Presentation 2009-10-23
Estimation of an un-mixed matrix for blind source separation with least-squares joint diagonalization problem
Shinya SAITO, Kunio OISHI, Hajime KUBOTA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We present a new learning algorithm for solving a least-squares joint diagonalization problem at every frequency bin in convolutive frequency-domain blind source separation. Conventional algorithm requires computational burden for computing inverse matrix at every frequency bin. The proposed algorithm employs the matrix inversion lemma. It is useful to recursively estimate un-mixed matrix without being affected by initializing the algorithm. The performance of the algorithm will be evaluated in real reverberant environments.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) frequency-domain blind source separation / joint diagonalization / inversion lemma / convolutive mixture
Paper # EA2009-68
Date of Issue

Conference Information
Committee EA
Conference Date 2009/10/15(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Engineering Acoustics (EA)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Estimation of an un-mixed matrix for blind source separation with least-squares joint diagonalization problem
Sub Title (in English)
Keyword(1) frequency-domain blind source separation
Keyword(2) joint diagonalization
Keyword(3) inversion lemma
Keyword(4) convolutive mixture
1st Author's Name Shinya SAITO
1st Author's Affiliation Graduate School of Engineering, Chiba Institute of Technology()
2nd Author's Name Kunio OISHI
2nd Author's Affiliation Graduate School of Bionics, Computer and Media Sciences, Tokyo University of Technology
3rd Author's Name Hajime KUBOTA
3rd Author's Affiliation Graduate School of Engineering, Chiba Institute of Technology
Date 2009-10-23
Paper # EA2009-68
Volume (vol) vol.109
Number (no) 240
Page pp.pp.-
#Pages 6
Date of Issue