Joint audio source separation and dereverberation based on multichannel factorial hidden Markov model | IEEE Conference Publication | IEEE Xplore