Gender Representation in French Broadcast Corpora and Its Impact on ASR Performance

TL;DR

This study examines gender imbalance in French broadcast corpora and its negative effect on ASR performance, highlighting under-representation of women and proposing insights into data sufficiency for bias mitigation.

Contribution

It provides a detailed analysis of gender bias in broadcast data and its impact on ASR, introducing the concept of speaker role to refine bias assessment.

Findings

Women are under-represented in broadcast data.

Gender imbalance decreases ASR performance for women.

Sufficient data can mitigate gender bias effects.

Abstract

This paper analyzes the gender representation in four major corpora of French broadcast. These corpora being widely used within the speech processing community, they are a primary material for training automatic speech recognition (ASR) systems. As gender bias has been highlighted in numerous natural language processing (NLP) applications, we study the impact of the gender imbalance in TV and radio broadcast on the performance of an ASR system. This analysis shows that women are under-represented in our data in terms of speakers and speech turns. We introduce the notion of speaker role to refine our analysis and find that women are even fewer within the Anchor category corresponding to prominent speakers. The disparity of available data for both gender causes performance to decrease on women. However this global trend can be counterbalanced for speaker who are used to speak in the media when sufficient amount of data is available.