Spatial location priors for Gaussian model based reverberant audio source separation
Identifieur interne : 004675 ( Hal/Corpus ); précédent : 004674; suivant : 004676Spatial location priors for Gaussian model based reverberant audio source separation
Auteurs : Ngoc Q. K. Duong ; Emmanuel Vincent ; Rémi GribonvalSource :
Abstract
We consider the Gaussian framework for reverberant audio source separation, where the sources are modeled in the time-frequency domain by their short-term power spectra and their spatial covariance matrices. We propose three alternative probabilistic priors over the spatial covariance matrices which are consistent with the theory of statistical room acoustics and we derive Expectation-Maximization (EM) algorithms for maximum a posteriori (MAP) estimation. We argue that these algorithms provide a statistically principled solution to the permutation problem and to the risk of overfitting resulting from conventional maximum likelihood (ML) estimation. We show experimentally that, in a semi-informed scenario where the source positions and certain room characteristics are known, the algorithms using respectively inverse-Wishart and Gaussian priors outperform their ML counterparts. This opens the way to rigorous statistical treatment of this family of models in other scenarios in the future.
Url:
Links to Exploration step
Hal:hal-00727781Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en">Spatial location priors for Gaussian model based reverberant audio source separation</title>
<author><name sortKey="Duong, Ngoc Q K" sort="Duong, Ngoc Q K" uniqKey="Duong N" first="Ngoc Q. K." last="Duong">Ngoc Q. K. Duong</name>
<affiliation><hal:affiliation type="laboratory" xml:id="struct-118587" status="VALID"><orgName>Technicolor R & I [Cesson Sévigné]</orgName>
<desc><address><addrLine>975 avenue des champs blancs CS 17616 35576 Cesson Sévigné</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.technicolor.com/en/innovation/research-innovation/rennes</ref>
</desc>
<listRelation><relation active="#struct-367121" type="direct"></relation>
</listRelation>
<tutelles><tutelle active="#struct-367121" type="direct"><org type="institution" xml:id="struct-367121" status="INCOMING"><orgName>Technicolor</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
<author><name sortKey="Vincent, Emmanuel" sort="Vincent, Emmanuel" uniqKey="Vincent E" first="Emmanuel" last="Vincent">Emmanuel Vincent</name>
<affiliation><hal:affiliation type="researchteam" xml:id="struct-2522" status="OLD"><idno type="RNSR">200118344P</idno>
<orgName>Speech and sound data modeling and processing</orgName>
<orgName type="acronym">METISS</orgName>
<date type="start">2001-11-01</date>
<date type="end">2012-12-31</date>
<desc><address><addrLine>Campus de Beaulieu 35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/equipes/metiss</ref>
</desc>
<listRelation><relation active="#struct-2494" type="direct"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-117606" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
<relation active="#struct-419153" type="direct"></relation>
</listRelation>
<tutelles><tutelle active="#struct-2494" type="direct"><org type="laboratory" xml:id="struct-2494" status="OLD"><orgName>Institut de Recherche en Informatique et Systèmes Aléatoires</orgName>
<orgName type="acronym">IRISA</orgName>
<desc><address><addrLine>Campus universitaire de Beaulieu - 35042 Rennes</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.irisa.fr/</ref>
</desc>
<listRelation><relation active="#struct-105160" type="direct"></relation>
<relation active="#struct-117606" type="direct"></relation>
<relation active="#struct-300009" type="direct"></relation>
<relation name="UMR6074" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105160" type="indirect"><org type="institution" xml:id="struct-105160" status="VALID"><orgName>Université de Rennes 1</orgName>
<orgName type="acronym">UR1</orgName>
<desc><address><addrLine>2 rue du Thabor - CS 46510 - 35065 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-rennes1.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-117606" type="indirect"><org type="institution" xml:id="struct-117606" status="VALID"><orgName>Institut National des Sciences Appliquées - Rennes</orgName>
