::resampling_factors (ﬂoat source_←-

sampling_rate, ﬂoat target_sampling_rate, ﬂoat factor=1.0f)

Computes rational resampling factor from two sampling rates.

5.6 mha_parser.hh File Reference

Header ﬁle for the MHA-Parser script language.

Classes

• class MHAParser::keyword_list_t

Keyword list class.

• class MHAParser::base_t

Base class for all parser items.

• class MHAParser::parser_t

Parser node class.

• class MHAParser::monitor_t

Base class for monitors and variable nodes.

• class MHAParser::variable_t

Base class for variable nodes.

• class MHAParser::range_var_t

Base class for all variables with a numeric value range.

• class MHAParser::kw_t

Variable with keyword list value.

• class MHAParser::string_t

Variable with a string value.

• class MHAParser::vstring_t

Vector variable with string values.

• class MHAParser::bool_t

294 CONTENTS

Variable with a boolean value ("yes"/"no")

• class MHAParser::int_t

Variable with integer value.

• class MHAParser::ﬂoat_t

Variable with ﬂoat value.

• class MHAParser::complex_t

Variable with complex value.

• class MHAParser::vint_t

Variable with vector<int>value.

• class MHAParser::vﬂoat_t

Vector variable with ﬂoat value.

• class MHAParser::vcomplex_t

Vector variable with complex value.

• class MHAParser::mﬂoat_t

Matrix variable with ﬂoat value.

• class MHAParser::mcomplex_t

Matrix variable with complex value.

• class MHAParser::int_mon_t

Monitor variable with int value.

• class MHAParser::bool_mon_t

Monitor with string value.

• class MHAParser::string_mon_t

Monitor with string value.

• class MHAParser::vstring_mon_t

Vector of monitors with string value.

• class MHAParser::vint_mon_t

Vector of ints monitor.

• class MHAParser::vﬂoat_mon_t

Vector of ﬂoats monitor.

• class MHAParser::mﬂoat_mon_t

Matrix of ﬂoats monitor.

• class MHAParser::ﬂoat_mon_t

Monitor with ﬂoat value.

• class MHAParser::complex_mon_t

Monitor with complex value.

• class MHAParser::vcomplex_mon_t

Monitor with vector of complex values.

• class MHAParser::mcomplex_mon_t

Matrix of complex numbers monitor.

• class MHAParser::commit_t<receiver_t >

Parser variable with event-emission functionality.

5.6 mha_parser.hh File Reference 295

Namespaces

•MHAParser

Name space for the openMHA-Parser conﬁguration language.

•MHAParser::StrCnv

String converter namespace.

Macros

• #deﬁne insert_member(x) insert_item(#x,&x)

Macro to insert a member variable into a parser.

Functions

• void MHAParser::strreplace (std::string &, const std::string &, const std::string &)

string replace function

• void MHAParser::StrCnv::str2val (const std::string &, bool &)

Convert from string.

• void MHAParser::StrCnv::str2val (const std::string &, ﬂoat &)

Convert from string.

• void MHAParser::StrCnv::str2val (const std::string &, mha_complex_t &)

Convert from string.

• void MHAParser::StrCnv::str2val (const std::string &, int &)

Convert from string.

• void MHAParser::StrCnv::str2val (const std::string &, keyword_list_t &)

Convert from string.

• void MHAParser::StrCnv::str2val (const std::string &, std::string &)

Convert from string.

•template<class arg_t >

void MHAParser::StrCnv::str2val (const std::string &s, std::vector<arg_t >&val)

Converter for vector types.

•template<>

void MHAParser::StrCnv::str2val<mha_real_t >(const std::string &s, std::vector<

mha_real_t >&v)

Converter for vector<mha_real_t>with Matlab-style expansion.

•template<class arg_t >

void MHAParser::StrCnv::str2val (const std::string &s, std::vector<std::vector<arg_t

> > &val)

Converter for matrix types.

