Abstract

Summary In recent years, there has been a trend towards training large vocabulary continuous speech recognition (LVCSR) systems on a large amount of found data. Found data is recorded from spontaneous speech without careful control of the recording acoustic conditions, for example, conversational telephone speech. Hence, it typically has greater variability in terms of speaker and acoustic conditions than specially collected data. Thus, in addition to the desired speech variability required to discriminate between words, it also includes various non-speech variabil-ities, for example, the change of speakers or acoustic environments. The standard approach to handle this type of data is to train hidden Markov models (HMMs) on the whole data set as if all data comes from a single acoustic condition. This is referred to as multi-style training, for exam-ple speaker-independent training. Effectively, the non-speech variabilities are ignored. Though good performance has been obtained with multi-style systems, these systems account for all variabilities. Improvement may be obtained if the two types of variabilities in the found data are modelled separately. Adaptive training has been proposed for this purpose. In contrast to multi-style training, a set of transforms is used to represent the non-speech variabilities. A canonical

Citations

6231	Maximum likelihood from incomplete data via the EM algorithm - Dempster, Laird, et al. - 1977
2107	Introduction to statistical pattern recognition - Fukunaga - 1990
1484	D: Pattern Classification - Duda, Hart, et al.
1180	and Biing-Hwang Juang. Fundamentals of speech recognition - Rabiner - 1993
917	Statistical decision theory and Bayesian analysis - Berger - 1985
896	BD: Pattern Recognition and Neural Networks - Ripley - 1996
874	Error bounds for convolutional codes and an asymptotically optimum decoding algorithm - Viterbi - 1967
588	Monte Carlo statistical methods - Robert, Casella - 2004
550	Comparison of parametric representations of monosyllabic word recognition in continuously spoken sentences - Davis, Mermelstein - 1980
512	Optimal Statistical Decisions - DeGroot - 1970
440	Maximum likelihood linear regression for speaker adaptation of continuous density hidden markov models - Leggetter, Woodland - 1995
396	A model for reasoning about persistence and causation - Dean, Kanazawa - 1990
372	Maximum a posteriori estimation for multivariate gaussian mixture observations of markov chains - Gauvain, Lee - 1994
286	The population frequencies of species and the estimation of population parameters - Good - 1953
275	Maximum likelihood linear transformations for HMMbased speech recognition - Gales - 1998
195	The zero-frequency problem: estimating the probabilities of novel events in adaptive text compression - Witten, Bell - 1991
181	Variational algorithms for approximate Bayesian inference - Beal - 2003
173	An inequality with applications to statistical estimation for probabilistic functions of a Markov process and to models for ecology - Baum, Eagon - 1967
156	Some statistical issues in the comparison of speech recognition algorithms - Gillick, Cox - 1989
153	Accurate approximations for posterior moments and marginal densities - Tierney, Kadane - 1986
152	Continuous speech recognition by statistical methods - Jelinek - 1976
146	Semi-tied covariance matrices for hidden Markov models - Gales - 1999
139	Tree-based state tying for high accuracy acoustic modelling - Young, Odell, et al. - 1994
133	Effectiveness of linear prediction characteristics of the speech wave for automatic speaker identification and verification - Atal - 1974
125	Minimum Phone Error and I-smoothing for Improved Discriminative Training - Povey, Woodland
123	R.: Maximum mutual information estimation of hidden Markov model parameters for speech recognition - Bahl, Brown, et al. - 2000
106	A compact model for speaker adaptive training - Anastasakos, McDonough, et al.
102	Minimum classification error rate methods for speech recognition - Juang, Hou, et al. - 1997
80	Mean and Variance Adaptation within the mllr Framework.” Computer Speech and Language 10 - Gales, Woodland - 1996
77	The N-best algorithm: an efficient and exact procedure for finding the N most likely sentence hypotheses - Schwartz, Chow - 1990
77	Large scale discriminative training of hidden Markov models for speech recognition - Woodland, Povey - 2002
71	Discriminative Training for Large Vocabulary Speech Recognition - Povey - 2004
70	Confidence measures for large vocabulary continuous speech recognition - Wessel, Schlüter, et al. - 2001
68	Speaker normalization using efficient frequency warping procedures - Lee, Rose - 1996
66	Investigation of Silicon-Auditory Models and Generalization of Linear Discriminant Analysis for Improved Speech Recognition - Kumar - 1997
65	Fast speaker adaptation using constrained estimation of Gaussian mixtures - Digalakis, Rtischev, et al. - 1995
65	Rapid speaker adaptation in eigenvoice space - Kuhn, Junqua, et al. - 2000
63	The DRAGON system – an overview - Baker - 1975
62	Flexible speaker adaptation using maximum likelihood linear regression - LEGGETTER, WOODLAND - 1995
58	Large scale discriminative training for speech recognition - Woodland, Povey
56	The 1994 HTK large vocabulary speech recognition system - Woodland, Leggetter, et al. - 1995
54	Maximum a posteriori linear regression for hidden Markov model adaptation - Chesta, Siohan, et al. - 1999
52	Maximum likelihood estimation for multivariate mixture observations of Markov chains - Juang, Levinson, et al. - 1986
51	The Generation and Use of Regression Class Trees for mllr Adaptation - Gales - 1996
51	The Empirical Bayes approach to statistics - Robbins - 1955
46	Large vocabulary continuous speech recognition using HTK - Woodland, Odell, et al. - 1994
42	Minimum Bayes-risk automatic speech recognition - Goel, Byrne - 2003
38	RASTA-PLP speech analysis technique - Hermansky, Morgan, et al. - 1992
38	Using word probabilities as confidence measures - Wessel, Macherey, et al. - 1998
36	Cluster adaptive training of hidden Markov models - Gales - 2000