DocumentCode :
1901965
Title :
A new class of fenonic Markov word models for large vocabulary continuous speech recognition
Author :
Bahl, Lalit R. ; Bellegarda, Jerome R. ; deSouza, P.V. ; Gopalakrishnan, P.S. ; Nahamoo, David ; Picheny, Michael A.
Author_Institution :
IBM, Thomas Watson Res. Center, Yorktown Heights, NY, USA
fYear :
1991
fDate :
14-17 Apr 1991
Firstpage :
177
Abstract :
A technique for constructing hidden Markov models for the acoustic representation of words is described. The models, built from combinations of acoustically based subword units called fenones, are derived automatically from one or more sample utterances of words. They are more flexible than previously reported fenone-based word models and lead to an improved capability of modeling variations in pronunciation. In addition, their construction is simplified, because it can be done using the well-known forward-backward algorithm for the parameter estimation of hidden Markov models. Experimental results obtained on a 5000-word vocabulary continuous speech recognition task are presented to illustrate some of the benefits associated with the new models. Multonic baseforms resulted in a reduction of 16% in the average error rate obtained for ten speakers
Keywords :
Markov processes; acoustic signal processing; speech analysis and processing; speech recognition; HMM; acoustic representation; acoustically based subword units; average error rate; continuous speech recognition; fenones; fenonic Markov word models; forward-backward algorithm; hidden Markov models; large vocabulary; multonic baseforms; parameter estimation; pronunciation variations modeling; Artificial intelligence; Decoding; Leg; Oils; Speech recognition; TV; Tin; Vocabulary;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1991. ICASSP-91., 1991 International Conference on
Conference_Location :
Toronto, Ont.
ISSN :
1520-6149
Print_ISBN :
0-7803-0003-3
Type :
conf
DOI :
10.1109/ICASSP.1991.150306
Filename :
150306
Link To Document :
بازگشت