Blind separation of convolutive mixtures

Author

Principe, Jose C. ; Wu, Hsiao-Chun

Author_Institution

Comput. Neuro-Eng. Lab., Florida Univ., Gainesville, FL, USA

Volume

2

fYear

1999

fDate

36342

Firstpage

1054

Abstract

Reverberant signals recorded by multiple microphones can be described as sums of sources convolved with different parameters. Blind source separation of this unknown linear system can be transformed to a set of instantaneous mixtures for every frequency band. In each frequency band, we may use the simultaneous diagonalization algorithms to separate the sources. In addition to our previous simultaneous diagonalization to minimize the Frobenius norm, we now propose another set of efficient simultaneous diagonalization algorithms based on Hadamard´s inequality to make the source separation feasible in the frequency domain

Keywords

Hadamard matrices; acoustic convolution; computational complexity; convolution; linear systems; minimisation; reverberation; signal resolution; Frobenius norm minimization; Hadamard inequality; acoustic signals; blind source separation; convolutive mixtures; efficient simultaneous diagonalization algorithms; instantaneous mixtures; multiple microphones; reverberant signals; source separation; unknown linear system; Blind source separation; Computational complexity; Delay; Equations; Finite impulse response filter; Frequency domain analysis; Laboratories; Neural engineering; Noise reduction; Source separation;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1999. IJCNN '99. International Joint Conference on

Conference_Location

Washington, DC

ISSN

1098-7576

Print_ISBN

0-7803-5529-6

Type

conf

DOI

10.1109/IJCNN.1999.831101

Filename

831101