Yi Luo, Columbia University, Clustering methods for speech separation

Thursday Feb 21, 2019, 10:00 AM → 11:30 AM America/New_York

7th floor conference room (Flatiron Institute)

Clustering methods for single-channel speech separation – a deep learning perspective

Abstract:

Clustering methods for speech separation has long been a hot topic in the audio processing community. With the recent progress in deep learning, new systems that integrate neural networks together with traditional clustering methods have greatly advanced the state-of-the-art in this problem. In this talk, I will make an overview about several recent deep learning methods that help traditional clustering methods in speech separation, and introduce a more general framework for integrating iterative clustering algorithms with neural networks to achieve a more stable separation performance.

Luo_IEEE2018.pdf