<orgName type="acronym">INSA Rennes</orgName>
<desc><address><addrLine>20, avenue des Buttes de Coësmes - CS 70839 - 35708 Rennes cedex 7</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.insa-rennes.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle name="UMR6074" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="IdRef">02636817X</idno>
<idno type="ISNI">0000000122597504</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-419153" type="direct"><org type="laboratory" xml:id="struct-419153" status="VALID"><idno type="RNSR">198018249C</idno>
<orgName>Inria Rennes – Bretagne Atlantique </orgName>
<desc><address><addrLine>Campus de beaulieu35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/centre/rennes</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
</listRelation>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
<author><name sortKey="Gribonval, Remi" sort="Gribonval, Remi" uniqKey="Gribonval R" first="Rémi" last="Gribonval">Rémi Gribonval</name>
<affiliation><hal:affiliation type="researchteam" xml:id="struct-210613" status="VALID"><idno type="RNSR">201321091Y</idno>
<orgName>Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio</orgName>
<orgName type="acronym">PANAMA</orgName>
<desc><address><addrLine>Campus de Beaulieu 35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/equipes/panama</ref>
</desc>
<listRelation><relation active="#struct-419153" type="direct"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-419366" type="direct"></relation>
<relation active="#struct-105128" type="indirect"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-172265" type="indirect"></relation>
<relation active="#struct-247362" type="indirect"></relation>
<relation active="#struct-300812" type="indirect"></relation>
<relation name="- RENNES" active="#struct-301232" type="indirect"></relation>
<relation active="#struct-301262" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
</listRelation>
<tutelles><tutelle active="#struct-419153" type="direct"><org type="laboratory" xml:id="struct-419153" status="VALID"><idno type="RNSR">198018249C</idno>
<orgName>Inria Rennes – Bretagne Atlantique </orgName>
<desc><address><addrLine>Campus de beaulieu35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/centre/rennes</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-419366" type="direct"><org type="department" xml:id="struct-419366" status="VALID"><orgName>SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE</orgName>
<orgName type="acronym">IRISA-D5</orgName>
<desc><address><country key="FR"></country>
</address>
<ref type="url">https://www.irisa.fr/fr/departements/d5-signaux-images-numeriques-robotique</ref>
</desc>
<listRelation><relation active="#struct-105128" type="direct"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-172265" type="indirect"></relation>
<relation active="#struct-247362" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-300812" type="indirect"></relation>
<relation name="- RENNES" active="#struct-301232" type="indirect"></relation>
<relation active="#struct-301262" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105128" type="indirect"><org type="laboratory" xml:id="struct-105128" status="VALID"><idno type="IdRef">026386909</idno>
<idno type="RNSR">200012163A</idno>
<orgName>Institut de Recherche en Informatique et Systèmes Aléatoires</orgName>
<orgName type="acronym">IRISA</orgName>
<date type="start">2000</date>
<desc><address><addrLine>Avenue du général LeclercCampus de Beaulieu 35042 RENNES CEDEX</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.irisa.fr</ref>
</desc>
<listRelation><relation active="#struct-105160" type="direct"></relation>
<relation active="#struct-172265" type="direct"></relation>
<relation active="#struct-247362" type="direct"></relation>
<relation active="#struct-300009" type="direct"></relation>
<relation active="#struct-300812" type="direct"></relation>
<relation name="- RENNES" active="#struct-301232" type="direct"></relation>
