Two-channel noise reduction with pitch-adaptive post-processing

Author

Cheikh-Rouhou, Fahmi ; Tilp, Jan

Author_Institution

Dept. of Signal Theory, Darmstadt Univ. of Technol., Germany

Volume

2

fYear

2000

fDate

Oct. 29 2000-Nov. 1 2000

Firstpage

1740

Abstract

For many applications such as hands-free telephony or voice-controlled operation of car equipment, the acoustic noise accompanying speech has to be attenuated in order to improve the system performance. This can be achieved by means of a noise reduction unit. A new method for a dual-channel enhancement of disturbed speech signals, based on spectral subtraction in combination with pitch-adaptive post-processing, is presented. Reliable speech and noise parameter estimates are obtained leading to a significant noise reduction practically without affecting the quality of speech.

Keywords

acoustic noise; adaptive filters; adaptive signal processing; automobiles; filtering theory; noise abatement; parameter estimation; spectral analysis; speech enhancement; speech intelligibility; telecommunication channels; telephony; voice equipment; acoustic noise attenuation; car equipment; correlation filter; disturbed speech signals; dual-channel speech enhancement; hands-free telephony; informal listening tests; noise reduction; noise reduction unit; pitch-adaptive post-filtering; pitch-adaptive post-processing; reliable noise parameter estimates; reliable speech parameter estimates; spectral subtraction; speech quality; system performance; two-channel noise reduction; voice-controlled operation; Filters; Geometry; Integrated circuit noise; Microphones; Noise reduction; Signal processing algorithms; Signal to noise ratio; Speech enhancement; Telephony; Working environment noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Signals, Systems and Computers, 2000. Conference Record of the Thirty-Fourth Asilomar Conference on

Conference_Location

Pacific Grove, CA, USA

ISSN

1058-6393

Print_ISBN

0-7803-6514-3

Type

conf

DOI

10.1109/ACSSC.2000.911286

Filename

911286