This is the MATLAB code of the speech enhancement algorithm developed during my master thesis at Universität Oldenburg in 2022.
To try it out, simply clone or download the repository, and run the thomsen2022_main.m script.
Notes:
- You must add the folder of the repository and all its subfolders to your MATLAB search path.
- You need to install the SOFA API for Matlab/Octave.
The main function of the speech enhancement algorithm is 01_base/speechEnhancement.m. Its subfunctions are stored in the folders 03_filterbank, 04_periodicityAnalysis, 05_directionOfArrivalEstimation and 06_enhancementStage.
The folder 02_testSignalGeneration contains a few test stimuli and functions used by thomsen2022_main.m for generating binaural multi-talker test signals.
The folder Hohmann2002 contains the original implementation of the gammatone filterbank that is used by the subbandDecomposition.m and subbandResynthesis.m functions. It is taken from the Auditory Modeling Toolbox.
testSuite.m and TestAlgorithm.m contain unit tests that were used during the development of the algorithm.