A Wideband Doubly-Sparse Approach for MITO Sparse Filter Estimation

Simon Arberet; Prasad Sudhakar; Rémi Gribonval

inria-00567210, version 2

A Wideband Doubly-Sparse Approach for MITO Sparse Filter Estimation

Simon Arberet () ¹, Prasad Sudhakar ()^a^, 2, Rémi Gribonval ()^a^, 2

Acoustics, Speech and Signal Processing, IEEE International Conference on (ICASSP 2011) (2011) To appear

Résumé : We propose an approach for the estimation of sparse filters from a convolutive mixture of sources, exploiting the time-domain sparsity of the mixing filters and the sparsity of the sources in the time-frequency (TF) domain. The proposed approach is based on a wideband formulation of the cross-relation (CR) in the TF domain and on a framework including two steps: (a) a clustering step, to determine the TF points where the CR is valid; (b) a filter estimation step, to recover the set of filters associated with each source. We propose for the first time a method to blindly perform the clustering step (a) and we show that the proposed approach based on the wideband CR outperforms the narrowband approach and the GCC-PHAT approach by between 5 dB and 20 dB.

a – INRIA
1 : LTS2 - EPFL
École Polytechnique Fédérale de Lausanne
2 : METISS (INRIA - IRISA)
CNRS : UMR6074 – INRIA – Institut National des Sciences Appliquées (INSA) - Rennes – Université de Rennes 1

inria-00567210, version 2
http://hal.inria.fr/inria-00567210
oai:hal.inria.fr:inria-00567210
Contributeur : Rémi Gribonval <>
Soumis le : Dimanche 20 Février 2011, 20:51:28
Dernière modification le : Vendredi 30 Septembre 2011, 11:45:26

Voir la fiche détaillée

Documents associés

PDF :

DOI : 10.1109/ICASSP.2011.5947085

Exporter

Bibtex EndNote TEI RefWorks

%F inria-00567210, version 2
%0 Conference Proceedings
%U http://hal.inria.fr/inria-00567210
%2 INFO:INFO_SD
%T A Wideband Doubly-Sparse Approach for MITO Sparse Filter Estimation
%A Arberet, Simon
%A Sudhakar, Prasad
%A Gribonval, Rémi
%+ LTS2 - EPFL , METISS - INRIA - IRISA
%X We propose an approach for the estimation of sparse filters from a convolutive mixture of sources, exploiting the time-domain sparsity of the mixing filters and the sparsity of the sources in the time-frequency (TF) domain. The proposed approach is based on a wideband formulation of the cross-relation (CR) in the TF domain and on a framework including two steps: (a) a clustering step, to determine the TF points where the CR is valid; (b) a filter estimation step, to recover the set of filters associated with each source. We propose for the first time a method to blindly perform the clustering step (a) and we show that the proposed approach based on the wideband CR outperforms the narrowband approach and the GCC-PHAT approach by between 5 dB and 20 dB.
%2 Nous proposons une approche pour l'estimation de filtres parcimonieux à partir d'un mélange convolutif de sources, en exploitant la parcimonie dans le domaine temps-fréquence (TF). L'approche proposée est fondée sur une formulation large-bande de la cross-relation (CR) dans le domaine TF, et sur un framework composé de deux étapes: (a) une étape de clustering afin de déterminer les points TF où la CR est valide; (b) une étape d'estimation des filtres, afin de retrouver l'ensemble des filtres associés à chaque source. Nous proposons pour la première fois une méthode qui estime de façon aveugle les clusters (a) et nous montrons que l'approche proposée, fondée sur la CR large-bande améliore de 5 dBs à 20 dBs l'approche à bande étroite ainsi que l'approche GCC-PHAT.
%G Anglais
%8 2011-05-22
%2 internationale
%B Acoustics, Speech and Signal Processing, IEEE International Conference on (ICASSP 2011)
%2 Prague
%2 Tchèque, République
%8 2011-05-22
%8 2011-05-27
%I IEEE
%P To appear
%R 10.1109/ICASSP.2011.5947085
%K Blind filter estimation; sparsity; convex optimisation; cross-relation; source separation
%2 18065

<?xml version="1.0" encoding="UTF-8"?>
<listBibl xmlns="http://www.tei-c.org/ns/1.0" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML">
<listBibl type="inproceedings"><head>Communications avec actes</head>
<biblStruct type="inproceedings" xml:id="inria-00567210"><analytic>
<author><persName><forename>Simon</forename><surname>Arberet</surname></persName>
<affiliation>
<orgName type="team">LTS2 - EPFL</orgName>
<country>CH</country>
<orgName type="institution">École Polytechnique Fédérale de Lausanne</orgName>
</affiliation>
</author>
<author><persName><forename>Prasad</forename><surname>Sudhakar</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA - IRISA</orgName>
<orgName type="team">METISS</orgName>
<country>FR</country>
<orgName type="institution">CNRS : UMR6074</orgName>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Institut National des Sciences Appliquées (INSA) - Rennes</orgName>
<orgName type="institution">Université de Rennes 1</orgName>
</affiliation>
</author>
<author><persName><forename>Rémi</forename><surname>Gribonval</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA - IRISA</orgName>
<orgName type="team">METISS</orgName>
<country>FR</country>
<orgName type="institution">CNRS : UMR6074</orgName>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Institut National des Sciences Appliquées (INSA) - Rennes</orgName>
<orgName type="institution">Université de Rennes 1</orgName>
</affiliation>
</author>
<title type="main" level="a">A Wideband Doubly-Sparse Approach for MITO Sparse Filter Estimation</title></analytic><monogr><meeting>Acoustics, Speech and Signal Processing, IEEE International Conference on (ICASSP 2011)</meeting><imprint><publisher>IEEE</publisher><pubPlace><settlement>Prague</settlement><country>Tchèque, République</country></pubPlace><biblScope type="publicationDate">2011-05-22</biblScope><biblScope type="pp">To appear</biblScope></imprint></monogr><idno type="doi">10.1109/ICASSP.2011.5947085</idno><idno type="HALid">http://hal.inria.fr/inria-00567210</idno><idno type="HALFile">http://hal.inria.fr/inria-00567210/PDF/2011_ICASSP_ArberetEtAl.pdf</idno></biblStruct>
</listBibl>
</listBibl>