Global optimization of a neural network-hidden Markov model hybrid

Author

Bengio, Yoshua ; de Mori, Renato ; Flammia, Giovanni ; Kompe, Ralf

Author_Institution

Sch. of Comput. Sci., McGill Univ., Montreal, Que., Canada

Volume

3

Issue

2

fYear

1992

fDate

3/1/1992 12:00:00 AM

Firstpage

252

Lastpage

259

Abstract

The integration of multilayered and recurrent artificial neural networks (ANNs) with hidden Markov models (HMMs) is addressed. ANNs are suitable for approximating functions that compute new acoustic parameters, whereas HMMs have been proven successful at modeling the temporal structure of the speech signal. In the approach described, the ANN outputs constitute the sequence of observation vectors for the HMM. An algorithm is proposed for global optimization of all the parameters. Results on speaker-independent recognition experiments using this integrated ANN-HMM system on the TIMIT continuous speech database are reported

Keywords

Markov processes; neural nets; optimisation; parameter estimation; speech recognition; TIMIT continuous speech database; acoustic parameters; global optimization; hidden Markov models; multilayered networks; neural network-hidden Markov model hybrid; observation vectors; parameter estimation; recurrent artificial neural networks; speaker-independent recognition experiments; speech recognition; temporal structure; Artificial neural networks; Automatic speech recognition; Computer science; Hidden Markov models; Intelligent robots; Multi-layer neural network; Neural networks; Parameter estimation; Speech analysis; Speech recognition;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/72.125866

Filename

125866