<relation active="#struct-301262" type="direct"></relation>
<relation name="UMR6074" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105160" type="indirect"><org type="institution" xml:id="struct-105160" status="VALID"><orgName>Université de Rennes 1</orgName>
<orgName type="acronym">UR1</orgName>
<desc><address><addrLine>2 rue du Thabor - CS 46510 - 35065 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-rennes1.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-172265" type="indirect"><org type="institution" xml:id="struct-172265" status="VALID"><orgName>Université de Bretagne Sud</orgName>
<orgName type="acronym">UBS</orgName>
<desc><address><addrLine>BP 92116 - 56321 Lorient cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-ubs.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-247362" type="indirect"><org type="institution" xml:id="struct-247362" status="VALID"><orgName>École normale supérieure - Rennes</orgName>
<orgName type="acronym">ENS Rennes</orgName>
<desc><address><addrLine>Campus de Ker Lann - avenue Robert Schuman - 35170 Bruz</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.ens-rennes.fr</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-300812" type="indirect"><org type="institution" xml:id="struct-300812" status="VALID"><orgName>SUPELEC</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle name="- RENNES" active="#struct-301232" type="indirect"><org type="institution" xml:id="struct-301232" status="VALID"><orgName>Institut National des Sciences Appliquées</orgName>
<orgName type="acronym">INSA</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle active="#struct-301262" type="indirect"><org type="institution" xml:id="struct-301262" status="VALID"><orgName>Télécom Bretagne</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle name="UMR6074" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="IdRef">02636817X</idno>
<idno type="ISNI">0000000122597504</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">HAL</idno>
<idno type="RBID">Hal:hal-00727781</idno>
<idno type="halId">hal-00727781</idno>
<idno type="halUri">https://hal.inria.fr/hal-00727781</idno>
<idno type="url">https://hal.inria.fr/hal-00727781</idno>
<date when="2012-09-04">2012-09-04</date>
<idno type="wicri:Area/Hal/Corpus">004675</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en">Spatial location priors for Gaussian model based reverberant audio source separation</title>
<author><name sortKey="Duong, Ngoc Q K" sort="Duong, Ngoc Q K" uniqKey="Duong N" first="Ngoc Q. K." last="Duong">Ngoc Q. K. Duong</name>
<affiliation><hal:affiliation type="laboratory" xml:id="struct-118587" status="VALID"><orgName>Technicolor R & I [Cesson Sévigné]</orgName>
<desc><address><addrLine>975 avenue des champs blancs CS 17616 35576 Cesson Sévigné</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.technicolor.com/en/innovation/research-innovation/rennes</ref>
</desc>
<listRelation><relation active="#struct-367121" type="direct"></relation>
</listRelation>
<tutelles><tutelle active="#struct-367121" type="direct"><org type="institution" xml:id="struct-367121" status="INCOMING"><orgName>Technicolor</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
<author><name sortKey="Vincent, Emmanuel" sort="Vincent, Emmanuel" uniqKey="Vincent E" first="Emmanuel" last="Vincent">Emmanuel Vincent</name>
<affiliation><hal:affiliation type="researchteam" xml:id="struct-2522" status="OLD"><idno type="RNSR">200118344P</idno>
<orgName>Speech and sound data modeling and processing</orgName>
<orgName type="acronym">METISS</orgName>
<date type="start">2001-11-01</date>
<date type="end">2012-12-31</date>
<desc><address><addrLine>Campus de Beaulieu 35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/equipes/metiss</ref>
</desc>
<listRelation><relation active="#struct-2494" type="direct"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-117606" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
<relation active="#struct-419153" type="direct"></relation>
</listRelation>
