A time-frequency blind source separation method based on segmented coherence function

Authors:
Benoit Albouy;Yannick Deville
Affiliations:
Laboratoire d'Acoustique, de Métrologie et d'instrumentation, Université Paul Sabatier, Toulouse Cedex, FRANCE 31062;Laboratoire d'Acoustique, de Métrologie et d'instrumentation, Université Paul Sabatier, Toulouse Cedex, FRANCE 31062
Venue:
IWANN '03 Proceedings of the 7th International Work-Conference on Artificial and Natural Neural Networks: Part II: Artificial Neural Nets Problem Solving Methods
Year:
2009

Citing 1
Cited 0

Blind separation of disjoint orthogonal signals: demixing N sources from 2 mixtures

ICASSP '00 Proceedings of the Acoustics, Speech, and Signal Processing, 2000. on IEEE International Conference - Volume 05

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we introduce a new blind source separation (BSS) method for linear instantaneous mixtures, which only assumes the sources to be uncorrelated. It is based on the time-segmented frequency- dependent real coherence function of the observed signals. This parameter makes it possible to detect time-frequency zones where only one source is active. Such zones are then used to identify the required separating coefficients by means of ratios of power spectral densities of the observed signals. This BSS method yields very high performance for mixtures of speech and/or noise signals, i.e. SNR improvements range from 50 dB to more than 90 dB.