Maximum conditional mutual information projection for speech recognition.

Mohamed Kamal Omar, Mark Hasegawa-Johnson

Browse the full Interspeech paper archive.