A new discriminative training algorithm for hidden Markov models.

Horacio Franco, Antnio Joaquim Serralheiro

Browse the full Interspeech paper archive.