• std::string MHAParser::StrCnv::val2str (const bool &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const ﬂoat &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const mha_complex_t &)

Convert to string.

296 CONTENTS

• std::string MHAParser::StrCnv::val2str (const int &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const keyword_list_t &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::string &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<ﬂoat >&)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<mha_complex_t >&)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<int >&)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<std::string >&)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<std::vector<ﬂoat > > &)

Convert to string.

• std::string MHAParser::StrCnv::val2str (const std::vector<std::vector<mha_←-

complex_t > > &)

Convert to string.

5.6.1 Macro Deﬁnition Documentation

5.6.1.1 #deﬁne insert_member( x) insert_item(#x,&x)

Parameters

xMember variable to be inserted. Name of member variable will be used as conﬁguration

name.

See also MHAParser::parser_t::insert_item() (p. 222).

5.7 mha_plugin.hh File Reference

Header ﬁle for MHA C++ plugin class templates.

Classes

• class MHAPlugin::conﬁg_t<runtime_cfg_t >

Template class for thread safe conﬁguration.

• class MHAPlugin::plugin_t<runtime_cfg_t >

The template class for C++ openMHA plugins.

5.8 mha_signal.hh File Reference 297

Namespaces

•MHAPlugin

Namespace for openMHA plugin class templates and thread-safe runtime conﬁgurations.

Macros

• #deﬁne MHAPLUGIN_CALLBACKS_PREFIX(preﬁx, classname, indom, outdom)

C++ wrapper macro for the plugin interface.

• #deﬁne MHAPLUGIN_CALLBACKS(plugname, classname, indom, outdom) MHAPLU←-

GIN_CALLBACKS_PREFIX(MHA_STATIC_ ## plugname ## _,classname,indom,outdom)

C++ wrapper macro for the plugin interface.

• #deﬁne MHAPLUGIN_DOCUMENTATION(plugname, cat, doc) MHAPLUGIN_DOCU←-

MENTATION_PREFIX(MHA_STATIC_ ## plugname ## _,cat,doc)

Wrapper macro for the plugin documentation interface.

Functions

•__attribute__ ((unused)) static const char ∗mha_git_commit_hash

store git commit hash in every binary plgin to support reproducible research

5.7.1 Detailed Description

This ﬁle deﬁnes useful macros and template classes for the development of MHA plugins. A

set of macros wraps a C++ interface around the ANSI-C plugin interface. The plugin_t template

class deﬁnes a corresponding C++ class with all required members. This class can make use

of thread safe conﬁgurations (conﬁg_t).

5.8 mha_signal.hh File Reference

Header ﬁle for audio signal handling and processing classes.

Classes

• class MHASignal::spectrum_t

a signal processing class for spectral data (based on mha_spec_t (p. 141))

• class MHASignal::waveform_t

signal processing class for waveform data (based on mha_wave_t (p. 154))

• class MHASignal::doublebuffer_t

Double-buffering class.

• class MHASignal::ringbuffer_t

298 CONTENTS

A ringbuffer class for time domain audio signal, which makes no assumptions with respect to

fragment size.

• class MHASignal::hilbert_t

Hilbert transformation of a waveform segment.

• class MHASignal::minphase_t

Minimal phase function.

• class MHAWindow::base_t

Common base for window types.

• class MHAWindow::fun_t

Generic window based on a generator function.

• class MHAWindow::rect_t

Rectangular window.

• class MHAWindow::bartlett_t

Bartlett window.

• class MHAWindow::hanning_t

von-Hann window

• class MHAWindow::hamming_t

Hamming window.

• class MHAWindow::blackman_t

Blackman window.

• class MHAWindow::user_t

User deﬁned window.

• class MHASignal::delay_wave_t

Delayline containing wave fragments.

• class MHASignal::async_rmslevel_t

Class for asynchronous level metering.