<tutelles><tutelle active="#struct-2494" type="direct"><org type="laboratory" xml:id="struct-2494" status="OLD"><orgName>Institut de Recherche en Informatique et Systèmes Aléatoires</orgName>
<orgName type="acronym">IRISA</orgName>
<desc><address><addrLine>Campus universitaire de Beaulieu - 35042 Rennes</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.irisa.fr/</ref>
</desc>
<listRelation><relation active="#struct-105160" type="direct"></relation>
<relation active="#struct-117606" type="direct"></relation>
<relation active="#struct-300009" type="direct"></relation>
<relation name="UMR6074" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105160" type="indirect"><org type="institution" xml:id="struct-105160" status="VALID"><orgName>Université de Rennes 1</orgName>
<orgName type="acronym">UR1</orgName>
<desc><address><addrLine>2 rue du Thabor - CS 46510 - 35065 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-rennes1.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-117606" type="indirect"><org type="institution" xml:id="struct-117606" status="VALID"><orgName>Institut National des Sciences Appliquées - Rennes</orgName>
<orgName type="acronym">INSA Rennes</orgName>
<desc><address><addrLine>20, avenue des Buttes de Coësmes - CS 70839 - 35708 Rennes cedex 7</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.insa-rennes.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle name="UMR6074" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="IdRef">02636817X</idno>
<idno type="ISNI">0000000122597504</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-419153" type="direct"><org type="laboratory" xml:id="struct-419153" status="VALID"><idno type="RNSR">198018249C</idno>
<orgName>Inria Rennes – Bretagne Atlantique </orgName>
<desc><address><addrLine>Campus de beaulieu35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/centre/rennes</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
</listRelation>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
<author><name sortKey="Gribonval, Remi" sort="Gribonval, Remi" uniqKey="Gribonval R" first="Rémi" last="Gribonval">Rémi Gribonval</name>
<affiliation><hal:affiliation type="researchteam" xml:id="struct-210613" status="VALID"><idno type="RNSR">201321091Y</idno>
<orgName>Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio</orgName>
<orgName type="acronym">PANAMA</orgName>
<desc><address><addrLine>Campus de Beaulieu 35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/equipes/panama</ref>
</desc>
<listRelation><relation active="#struct-419153" type="direct"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-419366" type="direct"></relation>
<relation active="#struct-105128" type="indirect"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-172265" type="indirect"></relation>
<relation active="#struct-247362" type="indirect"></relation>
<relation active="#struct-300812" type="indirect"></relation>
<relation name="- RENNES" active="#struct-301232" type="indirect"></relation>
<relation active="#struct-301262" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
</listRelation>
<tutelles><tutelle active="#struct-419153" type="direct"><org type="laboratory" xml:id="struct-419153" status="VALID"><idno type="RNSR">198018249C</idno>
<orgName>Inria Rennes – Bretagne Atlantique </orgName>
<desc><address><addrLine>Campus de beaulieu35042 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/centre/rennes</ref>
</desc>
<listRelation><relation active="#struct-300009" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-300009" type="indirect"><org type="institution" xml:id="struct-300009" status="VALID"><orgName>Institut National de Recherche en Informatique et en Automatique</orgName>
<orgName type="acronym">Inria</orgName>
<desc><address><addrLine>Domaine de VoluceauRocquencourt - BP 10578153 Le Chesnay Cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.inria.fr/en/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-419366" type="direct"><org type="department" xml:id="struct-419366" status="VALID"><orgName>SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE</orgName>
<orgName type="acronym">IRISA-D5</orgName>
<desc><address><country key="FR"></country>
</address>
