Towards single-channel unsupervised source separation of speech mixtures: the layered harmonics/formants separation-tracking model.

Manuel Reyes-Gomez, Nebojsa Jojic, Daniel P. W. Ellis

Browse the full Interspeech paper archive.