Feature Combination Using Multiple Spectral Cues for Robust Speech Recognition in Mobile Communications

Author

Addou, Djamel ; Selouani, Sid-Ahmed ; Boudraa, Malika ; Boudraa, Bachir

Author_Institution

Speech & Signal Process. Lab., USTHB Univ. of Sci. & Technol., Algiers

fYear

2009

fDate

27-29 April 2009

Firstpage

1256

Lastpage

1261

Abstract

This paper investigates a new front-end processing that aims at improving the performance of speech recognition in noisy mobile environments. This approach combines features based on conventional Mel-cepstral coefficients (MFCCs) and line spectral frequencies (LSFs) to constitute robust multivariate feature vectors. The proposed front-end constitutes an alternative to the DSRXAFE (XAFE: extended audio front-end) available in GSM mobile communications. Our results showed that for highly noisy speech, using the paradigm that combines LSF with MFCCs, leads to a significant improvement in recognition accuracy on the Aurora 2 task.

Keywords

cellular radio; cepstral analysis; feature extraction; speech recognition; DSRXAFE; GSM mobile communication; MFCC approach; Mel-frequency cepstral coeffecient; distributed speech recognition-extended audio front-end; front-end processing; line spectral frequency; multiple spectral cue; multivariate feature vector; Automatic speech recognition; Codecs; Frequency; GSM; Mobile communication; Robustness; Speech analysis; Speech recognition; Telecommunication standards; Working environment noise; Distributed Speech Recognition; GSM; Line Spectral Frequencies; Noisy mobile communications;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Technology: New Generations, 2009. ITNG '09. Sixth International Conference on

Conference_Location

Las Vegas, NV

Print_ISBN

978-1-4244-3770-2

Electronic_ISBN

978-0-7695-3596-8

Type

conf

DOI

10.1109/ITNG.2009.332

Filename

5070797