Emo-bias: A Large Scale Evaluation of Social Bias on Speech Emotion Recognition

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Lin, Yi-Cheng, Wu, Haibin, Chou, Huang-Cheng, Lee, Chi-Chun, Lee, Hung-yi
Format:	Preprint
Published:	2024
Subjects:	Audio and Speech Processing
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866913492315930624
author	Lin, Yi-Cheng Wu, Haibin Chou, Huang-Cheng Lee, Chi-Chun Lee, Hung-yi
author_facet	Lin, Yi-Cheng Wu, Haibin Chou, Huang-Cheng Lee, Chi-Chun Lee, Hung-yi
contents	The rapid growth of Speech Emotion Recognition (SER) has diverse global applications, from improving human-computer interactions to aiding mental health diagnostics. However, SER models might contain social bias toward gender, leading to unfair outcomes. This study analyzes gender bias in SER models trained with Self-Supervised Learning (SSL) at scale, exploring factors influencing it. SSL-based SER models are chosen for their cutting-edge performance. Our research pioneering research gender bias in SER from both upstream model and data perspectives. Our findings reveal that females exhibit slightly higher overall SER performance than males. Modified CPC and XLS-R, two well-known SSL models, notably exhibit significant bias. Moreover, models trained with Mandarin datasets display a pronounced bias toward valence. Lastly, we find that gender-wise emotion distribution differences in training data significantly affect gender bias, while upstream model representation has a limited impact.
format	Preprint
id	arxiv_https___arxiv_org_abs_2406_05065
institution	arXiv
publishDate	2024
record_format	arxiv
spellingShingle	Emo-bias: A Large Scale Evaluation of Social Bias on Speech Emotion Recognition Lin, Yi-Cheng Wu, Haibin Chou, Huang-Cheng Lee, Chi-Chun Lee, Hung-yi Audio and Speech Processing The rapid growth of Speech Emotion Recognition (SER) has diverse global applications, from improving human-computer interactions to aiding mental health diagnostics. However, SER models might contain social bias toward gender, leading to unfair outcomes. This study analyzes gender bias in SER models trained with Self-Supervised Learning (SSL) at scale, exploring factors influencing it. SSL-based SER models are chosen for their cutting-edge performance. Our research pioneering research gender bias in SER from both upstream model and data perspectives. Our findings reveal that females exhibit slightly higher overall SER performance than males. Modified CPC and XLS-R, two well-known SSL models, notably exhibit significant bias. Moreover, models trained with Mandarin datasets display a pronounced bias toward valence. Lastly, we find that gender-wise emotion distribution differences in training data significantly affect gender bias, while upstream model representation has a limited impact.
title	Emo-bias: A Large Scale Evaluation of Social Bias on Speech Emotion Recognition
topic	Audio and Speech Processing
url	https://arxiv.org/abs/2406.05065