Improved noise robustness by corrective and rival training

Author

Meyer, Carsten ; Rose, Georg

Author_Institution

Philips Res. Lab., Aachen, Germany

Volume

1

fYear

2001

fDate

2001

Firstpage

293

Abstract

We show that discriminative training methods have the potential to improve noise robustness even for high resolution acoustic models trained on noisy data. To this end, we compare the performance of acoustic models trained on noisy data using maximum likelihood (ML), corrective (CT) and rival training (RT). Experiments are performed on a German and a Dutch continuous digit string recognition task, yielding improvements in the range of 12% to 35% relative

Keywords

AWGN; acoustic noise; maximum likelihood estimation; noise pollution; speech recognition; AWGN; Dutch; German; acoustic model training; acoustic models; additive Gaussian white noise; additive real car noise; automatic speech recognition; continuous digit string recognition; corrective training; discriminative training; environmental noise; high resolution acoustic models; maximum likelihood training; noise robustness; noisy data; rival training; Acoustic noise; Automatic speech recognition; Laboratories; Noise robustness; Singular value decomposition; Speech enhancement; Testing; Training data; Wiener filter; Working environment noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 2001. Proceedings. (ICASSP '01). 2001 IEEE International Conference on

Conference_Location

Salt Lake City, UT

ISSN

1520-6149

Print_ISBN

0-7803-7041-4

Type

conf

DOI

10.1109/ICASSP.2001.940825

Filename

940825