Approximating Negentropy of Time-Frequency Series of Speech for Fixed Point ICA by Negentropy Maximization


NAIST \| Library	Advanced Search	Japanese \| English

Please use this identifier to cite or link to this item: http://hdl.handle.net/10061/8205

Title:	Approximating Negentropy of Time-Frequency Series of Speech for Fixed Point ICA by Negentropy Maximization
Authors:	Rajkishore Prasad Hiroshi Saruwatari Kiyohiro Shikano
Issue Date:	Sep-2005
Start page:	225
End page:	228
Abstract:	This paper presents study on the approximation of negentropy of Time-Frequency Series of Speech (TFSS) using generalized Higher Order Statistics (HOS) of the non-quadratic non-linear functions in light of their efficient usability in the frequency domain blind signal separation algorithms for the separation of convolutive mixture of speech. We also propose a new nonlinear function based on the statistical modeling of TFSS by exponential power functions. The results of standard error and bias, estimated using sequential delete-one Jackknifing method, in the approximation of negentropy of TFSS by different nonlinear functions along with their signal separation performances show the superlative power of the exponential power based nonlinear function.
Description:	IWAENC2005: International Workshop on Acoustic Echo and Noise Control , September 12-15, 2005, Eindhoven, The Netherlands.
URI:	http://hdl.handle.net/10061/8205
Text Version:	Publisher
Appears in Collections:	情報科学研究科 / Graduate School of Information Science

Files in This Item:

File	Size	Format
IWAENC_2005_225.pdf	341.18 kB	Adobe PDF	View/Open