Linguistic influences on bottom-up and top-down clustering for speaker diarization - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2011

Linguistic influences on bottom-up and top-down clustering for speaker diarization

Simon Bozonnet
  • Fonction : Auteur
  • PersonId : 903988
Nicholas Evans
  • Fonction : Auteur
  • PersonId : 938450
Raphaël Troncy

Résumé

While bottom-up approaches have emerged as the standard, default approach to clustering for speaker diarization we have always found the top-down approach gives equivalent or superior performance. Our recent work shows that significant gains in performance can be obtained when cluster purification is applied to the output of topdown systems but that it can degrade performance when applied to the output of bottom-up systems. This paper demonstrates that these observations can be accounted for by factors unrelated to the speaker and that they can impact more strongly on the performance of bottom-up clustering strategies than top-down strategies. Experimental results confirm that clusters produced through top-down clustering are better normalized against phone variation than those produced through bottom-up clustering and that this accounts for the observed inconsistencies in purification performance. The work highlights the need for marginalization strategies which should encourage convergence toward different speakers rather than toward nuisance factors such as that those related to the linguistic content.
Fichier principal
Vignette du fichier
LINGUISTIC_INFLUENCES_ON_BOTTOM-UP_AND_TOP-DOWN_1_.pdf (75.77 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00733392 , version 1 (18-09-2012)

Identifiants

  • HAL Id : hal-00733392 , version 1

Citer

Simon Bozonnet, Dong Wang, Nicholas Evans, Raphaël Troncy. Linguistic influences on bottom-up and top-down clustering for speaker diarization. ICASSP 2011, May 2011, Prague, Czech Republic. pp.1. ⟨hal-00733392⟩

Collections

EURECOM
79 Consultations
90 Téléchargements

Partager

Gmail Facebook X LinkedIn More