Suicide Risk Assessment Using Multimodal Speech Features: A Study on the SW1 Challenge Dataset

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Marie, Ambre, Maoudj, Ilias, Dardenne, Guillaume, Quellec, Gwenolé
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910968458510336
author Marie, Ambre
Maoudj, Ilias
Dardenne, Guillaume
Quellec, Gwenolé
author_facet Marie, Ambre
Maoudj, Ilias
Dardenne, Guillaume
Quellec, Gwenolé
contents The 1st SpeechWellness Challenge conveys the need for speech-based suicide risk assessment in adolescents. This study investigates a multimodal approach for this challenge, integrating automatic transcription with WhisperX, linguistic embeddings from Chinese RoBERTa, and audio embeddings from WavLM. Additionally, handcrafted acoustic features -- including MFCCs, spectral contrast, and pitch-related statistics -- were incorporated. We explored three fusion strategies: early concatenation, modality-specific processing, and weighted attention with mixup regularization. Results show that weighted attention provided the best generalization, achieving 69% accuracy on the development set, though a performance gap between development and test sets highlights generalization challenges. Our findings, strictly tied to the MINI-KID framework, emphasize the importance of refining embedding representations and fusion mechanisms to enhance classification reliability.
format Preprint
id arxiv_https___arxiv_org_abs_2505_13069
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Suicide Risk Assessment Using Multimodal Speech Features: A Study on the SW1 Challenge Dataset
Marie, Ambre
Maoudj, Ilias
Dardenne, Guillaume
Quellec, Gwenolé
Computation and Language
Machine Learning
Sound
Audio and Speech Processing
I.2.7; I.5.1
The 1st SpeechWellness Challenge conveys the need for speech-based suicide risk assessment in adolescents. This study investigates a multimodal approach for this challenge, integrating automatic transcription with WhisperX, linguistic embeddings from Chinese RoBERTa, and audio embeddings from WavLM. Additionally, handcrafted acoustic features -- including MFCCs, spectral contrast, and pitch-related statistics -- were incorporated. We explored three fusion strategies: early concatenation, modality-specific processing, and weighted attention with mixup regularization. Results show that weighted attention provided the best generalization, achieving 69% accuracy on the development set, though a performance gap between development and test sets highlights generalization challenges. Our findings, strictly tied to the MINI-KID framework, emphasize the importance of refining embedding representations and fusion mechanisms to enhance classification reliability.
title Suicide Risk Assessment Using Multimodal Speech Features: A Study on the SW1 Challenge Dataset
topic Computation and Language
Machine Learning
Sound
Audio and Speech Processing
I.2.7; I.5.1
url https://arxiv.org/abs/2505.13069