Here is the supplementary material from my E4810 (Digital Signal Processing) class project. The report can be found here, the slides from my presentation are here, and the source code is here.

The following table gives the figures and sound examples for several sound sources with differing amounts of reverb. The links labelled "eps" are eps files with spectrogram results, "clean" represents the original clean sounds, "reverb" are the reverberated versions of the clean sounds, "alg" is the result of the dereverberation algorithm, and "oracle" is the dereverberation result when using an ideal binary mask (see the report).

T60=0.25sec T60=0.50sec T60=1.00sec
Clapping [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]
Drums [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]
Digits(F) [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]
Digits(M) [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]
Speech(F) [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]
Speech(M) [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle] [eps][reverb][clean][alg][oracle]