A Hybrid HMM-Based Speech Recognizer Using Kernel-Based Discriminants as Acoustic Models

Author

Andelic, E. ; Schaffoner, M. ; Katz, Marcos ; Kruger, S.E. ; Wendemuth, Andreas

Author_Institution

Cognitive Syst. Group, Otto-von-Guericke Univ., Magdeburg

Volume

2

fYear

0

fDate

0-0 0

Firstpage

1158

Lastpage

1161

Abstract

In this paper, we propose a novel order-recursive training algorithm for kernel-based discriminants which is computationally efficient. We integrate this method in a hybrid HMM-based speech recognition system by translating the outputs of the kernel-based classifier into class-conditional probabilities and using them instead of Gaussian mixtures as production probabilities of a HMM-based decoder for speech recognition. The performance of the described hybrid structure is demonstrated on the DARPA resource management (RMI) corpus

Keywords

hidden Markov models; pattern classification; probability; speech recognition; DARPA resource management; HMM-based speech recognizer; acoustic model; class-conditional probability; hidden Markov model based decoder; kernel-based classifier; kernel-based discriminants; order-recursive training algorithm; speech recognition; Automatic speech recognition; Decoding; Hidden Markov models; Kernel; Pattern recognition; Production systems; Resource management; Speech processing; Speech recognition; Stochastic processes;

fLanguage

English

Publisher

ieee

Conference_Titel

Pattern Recognition, 2006. ICPR 2006. 18th International Conference on

Conference_Location

Hong Kong

ISSN

1051-4651

Print_ISBN

0-7695-2521-0

Type

conf

DOI

10.1109/ICPR.2006.82

Filename

1699415