DocumentCode :
454566
Title :
Discriminant Initialization for Factor Analyzed HMM Training
Author :
Lefevre, Fabrice ; Gauvain, Jean-Luc
Author_Institution :
Spoken Language Process. Group, LIMSI-CNRS
Volume :
1
fYear :
2006
fDate :
14-19 May 2006
Abstract :
Factor analysis has been recently used to model the covariance of the feature vector in speech recognition systems. Maximum likelihood estimation of the parameters of factor analyzed HMMs (FAHMMs) is usually done via the EM algorithm, meaning that initial estimates of the model parameters is a key issue. In this paper we report on experiments showing some evidence that the use of a discriminative criterion to initialize the FAHMM maximum likelihood parameter estimation can be effective. The proposed approach relies on the estimation of a discriminant linear transformation to provide initial values for the factor loading matrices, as well as appropriate initializations for the other model parameters. Speech recognition experiments were carried out on the Wall Street Journal LVCSR task with a 65k vocabulary. Contrastive results are reported with various model sizes using discriminant and non discriminant initialization
Keywords :
hidden Markov models; matrix algebra; maximum likelihood estimation; speech recognition; discriminant initialization; discriminant linear transformation; factor analyzed HMM training; factor loading matrices; maximum likelihood estimation; maximum likelihood parameter estimation; speech recognition systems; Automatic speech recognition; Covariance matrix; Gaussian processes; Hidden Markov models; Maximum likelihood estimation; Parameter estimation; Speech recognition; State-space methods; Vectors; Vocabulary;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on
Conference_Location :
Toulouse
ISSN :
1520-6149
Print_ISBN :
1-4244-0469-X
Type :
conf
DOI :
10.1109/ICASSP.2006.1660013
Filename :
1660013
Link To Document :
بازگشت