• class MHASignal::uint_vector_t

Vector of unsigned values, used for size and index description of n-dimensional matrixes.

• class MHASignal::matrix_t

n-dimensional matrix with real or complex ﬂoating point values.

• class MHASignal::schroeder_t

Schroeder tone complex class.

• class MHASignal::quantizer_t

Simple simulation of ﬁxpoint quantization.

• class MHASignal::loop_wavefragment_t

Copy a ﬁxed waveform fragment to a series of waveform fragments of other size.

• class MHASignal::delay_t

Class to realize a simple delay of waveform streams.

• class MHASignal::subsample_delay_t

implements subsample delay in spectral domain.

Namespaces

•MHASignal

Namespace for audio signal handling and processing classes.

•MHAWindow

Collection of Window types.

5.8 mha_signal.hh File Reference 299

Functions

• void MHASignal::for_each (mha_wave_t ∗s, mha_real_t(∗fun)(mha_real_t))

Apply a function to each element of a mha_wave_t (p. 154).

•mha_real_t MHASignal::lin2db (mha_real_t x)

Conversion from linear scale to dB (no SPL reference)

•mha_real_t MHASignal::db2lin (mha_real_t x)

Conversion from dB scale to linear (no SPL reference)

•mha_real_t MHASignal::pa2dbspl (mha_real_t x)

Conversion from linear Pascal scale to dB SPL.

•mha_real_t MHASignal::pa22dbspl (mha_real_t x, mha_real_t eps=1e-20f)

Conversion from squared Pascal scale to dB SPL.

•mha_real_t MHASignal::dbspl2pa (mha_real_t x)

Conversion from dB SPL to linear Pascal scale.

•mha_real_t MHASignal::smp2sec (mha_real_t n, mha_real_t srate)

conversion from samples to seconds

•mha_real_t MHASignal::sec2smp (mha_real_t sec, mha_real_t srate)

conversion from seconds to samples

•mha_real_t MHASignal::bin2freq (mha_real_t bin, unsigned fftlen, mha_real_t srate)

conversion from fft bin index to frequency

•mha_real_t MHASignal::freq2bin (mha_real_t freq, unsigned fftlen, mha_real_t srate)

conversion from frequency to fft bin index

•mha_real_t MHASignal::smp2rad (mha_real_t samples, unsigned bin, unsigned fftlen)

conversion from delay in samples to phase shift

•mha_real_t MHASignal::rad2smp (mha_real_t phase_shift, unsigned bin, unsigned

fftlen)

conversion from phase shift to delay in samples

•template<class elem_type >

std::vector<elem_type >MHASignal::dupvec (std::vector<elem_type >vec, unsigned

Duplicate last vector element to match desired size.

•template<class elem_type >

std::vector<elem_type >MHASignal::dupvec_chk (std::vector<elem_type >vec, un-

signed n)

Duplicate last vector element to match desired size, check for dimension.

• bool equal_dim (const mha_wave_t &a, const mha_wave_t &b)

Test for equal dimension of waveform structures.

• bool equal_dim (const mha_wave_t &a, const mhaconﬁg_t &b)

Test for match of waveform dimension with mhaconﬁg structure.

• bool equal_dim (const mha_spec_t &a, const mha_spec_t &b)

Test for equal dimension of spectrum structures.

• bool equal_dim (const mha_spec_t &a, const mhaconﬁg_t &b)

Test for match of spectrum dimension with mhaconﬁg structure.

• bool equal_dim (const mha_wave_t &a, const mha_spec_t &b)

Test for equal dimension of waveform/spectrum structures.

• bool equal_dim (const mha_spec_t &a, const mha_wave_t &b)

300 CONTENTS

Test for equal dimension of waveform/spectrum structures.

• void integrate (mha_wave_t &s)

Numeric integration of a signal vector (real values)

• void integrate (mha_spec_t &s)

Numeric integration of a signal vector (complex values)

• unsigned int size (const mha_wave_t &s)

Return size of a waveform structure.