<ref type="url">https://www.irisa.fr/fr/departements/d5-signaux-images-numeriques-robotique</ref>
</desc>
<listRelation><relation active="#struct-105128" type="direct"></relation>
<relation active="#struct-105160" type="indirect"></relation>
<relation active="#struct-172265" type="indirect"></relation>
<relation active="#struct-247362" type="indirect"></relation>
<relation active="#struct-300009" type="indirect"></relation>
<relation active="#struct-300812" type="indirect"></relation>
<relation name="- RENNES" active="#struct-301232" type="indirect"></relation>
<relation active="#struct-301262" type="indirect"></relation>
<relation name="UMR6074" active="#struct-441569" type="indirect"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105128" type="indirect"><org type="laboratory" xml:id="struct-105128" status="VALID"><idno type="IdRef">026386909</idno>
<idno type="RNSR">200012163A</idno>
<orgName>Institut de Recherche en Informatique et Systèmes Aléatoires</orgName>
<orgName type="acronym">IRISA</orgName>
<date type="start">2000</date>
<desc><address><addrLine>Avenue du général LeclercCampus de Beaulieu 35042 RENNES CEDEX</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.irisa.fr</ref>
</desc>
<listRelation><relation active="#struct-105160" type="direct"></relation>
<relation active="#struct-172265" type="direct"></relation>
<relation active="#struct-247362" type="direct"></relation>
<relation active="#struct-300009" type="direct"></relation>
<relation active="#struct-300812" type="direct"></relation>
<relation name="- RENNES" active="#struct-301232" type="direct"></relation>
<relation active="#struct-301262" type="direct"></relation>
<relation name="UMR6074" active="#struct-441569" type="direct"></relation>
</listRelation>
</org>
</tutelle>
<tutelle active="#struct-105160" type="indirect"><org type="institution" xml:id="struct-105160" status="VALID"><orgName>Université de Rennes 1</orgName>
<orgName type="acronym">UR1</orgName>
<desc><address><addrLine>2 rue du Thabor - CS 46510 - 35065 Rennes cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-rennes1.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-172265" type="indirect"><org type="institution" xml:id="struct-172265" status="VALID"><orgName>Université de Bretagne Sud</orgName>
<orgName type="acronym">UBS</orgName>
<desc><address><addrLine>BP 92116 - 56321 Lorient cedex</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.univ-ubs.fr/</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-247362" type="indirect"><org type="institution" xml:id="struct-247362" status="VALID"><orgName>École normale supérieure - Rennes</orgName>
<orgName type="acronym">ENS Rennes</orgName>
<desc><address><addrLine>Campus de Ker Lann - avenue Robert Schuman - 35170 Bruz</addrLine>
<country key="FR"></country>
</address>
<ref type="url">http://www.ens-rennes.fr</ref>
</desc>
</org>
</tutelle>
<tutelle active="#struct-300812" type="indirect"><org type="institution" xml:id="struct-300812" status="VALID"><orgName>SUPELEC</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle name="- RENNES" active="#struct-301232" type="indirect"><org type="institution" xml:id="struct-301232" status="VALID"><orgName>Institut National des Sciences Appliquées</orgName>
<orgName type="acronym">INSA</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle active="#struct-301262" type="indirect"><org type="institution" xml:id="struct-301262" status="VALID"><orgName>Télécom Bretagne</orgName>
<desc><address><country key="FR"></country>
</address>
</desc>
</org>
</tutelle>
<tutelle name="UMR6074" active="#struct-441569" type="indirect"><org type="institution" xml:id="struct-441569" status="VALID"><idno type="IdRef">02636817X</idno>
<idno type="ISNI">0000000122597504</idno>
<orgName>Centre National de la Recherche Scientifique</orgName>
<orgName type="acronym">CNRS</orgName>
<date type="start">1939-10-19</date>
<desc><address><country key="FR"></country>
</address>
<ref type="url">http://www.cnrs.fr/</ref>
</desc>
</org>
</tutelle>
</tutelles>
</hal:affiliation>
</affiliation>
</author>
</analytic>
</biblStruct>
</sourceDesc>
</fileDesc>
