DMCA
Discriminative models for speech recognition (1997)
| Venue: | In Information Theory and Applications Workshop |
| Citations: | 22 - 8 self |
BibTeX
@INPROCEEDINGS{Gales97discriminativemodels,
author = {M. J. F. Gales},
title = {Discriminative models for speech recognition},
booktitle = {In Information Theory and Applications Workshop},
year = {1997},
pages = {2--3}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract — The vast majority of automatic speech recognition systems use Hidden Markov Models (HMMs) as the underlying acoustic model. Initially these models were trained based on the maximum likelihood criterion. Significant performance gains have been obtained by using discriminative training criteria, such as maximum mutual information and minimum phone error. However, the underlying acoustic model is still generative, with the associated constraints on the state and transition probability distributions, and classification is based on Bayes ’ decision rule. Recently, there has been interest in examining discriminative, or direct, models for speech recognition. This paper briefly reviews the forms of discriminative models that have been investigated. These include maximum entropy Markov models, hidden conditional random fields and conditional augmented models. The relationships between the various models and issues with applying them to large vocabulary continuous speech recognition will be discussed. I.
Keyphrases
speech recognition discriminative model automatic speech recognition system associated constraint vast majority hidden conditional random field discriminative training criterion maximum entropy markov model hidden markov model acoustic model large vocabulary continuous speech recognition minimum phone error significant performance gain various model maximum mutual information maximum likelihood criterion paper briefly bayes decision rule underlying acoustic model conditional augmented model transition probability distribution
