Mikkel N. Schmidt

This talk was given at the Intelligent Sound Workshop in Sæby, May 2006. It presents some thoughts on speech separations of two known speakers using sparse NMF with smoothness constraints which are learned from a training set. The slides include an audio demo.