A frequency-weighted HMM based on minimum error classification for noisy speech recognition.

Hiroshi Matsumoto, Masanori Ono

Browse the full ICASSP paper archive.