Skip to content

An inverse-gamma source variance prior with factorized parameterization for audio source separation

Notifications You must be signed in to change notification settings

dkounadis/nmfig

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NMFIG

MATLAB code for the Variational EM algorithm for Source Separation via non-factorised full-rank PSD (NMF applies in the Bayesian prior) via Inverse-Gamma priors from the paper:

D. Kounades-Bastian, L. Girin, X. Alameda-Pineda, S. Gannot and R. Horaud, 
"An inverse-gamma source variance prior with factorized parameterization for audio source separation,"
2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Shanghai, China, pp. 136-140, doi: 10.1109/ICASSP.2016.7471652.

DEMO

# In Matlab command line
>> example

example.m generates a stereo mix of 3 sources (by loading trueSrc1.wav, ..). Then calls initNMF() to provide initialisation for the Variational EM params (via binary masking). Then applies nmfig() to separate the sources.

Separated sources will be saved as .wav files (estimatedSrc1.wav, etc.) in the directory ./results/. File nmfig.m implements the NMFIG algorithm.

PAPER

About

An inverse-gamma source variance prior with factorized parameterization for audio source separation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages