Some voices are too common: Building fair speech recognition systems using the Common Voice dataset - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Some voices are too common: Building fair speech recognition systems using the Common Voice dataset

Résumé

Automatic speech recognition (ASR) systems become increasingly efficient thanks to new advances in neural network training like self-supervised learning. However, they are known to be unfair toward certain groups, for instance, people speaking with an accent. In this work, we use the French Common Voice dataset to quantify the biases of a pre-trained wav2vec~2.0 model toward several demographic groups. By fine-tuning the pre-trained model on a variety of fixed-size, carefully crafted training sets, we demonstrate the importance of speaker diversity. We also run an in-depth analysis of the Common Voice corpus and identify important shortcomings that should be taken into account by users of this dataset.
Fichier principal
Vignette du fichier
Interspeech_2023_CommonVoice.pdf (310.24 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-04163615 , version 1 (17-07-2023)

Licence

Identifiants

Citer

Lucas Maison, Yannick Estève. Some voices are too common: Building fair speech recognition systems using the Common Voice dataset. Interspeech 2023, ISCA, Aug 2023, Dublin, Ireland. ⟨hal-04163615⟩

Collections

UNIV-AVIGNON LIA
16 Consultations
110 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More