Auditory scene analysis via application of ICA in a time-frequency domain

被引：0

作者：

Janku, L ^{[1
]}

机构：

[1] Czech Tech Univ, Fac Elect Engn, Dept Comp Sci, CR-16635 Prague, Czech Republic

[2] Tech Univ Brno, Fac Informat Technol, Inst Comp Graph & Multimedia, Brno, Czech Republic

来源：

TEXT, SPEECH AND DIALOGUE, PROCEEDINGS | 2004年 / 3206卷

关键词：

D O I：

暂无

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper deals with the auditory scene analysis via application of ICA in a time-frequency domain. An extension of an original algorithm is presented. This extension consists in Bayesian estimation of a number of independent components via direct implementation of selected grouping principles and via analysis of a structure of the previous time-spans. While the original algorithm is not capable to process sound scenes with fluctuating number of independent sound sources, the presented extension can operate also on sound scenes with the fluctuating number of sound sources.

引用

页码：347 / 353

页数：7

共 18 条

[1]

Albert S. Bregman, 1990, AUDITORY SCENE ANAL, P411, DOI [DOI 10.1121/1.408434, DOI 10.7551/MITPRESS/1486.001.0001]

[2]

[Anonymous], SIGNAL PROCESS

[3]

[Anonymous], 2002, Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications

[4]

[Anonymous], INDEPENDENT COMPONEN

[5] SOME EXPERIMENTS ON THE RECOGNITION OF SPEECH, WITH ONE AND WITH 2 EARS [J].

CHERRY, EC .

JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA, 1953, 25 (05) :975-979

[6]

Cooke M., 1993, MODELLING AUDITORY P

[7]

Duda R. O., 1990, P 1990 IEEE AS WORKS

[8]

Duda R. O., 1973, PATTERN CLASSIFICATI

[9]

ELLIS DPW, THESIS MIT DEPT ELEC

[10]

Hyvarinen A, 1999, IEEE T NEURAL NETWOR, V10

← 1 2 →