Skip to Main content Skip to Navigation

Séparation Aveugle de Mélanges Convolutifs de Sources

Abstract : In this thesis, Blind Source Separation (BSS) of Convolutive Mixtures of Sources is addressed. For separating audio mixtures, we have developed new methods for cases without noise and with noise in propagation environment. The method for case without noise is based on joint diagonalization of spectral matrices and exploit the non stationarity of signals. We have proposed two techniques in order to solve permutation problem. The second method where an additive noise is present, is based on maximum likelihood. Simulations were done with reel data of acoustics room.
Complete list of metadata

Cited literature [110 references]  Display  Hide  Download
Contributor : Hakim Boumaraf <>
Submitted on : Friday, February 17, 2006 - 5:55:37 PM
Last modification on : Wednesday, May 19, 2021 - 3:37:47 PM
Long-term archiving on: : Saturday, April 3, 2010 - 10:28:23 PM


  • HAL Id : tel-00011643, version 1



Hakim Boumaraf. Séparation Aveugle de Mélanges Convolutifs de Sources. Traitement du signal et de l'image [eess.SP]. Université Joseph-Fourier - Grenoble I; Institut National Polytechnique de Grenoble - INPG, 2005. Français. ⟨tel-00011643⟩



Record views


Files downloads