Global discrimination for neural predictive systems based on N-best algorithm

Author

Mellouk, Abdelhamid ; Gallinari, Patrick

Author_Institution

LRI, CNRS, Orsay, France

Volume

1

fYear

1995

fDate

9-12 May 1995

Firstpage

465

Abstract

We describe a general formalism for training neural predictive systems. We then introduce discrimination at the frame level and show how it relates to maximum mutual information training. Finally, we propose an approach for performing discrimination in predictive systems at the sequence level, it makes use of N-best sequence selection. The performance for acoustic-phonetic decoding showed a 77.4% phone accuracy on the 1988 version of the TIMIT database

Keywords

acoustic signal processing; decoding; learning (artificial intelligence); neural nets; prediction theory; speech processing; speech recognition; N-best algorithm; N-best sequence selection; TIMIT database; acoustic-phonetic decoding; frame level discrimination; global discrimination; maximum mutual information training; neural predictive systems; phone accuracy; sequence level discrimination; training; Context modeling; Decoding; Dynamic programming; Iterative algorithms; Mutual information; Neural networks; Predictive models; Production systems; Signal processing; Speech processing; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1995. ICASSP-95., 1995 International Conference on

Conference_Location

Detroit, MI

ISSN

1520-6149

Print_ISBN

0-7803-2431-5

Type

conf

DOI

10.1109/ICASSP.1995.479629

Filename

479629