Indexation en locuteur : utilisation d'informations lexicales - Archive ouverte HAL Access content directly
Conference Papers Year : 2006

Indexation en locuteur : utilisation d'informations lexicales

Abstract

The automatic speaker indexing consists in splitting the signal into homogeneous segments and cluster- ing them by speakers. However the speaker segments are speci ed with anonymous labels. This paper pro- pose to identify those speakers by extracting their full names pronounced in the show. With a semantic clas- si cation tree, the full names detected in the segment transcription are associated to this segment or to one of its neighbors. Then, a merging method associates a full name to a speaker cluster instead of the anony- mous label. The experiments are carried out over French broadcast news from the ESTER 2005 evalua- tion campaign. About 70% show duration is correctly processed for evaluation corpus.
Fichier principal
Vignette du fichier
final-108.pdf (368.72 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01434240 , version 1 (22-03-2017)

Identifiers

  • HAL Id : hal-01434240 , version 1

Cite

Julie Mauclair, Sylvain Meignier, Yannick Estève. Indexation en locuteur : utilisation d'informations lexicales. Les Journées d'Étude sur la Parole (JEP) 2006, 2006, Dinard, France. pp.5. ⟨hal-01434240⟩
101 View
53 Download

Share

Gmail Facebook X LinkedIn More