Sound examples for

Bayesian audio source separation

C. Févotte

Blind Speech Separation, Ed. S. Makino, T.-W. Lee, H. Sawada





Stereo mixture




a) Student t prior with full block update of the sources
estimate 1
estimate 2
estimate 3

b) Student t prior with individual update of the sources
estimate 1
estimate 2
estimate 3

c) Hierarchical prior with Bernoulli prior and individual update of the sources
estimate 1
estimate 2
estimate 3

d) Hierarchical prior with horizontal Markov prior and individual update of the sources
estimate 1
estimate 2
estimate 3

Original sources
source 1
source 2
source 3


The original sources are excerpts from the song Anabelle Lee from Alex Q, published under a Creative Commons License