openMHA

HörTech Open Master Hearing Aid (openMHA)

1. Content of the openMHA release 4.5.2 (2017-12-06)

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 [1]
• binaural coherence filter for feedback reduction and dereverberation [2]
• multi-band dynamic range compressor for hearing loss compensation [3]
• spatial filtering algorithms:
• a delay-and-sum beamformer
• a MVDR beamformer [4]
• single-channel noise reduction [5]
• simple upsampling and downsampling plugins
• STFT cyclic aliasing prevention
• adaptive feedback cancellation [6]

2. Citation in publications

In publications using openMHA, please cite

Herzke, T., Kayser, H., Loshaj, F., Grimm, G., Hohmann, V., Open signal processing software platform for hearing aid research (openMHA). Proceedings of the Linux Audio Conference. Université Jean Monnet, Saint-Étienne, pp. 35-42, 2017.

As we are working on an updated paper, please check back this section of the README for updates.

For individual algorithms, please also refer to the list of publications at the end of this README.

3. Installation from binary packages on Ubuntu.

First, add the package source with the openMHA Debian packages to your system:

under Ubuntu 16.04

sudo apt-add-repository 'deb http://mha.hoertech.de/hoertech/xenial /'

under Ubuntu 14.04

sudo apt-add-repository 'deb http://mha.hoertech.de/hoertech/trusty /'

Update the list of available packages:

sudo apt-get update

This will give you a warning:

W: The repository 'http://mha.hoertech.de/hoertech/xenial Release' does not have a Release file.
N: Data from such a repository can't be authenticated and is therefore potentially dangerous to use.
N: See apt-secure(8) manpage for repository creation and user configuration details.

Install openMHA:

sudo apt-get install openmha

This will give you again an authentication warning: WARNING: The following packages cannot be authenticated! openmha libopenmha Install these packages without verification? [y/N]

To install openMHA you have to type "y".

The authentification issue will be resolved in the future.

After installation, openMHA dcoumentation is found in /usr/share/doc/openmha and tools for GNU Octave/Matlab here: /usr/lib/openmha/mfiles

To update openMHA when a new release is available run

sudo apt-get install --only-upgrade openmha

4. 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-directory mha/frameworks. The plugins contained in this 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.

5. 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 release we concentrate on compilation on Ubuntu 16.04 for 64-bit PC processors (x86_64) and on Debian 8 (jessie) for the Beaglebone Black single-board ARM computer.

Prerequisites: 64-bit version of Ubuntu 16.04 or later, or a Beaglebone Black with Debian jessie installed.

with the following software packages installed:

• g++-5 for Ubuntu, g++-4.9 for Debian
• make
• libsndfile1-dev,
• libjack-jackd2-dev
• jackd2
• portaudio19-dev
• optional:
  • octave with the signal package and default-jre (e.g. openjdk-8-jre and openjdk-8-jdk for Debian 9)

octave and default-jre are not essential for building or running the openMHA. The build process uses octave + java to run some tests after building openMHA. If octave is not available, this test will fail. But the produced openMHA will still work.

The optional GUI (cf. openMHA_gui_manual.pdf) requires java-enabled octave in version >= 4.2.1.

6. Compilation instructions:

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

6.1 Installation instructions:

A very simple installation routine is provided together with the source code. To collect the relevant binaries in a subdirectory "bin", execute

make install

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.

7. Usage instructions:

We provide with this release several examples of configuration files and sound examples. These are contained in the directory ./mha/configurations

For example, we can start an example featuring multiple algorithms together with the following command:

mha ?read:mha/configurations/prerelease_combination.cfg cmd=start cmd=quit

8. Documentation:

User manuals are provided in PDF format. Recreating them from source is normally not necessary.

User manuals for different levels of usability in PDF format are provided with this release. These files can also be generated by typing 'make doc' in the terminal. The new manuals will be created in the ./mha/doc/ directory. In addition, html documentation is generated in ./mha/doc/mhadoc/html/ and ./mha/doc/mhaplugins/html/

Prerequisites for recreating the documents (Optional!): Extra packages are needed for generating documentation:

• doxygen
• xfig
• graphviz
• texlive
• texlive-latex-extra

9. References for individual algorithms.

[1] Elko GW, Pong ATN. A Simple Adaptive First-order Differential Microphone. In: Proceedings of 1995 Workshop on Applications of Signal Processing to Audio and Accoustics; 1995. p. 169–172.

[2] Grimm G, Hohmann V, Kollmeier B. Increase and Subjective Evaluation of Feedback Stability in Hearing Aids by a Binaural Coherence-based Noise Reduction Scheme. IEEE Transactions on Audio, Speech, and Language Processing. 2009;17(7):1408–1419.

[3] Grimm G, Herzke T, Ewert S, Hohmann V. Implementation and Evaluation of an Experimental Hearing Aid Dynamic Range Compressor Gain Prescription. In: DAGA 2015; 2015. p. 996–999.

[4] Adiloğlu K, Kayser H, Baumgärtel RM, Rennebeck S, Dietz M, Hohmann V. A Binaural Steering Beamformer System for Enhancing a Moving Speech Source. Trends in Hearing. 2015;19:2331216515618903

[5] Gerkmann T, Hendriks RC. Unbiased MMSE-Based Noise Power Estimation With Low Complexity and Low Tracking Delay. IEEE Transactions on Audio, Speech, and Language Processing. 2012;20(4):1383–1393.

[6] Schepker H, Doclo S, A semidefinite programming approach to min-max estimation of the common part of acoustic feedback paths in hearing aids. IEEE Transactions on Audio, Speech, and Language Processing. 2016;24(2):366-377.