A method of extracting time-varying acoustic features effective for speech recognition.

Kazuyo Tanaka, Hiroaki Kojima

Browse the full ICASSP paper archive.