Audio-visual speech enhancement with AVCDCN (audio-visual codebook dependent cepstral normalization)

Author

Deligne, Sabine ; Potamianos, Gerasimos ; Neti, Chalapathy

Author_Institution

IBM T. J. Watson Res. Center, Yorktown Heights, NY, USA

fYear

2002

fDate

4-6 Aug. 2002

Firstpage

68

Lastpage

71

Abstract

We introduce a non-linear enhancement technique called audio-visual codebook dependent cepstral normalization (AVCDCN) and we consider its use with both audio-only and audio-visual speech recognition. AVCDCN is inspired from CDCN, an audio-only enhancement technique that approximates the nonlinear effect of noise on speech with a piecewise constant function. Our experiments show that the use of visual information in AVCDCN allows significant performance gains over CDCN.

Keywords

audio coding; cepstral analysis; data compression; piecewise constant techniques; speech enhancement; speech recognition; video coding; AVCDCN; CDCN; audio-only speech recognition; audio-visual codebook dependent cepstral normalization; audio-visual speech enhancement; audio-visual speech recognition; nonlinear effect; parameter estimation; piecewise constant function; video compression; visual information; Automatic speech recognition; Cepstral analysis; Degradation; Integral equations; Noise level; Noise robustness; Performance gain; Speech enhancement; Speech recognition; Working environment noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Sensor Array and Multichannel Signal Processing Workshop Proceedings, 2002

Print_ISBN

0-7803-7551-3

Type

conf

DOI

10.1109/SAM.2002.1191001

Filename

1191001