PHONETICALLY-CONSTRAINED PLDA MODELING FOR TEXT-DEPENDENT SPEAKER VERIFICATION WITH MULTIPLE SHORT UTTERANCES - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2013

PHONETICALLY-CONSTRAINED PLDA MODELING FOR TEXT-DEPENDENT SPEAKER VERIFICATION WITH MULTIPLE SHORT UTTERANCES

Résumé

The importance of phonetic variability for short duration speaker verification is widely acknowledged. This paper assesses the performance of Probabilistic Linear Discriminant Analysis (PLDA) and i-vector normalization for a text-dependent verification task. We show that using a class definition based on both speaker and pho-netic content significantly improves the performance of a state-of-the-art system. We also compare four models for computing the verification scores using multiple enrollment utterances and show that using PLDA intrinsic scoring obtains the best performance in this context. This study suggests that such scoring regime remains to be optimized.
Fichier principal
Vignette du fichier
Icassp13_TD-IV_v0.pdf (185.18 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01927589 , version 1 (19-11-2018)

Identifiants

  • HAL Id : hal-01927589 , version 1

Citer

Anthony Larcher, Kong Aik Lee, Bin Ma, Haizhou Li. PHONETICALLY-CONSTRAINED PLDA MODELING FOR TEXT-DEPENDENT SPEAKER VERIFICATION WITH MULTIPLE SHORT UTTERANCES. IEEE International Conference on Acoustic Speech and Signal Processing, May 2013, Vancouver, Canada. ⟨hal-01927589⟩

Collections

UNIV-LEMANS LIUM
40 Consultations
216 Téléchargements

Partager

Gmail Facebook X LinkedIn More