fMPE: Discriminatively Trained Features for Speech Recognition.

Daniel Povey, Brian Kingsbury, Lidia Mangu, George Saon, Hagen Soltau, Geoffrey Zweig

Browse the full ICASSP paper archive.