Joint Discriminative Front End and Back End Training for Improved Speech Recognition Accuracy

Author

Droppo, Jasha ; Acero, Alex

Author_Institution

Speech Technol. Group, Microsoft Res., Redmond, WA

Volume

1

fYear

2006

fDate

14-19 May 2006

Abstract

This paper presents a general discriminative training method for both the front end feature extractor and back end acoustic model of an automatic speech recognition system. The front end and back end parameters are jointly trained using the Rprop algorithm against a maximum mutual information (MMI) objective function. Results are presented on the Aurora 2 noisy English digit recognition task. It is shown that discriminative training of the front end or back end alone can improve accuracy, but joint training is considerably better

Keywords

feature extraction; speech recognition; Aurora 2 noisy English digit recognition task; Rprop algorithm; automatic speech recognition system; back end acoustic model; discriminative training method; feature extractor; front end feature extractor; maximum mutual information; speech recognition accuracy; Acceleration; Acoustic noise; Automatic speech recognition; Cepstral analysis; Data mining; Feature extraction; Hidden Markov models; Mutual information; Neural networks; Speech recognition;

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.2006.1660012

Filename

1660012