Time-Domain Blind Separation of Audio Sources on the Basis of a Complete ICA Decomposition of an Observation Space

0356661 - ÚTIA 2011 RIV US eng J - Journal Article
Koldovský, Zbyněk - Tichavský, Petr
Time-Domain Blind Separation of Audio Sources on the Basis of a Complete ICA Decomposition of an Observation Space.
IEEE Transactions on Audio Speech and Language Processing. Roč. 19, č. 2 (2011), s. 406-416. ISSN 1558-7916
R&D Projects: GA MŠMT 1M0572; GA ČR GA102/09/1278
Grant - others:GAČR(CZ) GP102/07/P384
Program: GP
Institutional research plan: CEZ:AV0Z10750506
Keywords : blind source separation * audio * speech
Subject RIV: BI - Acoustics
Impact factor: 1.498, year: 2011
http://library.utia.cas.cz/separaty/2011/SI/tichavsky-0356661.pdf

A novel time-domain algorithm for blind separation of audio sources that is based on a complete unconstrained decomposition of the observation space is proposed. The observation space may be defined in a general way, which allows application of long separating filters, although its dimension is low. The decomposition is done by an appropriate independent component analysis (ICA) algorithm giving independent components that are grouped into clusters corresponding to the original sources. Components of the clusters are combined by a reconstruction procedure after estimating responses of the original sources. The authors demonstrate by experiments that the method works effectively with short data, compared to other methods.
Permanent Link: http://hdl.handle.net/11104/0195125