Blind Separation of Single Channel Mixed Speech Signals

DOI

10.2991/meic-14.2014.226How to use a DOI?

Keywords

blind separation; single channel; informax; SVD

Abstract

Current signal separation of single channel mixed speech principally depends on the training samples of the source, for this problem, a new method of blind separation is proposed. According to distribution of different source vary in band, we use SVD method recombine the single channel of dual-source mixed signals as double channels, in that case, an complex underdetermined problem is translated into a simple even-determined problem, and ultimately, the source signals can be separated by the informax algorithm. The simulation results show that the method do not require more prior knowledge of the source, and which have a better separation performance for the mixed source signal partial overlapped in band. The method can separate the single channel of dual-source mixed speech signals blindly. The study provides important theoretical significance and application value.

Copyright

© 2014, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Dongxu Han
AU  - Zheyuan Fu
PY  - 2014/11
DA  - 2014/11
TI  - Blind Separation of Single Channel Mixed Speech Signals
BT  - Proceedings of the 2014 International Conference on Mechatronics, Electronic, Industrial and Control Engineering
PB  - Atlantis Press
SP  - 1010
EP  - 1015
SN  - 2352-5401
UR  - https://doi.org/10.2991/meic-14.2014.226
DO  - 10.2991/meic-14.2014.226
ID  - Han2014/11
ER  -