Multi-pass feature enhancement based on generative-discriminative hybrid approach for noise robust speech recognition.

Masakiyo Fujimoto, Tomohiro Nakatani

Browse the full ICASSP paper archive.