Skip to Main content Skip to Navigation
Master Thesis

Répartition hommes/femmes dans les systèmes d’IA : une étude pilote sur les grands corpus pour la transcription automatique de la parole

Abstract : AI systems are trained on a huge amount of data, and speech processing technologies are no exception to the rule. However corpora may be statistically imbalanced regarding genders and this can lead to discriminative algorithms. With AI becoming ever more present in our everyday life, it seems more than necessary to be aware of the impact of gender representation in training data on the system’s performances.
This masters’ thesis proposes to study gender representation in large spoken french corpora and to estimate the impact of this distribution on the performances of an automatic speech recognition system.
Document type :
Master Thesis
Complete list of metadatas

Cited literature [30 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-01835333
Contributor : Uga - Bulles <>
Submitted on : Wednesday, July 11, 2018 - 12:17:37 PM
Last modification on : Wednesday, July 15, 2020 - 1:06:02 PM

File

GARNERIN Mahault_M2_IDL.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-01835333, version 1

Citation

Mahault Garnerin. Répartition hommes/femmes dans les systèmes d’IA : une étude pilote sur les grands corpus pour la transcription automatique de la parole. Sciences de l'Homme et Société. 2018. ⟨dumas-01835333⟩

Share

Metrics

Record views

118

Files downloads

163