<profileDesc><textClass></textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">We consider the Gaussian framework for reverberant audio source separation, where the sources are modeled in the time-frequency domain by their short-term power spectra and their spatial covariance matrices. We propose three alternative probabilistic priors over the spatial covariance matrices which are consistent with the theory of statistical room acoustics and we derive Expectation-Maximization (EM) algorithms for maximum a posteriori (MAP) estimation. We argue that these algorithms provide a statistically principled solution to the permutation problem and to the risk of overfitting resulting from conventional maximum likelihood (ML) estimation. We show experimentally that, in a semi-informed scenario where the source positions and certain room characteristics are known, the algorithms using respectively inverse-Wishart and Gaussian priors outperform their ML counterparts. This opens the way to rigorous statistical treatment of this family of models in other scenarios in the future.</div>
</front>
</TEI>
<hal api="V3"><titleStmt><title xml:lang="en">Spatial location priors for Gaussian model based reverberant audio source separation</title>
<author role="aut"><persName><forename type="first">Ngoc Q. K.</forename>
<surname>Duong</surname>
</persName>
<email></email>
<idno type="halauthor">757380</idno>
<affiliation ref="#struct-118587"></affiliation>
</author>
<author role="aut"><persName><forename type="first">Emmanuel</forename>
<surname>Vincent</surname>
</persName>
<email>emmanuel.vincent@inria.fr</email>
<idno type="idhal">emmanuelv</idno>
<idno type="halauthor">571022</idno>
<affiliation ref="#struct-2522"></affiliation>
<affiliation ref="#struct-205127"></affiliation>
</author>
<author role="aut"><persName><forename type="first">Rémi</forename>
<surname>Gribonval</surname>
</persName>
<email>Remi.Gribonval@inria.fr</email>
<ptr type="url" target="http://people.irisa.fr/Remi.Gribonval/"></ptr>
<idno type="idhal">remi-gribonval</idno>
<idno type="halauthor">888631</idno>
<orgName ref="#struct-300009"></orgName>
<affiliation ref="#struct-210613"></affiliation>
</author>
<editor role="depositor"><persName><forename>Emmanuel</forename>
<surname>Vincent</surname>
</persName>
<email>emmanuel.vincent@inria.fr</email>
</editor>
</titleStmt>
<editionStmt><edition n="v1"><date type="whenSubmitted">2012-09-04 13:42:01</date>
</edition>
<edition n="v2" type="current"><date type="whenSubmitted">2013-04-02 23:26:25</date>
<date type="whenModified">2015-09-22 01:13:17</date>
<date type="whenReleased">2013-04-03 14:16:30</date>
<date type="whenProduced">2012-09-04</date>
<date type="whenEndEmbargoed">2013-04-02</date>
<ref type="file" target="https://hal.inria.fr/hal-00727781v2/document"><date notBefore="2013-04-02"></date>
</ref>
<ref type="file" subtype="author" n="1" target="https://hal.inria.fr/hal-00727781/file/RR-8057.pdf"><date notBefore="2013-04-02"></date>
</ref>
</edition>
<respStmt><resp>contributor</resp>
<name key="150572"><persName><forename>Emmanuel</forename>
<surname>Vincent</surname>
</persName>
<email>emmanuel.vincent@inria.fr</email>
</name>
</respStmt>
</editionStmt>
<publicationStmt><distributor>CCSD</distributor>
<idno type="halId">hal-00727781</idno>
<idno type="halUri">https://hal.inria.fr/hal-00727781</idno>
<idno type="halBibtex">duong:hal-00727781</idno>
<idno type="halRefHtml">[Research Report] RR-8057, INRIA. 2012</idno>
<idno type="halRef">[Research Report] RR-8057, INRIA. 2012</idno>
</publicationStmt>
<seriesStmt><idno type="stamp" n="UNIV-RENNES1">Université de Rennes 1</idno>
<idno type="stamp" n="CNRS">CNRS - Centre national de la recherche scientifique</idno>
<idno type="stamp" n="INRIA">INRIA - Institut National de Recherche en Informatique et en Automatique</idno>
<idno type="stamp" n="IRISA">Irisa</idno>
<idno type="stamp" n="IRISA_SET">IRISA_SET</idno>
<idno type="stamp" n="INRIA-LORRAINE">INRIA Nancy - Grand Est</idno>
<idno type="stamp" n="LORIA2">Publications du LORIA</idno>
<idno type="stamp" n="INRIA-NANCY-GRAND-EST">INRIA Nancy - Grand Est</idno>
