Penalty function maximization for large margin HMM training.

George Saon, Daniel Povey

Browse the full Interspeech paper archive.