@inproceedings{le-lan-etal-2016-autoapprentissage,
title = "Autoapprentissage pour le regroupement en locuteurs : premi{\`e}res investigations (First investigations on self trained speaker diarization )",
author = {Le Lan, Ga{\"e}l and
Meignier, Sylvain and
Charlet, Delphine and
Larcher, Anthony},
editor = "Danlos, Laurence and
Hamon, Thierry",
booktitle = "Actes de la conf{\'e}rence conjointe JEP-TALN-RECITAL 2016. volume 1 : JEP",
month = "7",
year = "2016",
address = "Paris, France",
publisher = "AFCP - ATALA",
url = "https://aclanthology.org/2016.jeptalnrecital-jep.10",
pages = "82--90",
abstract = "This paper investigates self trained cross-show speaker diarization applied to collections of French TV archives, based on an \textit{i-vector/PLDA} framework. The parameters used for i-vectors extraction and PLDA scoring are trained in a unsupervised way, using the data of the collection itself. Performances are compared, using combinations of target data and external data for training. The experimental results on two distinct target corpora show that using data from the corpora themselves to perform unsupervised iterative training and domain adaptation of PLDA parameters can improve an existing system, trained on external annotated data. Such results indicate that performing speaker indexation on small collections of unlabeled audio archives should only rely on the availability of a sufficient external corpus, which can be specifically adapted to every target collection. We show that a minimum collection size is required to exclude the use of such an external bootstrap.",
language = "French",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="le-lan-etal-2016-autoapprentissage">
<titleInfo>
<title>Autoapprentissage pour le regroupement en locuteurs : premières investigations (First investigations on self trained speaker diarization )</title>
</titleInfo>
<name type="personal">
<namePart type="given">Gaël</namePart>
<namePart type="family">Le Lan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sylvain</namePart>
<namePart type="family">Meignier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Delphine</namePart>
<namePart type="family">Charlet</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anthony</namePart>
<namePart type="family">Larcher</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2016-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<language>
<languageTerm type="text">French</languageTerm>
<languageTerm type="code" authority="iso639-2b">fre</languageTerm>
</language>
<relatedItem type="host">
<titleInfo>
<title>Actes de la conférence conjointe JEP-TALN-RECITAL 2016. volume 1 : JEP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Laurence</namePart>
<namePart type="family">Danlos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Thierry</namePart>
<namePart type="family">Hamon</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>AFCP - ATALA</publisher>
<place>
<placeTerm type="text">Paris, France</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper investigates self trained cross-show speaker diarization applied to collections of French TV archives, based on an i-vector/PLDA framework. The parameters used for i-vectors extraction and PLDA scoring are trained in a unsupervised way, using the data of the collection itself. Performances are compared, using combinations of target data and external data for training. The experimental results on two distinct target corpora show that using data from the corpora themselves to perform unsupervised iterative training and domain adaptation of PLDA parameters can improve an existing system, trained on external annotated data. Such results indicate that performing speaker indexation on small collections of unlabeled audio archives should only rely on the availability of a sufficient external corpus, which can be specifically adapted to every target collection. We show that a minimum collection size is required to exclude the use of such an external bootstrap.</abstract>
<identifier type="citekey">le-lan-etal-2016-autoapprentissage</identifier>
<location>
<url>https://aclanthology.org/2016.jeptalnrecital-jep.10</url>
</location>
<part>
<date>2016-7</date>
<extent unit="page">
<start>82</start>
<end>90</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Autoapprentissage pour le regroupement en locuteurs : premières investigations (First investigations on self trained speaker diarization )
%A Le Lan, Gaël
%A Meignier, Sylvain
%A Charlet, Delphine
%A Larcher, Anthony
%Y Danlos, Laurence
%Y Hamon, Thierry
%S Actes de la conférence conjointe JEP-TALN-RECITAL 2016. volume 1 : JEP
%D 2016
%8 July
%I AFCP - ATALA
%C Paris, France
%G French
%F le-lan-etal-2016-autoapprentissage
%X This paper investigates self trained cross-show speaker diarization applied to collections of French TV archives, based on an i-vector/PLDA framework. The parameters used for i-vectors extraction and PLDA scoring are trained in a unsupervised way, using the data of the collection itself. Performances are compared, using combinations of target data and external data for training. The experimental results on two distinct target corpora show that using data from the corpora themselves to perform unsupervised iterative training and domain adaptation of PLDA parameters can improve an existing system, trained on external annotated data. Such results indicate that performing speaker indexation on small collections of unlabeled audio archives should only rely on the availability of a sufficient external corpus, which can be specifically adapted to every target collection. We show that a minimum collection size is required to exclude the use of such an external bootstrap.
%U https://aclanthology.org/2016.jeptalnrecital-jep.10
%P 82-90
Markdown (Informal)
[Autoapprentissage pour le regroupement en locuteurs : premières investigations (First investigations on self trained speaker diarization )](https://aclanthology.org/2016.jeptalnrecital-jep.10) (Le Lan et al., JEP/TALN/RECITAL 2016)
ACL