Audio Degradation Toolbox

The Audio Degradation Toolbox (ADT) consists of Matlab code for the controlled degradation of audio signals, and for the adaptation of ground-truth to the degraded audio. Its main purpose is to test the robustness of audio analysis methods against certain classes of degradations of the audio quality.

Degradation units include: add noise, add sound, aliasing, clipping, dynamic range compression, harmonic distortion, highpass filter, apply impulse response, mp3 compression, speedup, wow resampling. These units can be combined into more complex degradations. For more info see our ISMIR paper. For a brief intro check out our ISMIR poster or the slides of Matthias's talk at LabROSA.

Examples¶

You can find examples of degradtions, as produced by the demo script here.

Getting Started¶

Get the most current version of the toolbox from our Mercurial repository. Alternatively, download it from our Downloads page keeping in mind that this may not include recent bug fixes or extensions.

Requirements: the toolbox requires Matlab and the Matlab Signal Processing Toolbox. We've tested the ADT with Matlab R2011b on Mac OSX, Matlab R2013a on Linux.

Actually getting started: try the demo scripts in the root directory of the distribution.

License¶

The Audio Degradation Toolbox (ADT) is released under the GNU General Public License 2.0 or later. Find more information in the source files and in the file COPYING that is part of the distribution.

Cite/Credit the ADT¶

In addition to the license requirements:

1. If you use the ADT for a publication please cite the following paper: Matthias Mauch and Sebastian Ewert, The Audio Degradation Toolbox and its Application to Robustness Evaluation, Proceedings of the 14th International Society for Music Information Retrieval Conference (ISMIR 2013).
2. If your use the ADT to prepare data for commercial or other public use, please credit the ADT clearly and obviously as "Powered by the Audio Degradation Toolbox, Queen Mary, University of London".

Related Software:

• FaNT toolkit
• SNR tool
• A good overview of other related stuff from the Speech group at Berkeley
• For more flexible reverberation simulations (e.g. moving sound sources):
  • http://www.odeon.dk/ (commercial software)
  • http://ease.afmg.eu/ (commercial software)
  • http://www.otlterrain.com/ (commercial software)

External Databases with additional Noise Recordings and Impulse Responses

• High-Res Spatial Room Impulse Responses: http://www.isophonics.net/content/room-impulse-response-data-set
• Recordings of Acoustic Scenes (useful as additive noise): http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/
• Multi-channel Acoustic Scene Recordings: http://parole.loria.fr/DEMAND/
• Multi-Channel (microphone array) Room Impulse Responses: http://www.commsp.ee.ic.ac.uk/~sap/resources/mardy-multichannel-acoustic-reverberation-database-at-york-database/
• Aircraft Noises: http://noisedb.stac.aviation-civile.gouv.fr/
• Ambient Noises: http://www.ntt-at.com/product/noise-DB/
• Large Collection of Sounds (mostly unstructured; could be useful for additive noise though): https://www.freesound.org/