Ce Peng defends his PhD thesis

Ce Peng successfully defends his PhD thesis

Monaural Music Separation via Supervised Non-negative Matrix Factorization with Side-information

In this thesis, we develop a single-channel sound separation technique that can isolate different musical instruments in a recording. The technique relies on machine learning with non-negative matrix factorization (NMF) of tensors where we also use a small set of exemplar instrument sounds as side information. We had great support from Anystone Technologies, Inc., an NSERC Discovery grant, and NSERC Engage grants during this research.

Congratulations Ce Peng!