Extracting true speaker identities from transcriptions - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2007

Extracting true speaker identities from transcriptions

Résumé

Automatic speaker diarization generally produces a generic label such a spkr1 rather than the true identity of the speaker. Recently, two approaches based on lexical rules were proposed to extract the true identity of the speaker from the transcriptions of the audio recording without any a priori acoustic information: one uses n-gram, the other one uses semantic classification trees (SCT). The latter was proposed by the authors of this paper. In this paper, the two methods are compared in experiments carried out on French broadcast news records from the ESTER 2005 evaluation campaign. Experiments are processed on manual and automatic transcriptions. On manual transcriptions, the n-gram-based approach can be more precise, but the automatic transcriptions, the SCT-based approach gives significantly the best results in terms of recall and precision.
Fichier principal
Vignette du fichier
i07_2601.pdf (260.95 Ko) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte
Loading...

Dates et versions

hal-01434096 , version 1 (03-04-2017)

Identifiants

  • HAL Id : hal-01434096 , version 1

Citer

Yannick Estève, Sylvain Meignier, Paul Deléglise, Julie Mauclair. Extracting true speaker identities from transcriptions. Interspeech 2007, 2007, Antwerp, Belgium. ⟨hal-01434096⟩
128 Consultations
130 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More