GMM-Based a Priori SNR Estimation in Speech Enhancement

Author

Lei, Jianjun ; Wang, Jian ; Liu, Gang ; Guo, Jun

Author_Institution

Sch. of Inf. Eng., Beijing Univ. of Posts & Telecommun.

Volume

1

fYear

0

fDate

0-0 0

Firstpage

4293

Lastpage

4296

Abstract

A new approach based on Gaussian mixture model (GMM) is presented to estimating the a priori SNR in speech enhancement. In the proposed method, a GMM for clean speech spectra is trained beforehand. When enhancement, the speech spectra and the a priori SNR are estimated based on the GMM, and then the a priori SNR is used to speech enhancement system. We have evaluated performance under noisy environments using NOISEX-92 database and recorded speech signals in speech enhancement task. Compared with commonly recursive averaging method, the GMM-based method improves the enhancement performance at various SNRs

Keywords

Gaussian processes; performance evaluation; recursive estimation; speech enhancement; Gaussian mixture model; NOISEX-92 database; SNR estimation; performance evaluation; recorded speech signals; speech enhancement; speech spectra; Automation; Databases; Electronic mail; Intelligent control; Speech analysis; Speech enhancement; Working environment noise; Gaussian mixture model; a priori SNR; speech enhancement;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Control and Automation, 2006. WCICA 2006. The Sixth World Congress on

Conference_Location

Dalian

Print_ISBN

1-4244-0332-4

Type

conf

DOI

10.1109/WCICA.2006.1713185

Filename

1713185