Title :
Speech enhancement through nonlinear adaptive source separation methods
Author :
Doukas, Nikos ; Stathaki, Tania ; Naylor, Patrick
Author_Institution :
Dept. of Electr. Eng., Imperial Coll. of Sci., Technol. & Med., London, UK
Abstract :
A new method that exploits the ideas of independent source separation in the context of speech enhancement in single sensor signals, is developed and tested in various situations. The channel distortions of the two sensor case are artificially reproduced by suitable linear and nonlinear filters. Separation is implemented via a Lagrange neural network. Results on speech signals are shown
Keywords :
adaptive signal processing; filtering theory; neural nets; nonlinear filters; speech enhancement; telecommunication channels; Lagrange neural network; channel distortions; independent source separation; linear filters; nonlinear adaptive source separation methods; nonlinear filters; sensor signals; speech enhancement; Adaptive signal processing; Artificial neural networks; Educational institutions; Lagrangian functions; Neural networks; Nonlinear distortion; Nonlinear filters; Source separation; Speech enhancement; Testing;
Conference_Titel :
Statistical Signal and Array Processing, 1996. Proceedings., 8th IEEE Signal Processing Workshop on (Cat. No.96TB10004
Conference_Location :
Corfu
Print_ISBN :
0-8186-7576-4
DOI :
10.1109/SSAP.1996.534871
