A Two-Step Blind Extraction Algorithm of Underdetermined Speech Mixtures

Authors:
Ming Xiao;Fuquan Wang;Jianping Xiong
Affiliations:
Department of Electric and Information Engineering, Maoming University, Maoming, P. R. China 525000 and School of Electric and Information Engineering, South China University of Technology, Guangz ...;Department of Electric and Information Engineering, Maoming University, Maoming, P. R. China 525000;Department of Electric and Information Engineering, Maoming University, Maoming, P. R. China 525000
Venue:
ISNN '08 Proceedings of the 5th international symposium on Neural Networks: Advances in Neural Networks, Part II
Year:
2008

Citing 5
Cited 0

Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications

Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Analysis of sparse representation and blind source separation

Neural Computation
Blind Source Separation by Sparse Decomposition in a Signal Dictionary

Neural Computation
Sequential blind extraction of instantaneously mixed sources

IEEE Transactions on Signal Processing
Blind extraction of singularly mixed source signals

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

Underdetermined blind extraction problem of speech mixtures is discussed in this paper. A two-step blind extraction algorithm is proposed. It firstly estimates one basis vector of the mixing matrix using the samples in single source intervals (SSI), and then extracts or reconstructs the corresponding source. Thus, the algorithm can sequentially recovery partial sources using the SSI and decorrelative when the recoverability of the matrix is unknown. Compared with ICA algorithm, it is a non-iteration method. Several speech signals experiments demonstrate its performances.