• unsigned int size (const mha_spec_t &s)

Return size of a spectrum structure.

• unsigned int size (const mha_wave_t ∗s)

Return size of a waveform structure.

• unsigned int size (const mha_spec_t ∗s)

Return size of a spectrum structure.

• void clear (mha_wave_t &s)

Set all values of waveform to zero.

• void clear (mha_wave_t ∗s)

Set all values of waveform to zero.

• void clear (mha_spec_t &s)

Set all values of spectrum to zero.

• void clear (mha_spec_t ∗s)

Set all values of spectrum to zero.

• void assign (mha_wave_t self, mha_real_t val)

Set all values of waveform 'self' to 'val'.

• void assign (mha_wave_t self, const mha_wave_t &val)

Set all values of waveform 'self' to 'val'.

• void assign (mha_spec_t self, const mha_spec_t &val)

Set all values of spectrum 'self' to 'val'.

• void timeshift (mha_wave_t &self, int shift)

Time shift of waveform chunk.

•mha_wave_t range (mha_wave_t s, unsigned int k0, unsigned int len)

Return a time interval from a waveform chunk.

•mha_spec_t channels (mha_spec_t s, unsigned int ch_start, unsigned int nch)

Return a channel interval from a spectrum.

•mha_real_t &value (mha_wave_t ∗s, unsigned int fr, unsigned int ch)

Access an element of a waveform structure.

• const mha_real_t &value (const mha_wave_t ∗s, unsigned int fr, unsigned int ch)

Constant access to an element of a waveform structure.

•mha_complex_t &value (mha_spec_t ∗s, unsigned int fr, unsigned int ch)

Access to an element of a spectrum.

• const mha_complex_t &value (const mha_spec_t ∗s, unsigned int fr, unsigned int ch)

Constant access to an element of a spectrum.

•mha_real_t &value (mha_wave_t &s, unsigned int fr, unsigned int ch)

Access to an element of a waveform structure.

• const mha_real_t &value (const mha_wave_t &s, unsigned int fr, unsigned int ch)

Constant access to an element of a waveform structure.

5.8 mha_signal.hh File Reference 301

•mha_complex_t &value (mha_spec_t &s, unsigned int fr, unsigned int ch)

Access to an element of a spectrum.

• const mha_complex_t &value (const mha_spec_t &s, unsigned int fr, unsigned int ch)

Constant access to an element of a spectrum.

• std::vector<ﬂoat >std_vector_ﬂoat (const mha_wave_t &)

Converts a mha_wave_t (p. 154) structure into a std::vector<ﬂoat>(interleaved order).

• std::vector<std::vector<ﬂoat > > std_vector_vector_ﬂoat (const mha_wave_t &)

Converts a mha_wave_t (p. 154) structure into a std::vector<std::vector<ﬂoat> > (outer vec-

tor represents channels).

• std::vector<std::vector<mha_complex_t > > std_vector_vector_complex (const

mha_spec_t &)

Converts a mha_spec_t (p. 141) structure into a std::vector<std::vector<mha_complex_t> >

(outer vector represents channels).

•mha_wave_t &operator+= (mha_wave_t &, const mha_real_t &)

Addition operator.

•mha_wave_t &operator+= (mha_wave_t &, const mha_wave_t &)

Addition operator.

•mha_wave_t &operator-= (mha_wave_t &, const mha_wave_t &)

Subtraction operator.

•mha_spec_t &operator-= (mha_spec_t &, const mha_spec_t &)

Subtraction operator.

•mha_wave_t &operator∗=(mha_wave_t &, const mha_real_t &)

Element-wise multiplication operator.

•mha_wave_t &operator∗=(mha_wave_t &, const mha_wave_t &)

Element-wise multiplication operator.

•mha_spec_t &operator∗=(mha_spec_t &, const mha_real_t &)

Element-wise multiplication operator.

