A generative-discriminative hybrid approach to multi-channel noise reduction for robust automatic speech recognition.

Hendrik Meutzner, Shoko Araki, Masakiyo Fujimoto, Tomohiro Nakatani

Browse the full ICASSP paper archive.