Real world source separation by combining ICA and VD-CDWT in time-frequency domain

Author

Zhang, Zhong ; Aoki, Yasudake ; Toda, Hiroshi ; Miyake, Tetsuo ; Imamura, Takashi

Author_Institution

Dept. of Production Syst. Eng., Toyohashi Univ. of Technol., Toyohashi, Japan

fYear

2009

fDate

12-15 July 2009

Firstpage

247

Lastpage

252

Abstract

It is well known that in real world source separation, the environment noise removal must be considered with complex reverberating sound, and various noises. In this study, in order to improve the voice recognition accuracy in real world source separation, a new method that uses independent component analysis (ICA) in the time-frequency domain using the variable density complex discrete wavelet transform (VD-CDWT) and the subspace method has been proposed. Through comparison of the results according to signal noise ratio (SNR), the effectiveness of the proposed method is confirmed.

Keywords

discrete wavelet transforms; independent component analysis; signal denoising; source separation; speech processing; speech recognition; time-frequency analysis; environment noise removal; independent component analysis; reverberating sound; signal noise ratio; source separation; subspace method; time-frequency domain; variable density complex discrete wavelet transform; voice recognition; Acoustic noise; Discrete wavelet transforms; Independent component analysis; Signal to noise ratio; Source separation; Speech recognition; Time frequency analysis; Wavelet analysis; Wavelet domain; Working environment noise; Independent component analysis; Sound source; Time-frequency analysis; Wavelet transform;

fLanguage

English

Publisher

ieee

Conference_Titel

Wavelet Analysis and Pattern Recognition, 2009. ICWAPR 2009. International Conference on

Conference_Location

Baoding

Print_ISBN

978-1-4244-3728-3

Electronic_ISBN

978-1-4244-3729-0

Type

conf

DOI

10.1109/ICWAPR.2009.5207450

Filename

5207450