Heteroscedastic discriminant analysis and reduced rank HMMs for improved speech recognition

We present the theory for heteroscedastic discriminant analysis (HDA), a model-based generalization of linear discriminant analysis (LDA) derived in the maximum-likelihood framework to handle heteroscedastic-unequal variance-classi®er models. We show how to estimate the heteroscedastic Gaussian model parameters jointly with the dimensionality reducing transform, using the EM algorithm. In doing so, we alleviate the need for an a priori ad hoc class assignment. We apply the theoretical results to the problem of speech recognition and observe word-error reduction in systems that employed both diagonal and full covariance heteroscedastic Gaussian models tested on the TI-DIGITS database. Ó 1998 Elsevier Science B.V. All rights reserved.

• Type: Journal
• Year: 1998
• Where: SPEECH
• Authors: Nagendra Kumar, Andreas G. Andreou
• Keywords: Discriminant Analysis, Heteroscedastic Discriminant Analysis, Heteroscedastic Gaussian Models, SPEECH 1998