•mha_spec_t &operator∗=(mha_spec_t &, const mha_wave_t &)

Element-wise multiplication operator.

•mha_spec_t &operator∗=(mha_spec_t &, const mha_spec_t &)

Element-wise multiplication operator.

•mha_spec_t &operator/= (mha_spec_t &, const mha_spec_t &)

Element-wise division operator.

•mha_wave_t &operator/= (mha_wave_t &, const mha_wave_t &)

Element-wise division operator.

•mha_spec_t &operator+= (mha_spec_t &, const mha_spec_t &)

Addition operator.

•mha_spec_t &operator+= (mha_spec_t &, const mha_real_t &)

Addition operator.

•mha_spec_t &safe_div (mha_spec_t &self, const mha_spec_t &v, mha_real_t eps)

In-Place division with lower limit on divisor.

•mha_wave_t &operator∧=(mha_wave_t &self, const mha_real_t &arg)

Exponent operator.

• void MHASignal::copy_channel (mha_spec_t &self, const mha_spec_t &src, unsigned

sch, unsigned dch)

302 CONTENTS

Copy one channel of a source signal.

• void MHASignal::copy_channel (mha_wave_t &self, const mha_wave_t &src, un-

signed src_channel, unsigned dest_channel)

Copy one channel of a source signal.

•mha_real_t MHASignal::rmslevel (const mha_spec_t &s, unsigned int channel, un-

signed int fftlen)

Return RMS level of a spectrum channel.

•mha_real_t MHASignal::colored_intensity (const mha_spec_t &s, unsigned int chan-

nel, unsigned int fftlen, mha_real_t sqfreq_response[ ])

Colored spectrum intensity.

•mha_real_t MHASignal::maxabs (const mha_spec_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t MHASignal::rmslevel (const mha_wave_t &s, unsigned int channel)

Return RMS level of a waveform channel.

•mha_real_t MHASignal::maxabs (const mha_wave_t &s, unsigned int channel)

Find maximal absolute value.

•mha_real_t MHASignal::maxabs (const mha_wave_t &s)

Find maximal absolute value.

•mha_real_t MHASignal::max (const mha_wave_t &s)

Find maximal value.

•mha_real_t MHASignal::min (const mha_wave_t &s)

Find minimal value.

•mha_real_t MHASignal::sumsqr_channel (const mha_wave_t &s, unsigned int chan-

nel)

Calculate sum of squared values in one channel.

•mha_real_t MHASignal::sumsqr_frame (const mha_wave_t &s, unsigned int frame)

Calculate sum over all channels of squared values.

• void MHASignal::limit (mha_wave_t &s, const mha_real_t &min, const mha_real_←-

t&max)

Limit the singal in the waveform buffer to the range [min, max].

•mha_complex_t &set (mha_complex_t &self, mha_real_t real, mha_real_t imag=0)

Assign real and imaginary parts to a mha_complex_t (p. 123) variable.

•mha_complex_t mha_complex (mha_real_t real, mha_real_t imag=0)

Create a new mha_complex_t (p. 123) with speciﬁed real and imaginary parts.

•mha_complex_t &set (mha_complex_t &self, const std::complex<mha_real_t >

&stdcomplex)

Assign a mha_complex_t (p. 123) variable from a std::complex.

• std::complex<mha_real_t >stdcomplex (const mha_complex_t &self)

Create a std::complex from mha_complex_t (p. 123).

•mha_complex_t &expi (mha_complex_t &self, mha_real_t angle)

replaces the value of the given mha_complex_t (p. 123) with exp(i∗b).

• double angle (const mha_complex_t &self)

Computes the angle of a complex number in the complex plane.

•mha_complex_t &operator+= (mha_complex_t &self, const mha_complex_t &other)

Addition of two complex numbers, overwriting the ﬁrst.

5.8 mha_signal.hh File Reference 303