openMHA

HörTech Open Master Hearing Aid (openMHA)

1. Content of the Pre-Release

The software contains the source code of the openMHA Toolbox library, of the openMHA framework and command line application, and of a selection of algorithm plugins forming a basic hearing aid processing chain featuring

• calibration
• bilateral adaptive differential microphones for noise suppression
• binaural coherence filter foo feedback reduction and dereverberation
• multi-band dynamic range compressor for hearing loss compensation

2. Source Code Overview

Together with the MHA source code, we deliver the source code for the FFTW library, version 2.1.5, below the directory external_libs. All openMHA source code lives under the directory mha: The MHA Toolbox Library can be found in the sub-directory mha/libmha. It contains the implementation of the MHA configuration language, of signal processing primitives and also many signal processing algorithms to by used in MHA plugins. The MHA command line application and its support framework can be found in the sub-director mha/frameworks. The plugins contained in this pre-release can be found in the sub-directory mha/plugins. The IO libraries, that connect the MHA e.g. to the sound card for live processing, or to sound files for offline processing, are also found here.

3. Build Instructions

The MHA source code has to be compiled before the MHA can be used. While MHA in general can be compiled for many operating systems and hardware platforms, in this pre-release we concentrate on compilation on Ubuntu 16.04 for 64-bit processors (x86_64).

Prerequisites: 64-bit version of Ubuntu 16.04 or later,

with the following software packages installed:

• g++-5
• make
• libsndfile1-dev,
• libjack-jackd2-dev
• jackd2

4. Compilation instructions:

After downloading and unpacking the prerelease package or cloning from github compile the MHA with ./configure && make

4.1 Installation instructions:

No installation routine is provided together with the source code. The binaries will be generated in the source tree. To collect the relevant binaries in one place, you may e.g. create a bin directory and copy all the binaries there with

mkdir bin; cp $(find . -name *.so) mha/frameworks/x86_64-linux-gcc-5/mha bin

Then, you should set the environment variable LD_LIBRARY_PATH to point to your bin directory, like this:

export LD_LIBRARY_PATH=$PWD/bin

You can also add the bin directory to the PATH environment variable:

export PATH=$PATH:$PWD/bin

After this, you can invoke the MHA command line. Perform a quick test with

mha ? cmd=quit

Which should print the default configuration of the MHA without any plugins loaded.

5. Usage instructions:

We provide with this release several examples of configuration files containing detailed explations. The files included are:

a) prerelease_adm.cfg Adaptive Differential Microphones for signal enhancement. b) prerelease_coherence.cfg Binaural coherence filter for acoustic feedback and noise reduction. c) prerelease_dynamiccompression.cfg Multichannel dynamic range compressor for hearing loss compensation. d) prerelease_allplugins.cfg Implementation of the full hearing aid processing chain combining all plugins mentioned above.

For example, we can start the example d) above by the following command:

mha ?read:configuration/prerelease_allplugins.cfg cmd=start cmd=quit

Together with these configuration files we provide also example audio data from a binaural four-channel hearing aid setup in an anechoic environment:

• 1speaker_diffNoise_<2ch,4ch,binaural>.wav A female speaker positioned 20 degree to the right of the front in a spatially diffuse noise field.
• 2speaker_<2ch,4ch,binaural>.wav A female speaker positioned as above and a male speaker left behind the head, 135 degree to the left.
• 2speaker_diffNoise_<2ch,4ch,binaural>.wav Both speakers in a spatially diffuse noise.

The '2ch' versions are suitable input for the 'dynamiccompression' and the 'coherence' configurations, whereas the '4ch' versions can be used for the 'adm' and the 'allplugins' configurations. (The four-channel input is reduced to two channels by the adm processing.)

You can find the setting for the input audio files at the end in each configuration file - feel free to use your own audio examples. Note that for the adm a correct setting of the microphone distances is required for a successful application of the algorithm.