<idno type="stamp" n="LORIA-TALC" p="LORIA">Traitement automatique des langues et des connaissances</idno>
<idno type="stamp" n="LORIA">LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications</idno>
<idno type="stamp" n="INRIA-RRRT">Rapports de recherche et Technique de l'Inria</idno>
<idno type="stamp" n="IRISA-D5">SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE</idno>
<idno type="stamp" n="UNIV-LORRAINE">Université de Lorraine</idno>
<idno type="stamp" n="UNIV-UBS">Université de Bretagne Sud</idno>
<idno type="stamp" n="INSTITUT-TELECOM">Institut Télécom</idno>
<idno type="stamp" n="SUPELEC">SUPELEC</idno>
<idno type="stamp" n="INRIA_TEST">INRIA - Institut National de Recherche en Informatique et en Automatique</idno>
</seriesStmt>
<notesStmt><note type="audience" n="1">Not set</note>
<note type="report" n="6">Research Report</note>
</notesStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en">Spatial location priors for Gaussian model based reverberant audio source separation</title>
<author role="aut"><persName><forename type="first">Ngoc Q. K.</forename>
<surname>Duong</surname>
</persName>
<idno type="halAuthorId">757380</idno>
<affiliation ref="#struct-118587"></affiliation>
</author>
<author role="aut"><persName><forename type="first">Emmanuel</forename>
<surname>Vincent</surname>
</persName>
<email>emmanuel.vincent@inria.fr</email>
<idno type="idHal">emmanuelv</idno>
<idno type="halAuthorId">571022</idno>
<affiliation ref="#struct-2522"></affiliation>
<affiliation ref="#struct-205127"></affiliation>
</author>
<author role="aut"><persName><forename type="first">Rémi</forename>
<surname>Gribonval</surname>
</persName>
<email>Remi.Gribonval@inria.fr</email>
<ptr type="url" target="http://people.irisa.fr/Remi.Gribonval/"></ptr>
<idno type="idHal">remi-gribonval</idno>
<idno type="halAuthorId">888631</idno>
<orgName ref="#struct-300009"></orgName>
<affiliation ref="#struct-210613"></affiliation>
</author>
</analytic>
<monogr><idno type="reportNumber">RR-8057</idno>
<imprint><date type="datePub">2012-09-04</date>
</imprint>
<authority type="institution">INRIA</authority>
</monogr>
</biblStruct>
</sourceDesc>
<profileDesc><langUsage><language ident="en">English</language>
</langUsage>
<textClass><classCode scheme="halDomain" n="info.info-ts">Computer Science [cs]/Signal and Image Processing</classCode>
<classCode scheme="halDomain" n="spi.signal">Engineering Sciences [physics]/Signal and Image processing</classCode>
<classCode scheme="halTypology" n="REPORT">Reports</classCode>
</textClass>
<abstract xml:lang="en">We consider the Gaussian framework for reverberant audio source separation, where the sources are modeled in the time-frequency domain by their short-term power spectra and their spatial covariance matrices. We propose three alternative probabilistic priors over the spatial covariance matrices which are consistent with the theory of statistical room acoustics and we derive Expectation-Maximization (EM) algorithms for maximum a posteriori (MAP) estimation. We argue that these algorithms provide a statistically principled solution to the permutation problem and to the risk of overfitting resulting from conventional maximum likelihood (ML) estimation. We show experimentally that, in a semi-informed scenario where the source positions and certain room characteristics are known, the algorithms using respectively inverse-Wishart and Gaussian priors outperform their ML counterparts. This opens the way to rigorous statistical treatment of this family of models in other scenarios in the future.</abstract>
</profileDesc>
</hal>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Hal/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 004675 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Hal/Corpus/biblio.hfd -nk 004675 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Lorraine |area= InforLorV4 |flux= Hal |étape= Corpus |type= RBID |clé= Hal:hal-00727781 |texte= Spatial location priors for Gaussian model based reverberant audio source separation }}
![]() | This area was generated with Dilib version V0.6.33. | ![]() |