Accéder directement au contenu Accéder directement à la navigation
Communication dans un congrès

Speaker detection in the wild: Lessons learned from JSALT 2019

Paola García 1 Jesus Villalba 2 Hervé Bredin 2 Jun Du 3 Diego Castan 4 Alejandrina Cristia 5 Latane Bullock 6 Ling Guo 7 Koji Okabe 7 Phani Sankar Nidadavolu 8 Saurabh Kataria 9 Sizhu Chen 10 Léo Galmant 2 Marvin Lavechin 11, 5 Lei Sun 3 Marie-Philippe Gill 12 Bar Ben-Yair 8 Sajjad Abdoli 12 Xin Wang 13 Wassim Bouaziz 11, 5 Hadrien Titeux 11, 5 Emmanuel Dupoux 5, 11 Kong Aik Lee 14 Najim Dehak 8
1 CRIIA - Centre de Recherches Ibériques et Ibéro-Américaines
Études romanes
2 LIMSI - Laboratoire d'Informatique pour la Mécanique et les Sciences de l'Ingénieur
3 USTC - University of Science and Technology of China [Hefei]
4 SRI - SRI International [Menlo Park]
5 LSCP - Laboratoire de sciences cognitives et psycholinguistique
6 Rice University [Houston]
7 NEC Corporation
8 JHU - Johns Hopkins University
9 IIT Kanpur - Indian Institute of Technology Kanpur
10 UC San Diego - University of California [San Diego]
11 CoML - Apprentissage machine et développement cognitif
LSCP - Laboratoire de sciences cognitives et psycholinguistique, Inria de Paris
12 UQAM - Université du Québec à Montréal = University of Québec in Montréal
13 CDC - Centers for Disease Control and Prevention [Atlanta]
14 Institute for Infocomm Research - I²R [Singapore]
Abstract : This paper presents the problems and solutions addressed at the JSALT workshop when using a single microphone for speaker detection in adverse scenarios. The main focus was to tackle a wide range of conditions that go from meetings to wild speech. We describe the research threads we explored and a set of modules that was successful for these scenarios. The ultimate goal was to explore speaker detection; but our first finding was that an effective diarization improves detection, and not having a diarization stage impoverishes the performance. All the different configurations of our research agree on this fact and follow a main backbone that includes diarization as a previous stage. With this backbone, we analyzed the following problems: voice activity detection, how to deal with noisy signals, domain mismatch, how to improve the clustering; and the overall impact of previous stages in the final speaker detection. In this paper, we show partial results for speaker diarizarion to have a better understanding of the problem and we present the final results for speaker detection.
Keywords : voice ac- tivity detection Voice activity detection Speaker diarization Speaker detection Index Terms-speaker detection Speech enhancement Resegmentation
Type de document :
Communication dans un congrès
Liste complète des métadonnées

Littérature citée [29 références]  Voir  Masquer  Télécharger
https://hal.archives-ouvertes.fr/hal-02417632
Contributeur : Emmanuel Dupoux <>
Soumis le : vendredi 20 décembre 2019 - 15:24:19
Dernière modification le : mardi 2 mars 2021 - 10:25:03
Archivage à long terme le : : samedi 21 mars 2020 - 19:57:49

Fichier

Vignette du document
1912.00938.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-02417632, version 1
  • ARXIV : 1912.00938

Collections

UNIV-PARIS-LUMIERES | CNRS | INRIA | EHESS | SORBONNE-UNIVERSITE | LIMSI | PSL | SU-TI | LISN | UNIV-PARIS-NANTERRE

Citation

Paola García, Jesus Villalba, Hervé Bredin, Jun Du, Diego Castan, et al.. Speaker detection in the wild: Lessons learned from JSALT 2019. Odyssey 2020 The Speaker and Language Recognition Workshop, Nov 2020, Tokyo, Japan. ⟨hal-02417632⟩

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

447

Téléchargements de fichiers

172