Saved in:
Bibliographic Details
Main Authors: Yeh, Hsiang-Chen, Sun, Luqi, Mahapatra, Aurosweta, Chandra, Shreeram Suresh, Provost, Emily Mower, Sisman, Berrak
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.14354
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914477034700800
author Yeh, Hsiang-Chen
Sun, Luqi
Mahapatra, Aurosweta
Chandra, Shreeram Suresh
Provost, Emily Mower
Sisman, Berrak
author_facet Yeh, Hsiang-Chen
Sun, Luqi
Mahapatra, Aurosweta
Chandra, Shreeram Suresh
Provost, Emily Mower
Sisman, Berrak
contents This study investigates whether speech-based depression detection models learn depression-related acoustic biomarkers or instead rely on speaker identity cues. Using the DAIC-WOZ dataset, we propose a data-splitting strategy that controls speaker overlap between training and test sets while keeping the training size constant, and evaluate three models of varying complexity. Results show that speaker overlap significantly boosts performance, whereas accuracy drops sharply on unseen speakers. Even with a Domain-Adversarial Neural Network, a substantial performance gap remains. These findings indicate that depression-related features extracted by current speech models are highly entangled with speaker identity. Conventional evaluation protocols may therefore overestimate generalization and clinical utility, highlighting the need for strictly speaker-independent evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2604_14354
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Who is Speaking or Who is Depressed? A Controlled Study of Speaker Leakage in Speech-Based Depression Detection
Yeh, Hsiang-Chen
Sun, Luqi
Mahapatra, Aurosweta
Chandra, Shreeram Suresh
Provost, Emily Mower
Sisman, Berrak
Audio and Speech Processing
This study investigates whether speech-based depression detection models learn depression-related acoustic biomarkers or instead rely on speaker identity cues. Using the DAIC-WOZ dataset, we propose a data-splitting strategy that controls speaker overlap between training and test sets while keeping the training size constant, and evaluate three models of varying complexity. Results show that speaker overlap significantly boosts performance, whereas accuracy drops sharply on unseen speakers. Even with a Domain-Adversarial Neural Network, a substantial performance gap remains. These findings indicate that depression-related features extracted by current speech models are highly entangled with speaker identity. Conventional evaluation protocols may therefore overestimate generalization and clinical utility, highlighting the need for strictly speaker-independent evaluation.
title Who is Speaking or Who is Depressed? A Controlled Study of Speaker Leakage in Speech-Based Depression Detection
topic Audio and Speech Processing
url https://arxiv.org/abs/2604.14354