The design of a digital filter for noise reduction in an encoded speech signal

Author

Li, Yunhong ; Ho, K.C. ; Wang, Tong

Author_Institution

Dept. of Electr. & Comput. Eng., Univ. of Missouri, Columbia, MO, USA

Volume

1

fYear

2002

fDate

4-7 Aug. 2002

Abstract

Coding/decoding of noisy speech introduces correlation between speech and noise, which makes the conventional Wiener filter derived from the decoded signal inefficient for noise reduction. This paper proposes a filter design method by using the linear prediction coefficients from the encoded speech data to reduce noise in an encoded speech signal. The proposed design method reduces the correlation between speech and noise, which provides better quality on the enhanced speech signal.

Keywords

FIR filters; Wiener filters; correlation theory; digital filters; linear predictive coding; speech enhancement; TDWF; causal FIR Wiener filter; encoded speech data LP coefficients; encoded speech signal noise reduction; linear prediction coefficients; linear predictive coding; noisy speech coding; noisy speech decoding; speech enhancement techniques; speech signal digital filter; speech/noise correlation; time domain Wiener filter; Autocorrelation; Background noise; Decoding; Design methodology; Digital filters; Frequency domain analysis; Noise reduction; Speech coding; Speech enhancement; Wiener filter;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 2002. MWSCAS-2002. The 2002 45th Midwest Symposium on

Print_ISBN

0-7803-7523-8

Type

conf

DOI

10.1109/MWSCAS.2002.1187305

Filename

1187305