Saved in:
Bibliographic Details
Main Authors: Bonafos, Guillem, Rouch, Jéremy, Lego, Lény, Reby, David, Patural, Hugues, Mathevon, Nicolas, Emonet, Rémy
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.02259
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918134324133888
author Bonafos, Guillem
Rouch, Jéremy
Lego, Lény
Reby, David
Patural, Hugues
Mathevon, Nicolas
Emonet, Rémy
author_facet Bonafos, Guillem
Rouch, Jéremy
Lego, Lény
Reby, David
Patural, Hugues
Mathevon, Nicolas
Emonet, Rémy
contents Transfer learning using latent representations from pre-trained speech models achieves outstanding performance in tasks where labeled data is scarce. However, their applicability to non-speech data and the specific acoustic properties encoded in these representations remain largely unexplored. In this study, we investigate both aspects. We evaluate five pre-trained speech models on eight baby cries datasets, encompassing 115 hours of audio from 960 babies. For each dataset, we assess the latent representations of each model across all available classification tasks. Our results demonstrate that the latent representations of these models can effectively classify human baby cries and encode key information related to vocal source instability and identity of the crying baby. In addition, a comparison of the architectures and training strategies of these models offers valuable insights for the design of future models tailored to similar tasks, such as emotion detection.
format Preprint
id arxiv_https___arxiv_org_abs_2509_02259
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Speech transformer models for extracting information from baby cries
Bonafos, Guillem
Rouch, Jéremy
Lego, Lény
Reby, David
Patural, Hugues
Mathevon, Nicolas
Emonet, Rémy
Sound
Machine Learning
Applications
Transfer learning using latent representations from pre-trained speech models achieves outstanding performance in tasks where labeled data is scarce. However, their applicability to non-speech data and the specific acoustic properties encoded in these representations remain largely unexplored. In this study, we investigate both aspects. We evaluate five pre-trained speech models on eight baby cries datasets, encompassing 115 hours of audio from 960 babies. For each dataset, we assess the latent representations of each model across all available classification tasks. Our results demonstrate that the latent representations of these models can effectively classify human baby cries and encode key information related to vocal source instability and identity of the crying baby. In addition, a comparison of the architectures and training strategies of these models offers valuable insights for the design of future models tailored to similar tasks, such as emotion detection.
title Speech transformer models for extracting information from baby cries
topic Sound
Machine Learning
Applications
url https://arxiv.org/abs/2509.02259