MFE-Former: Disentangling Emotion-Identity Dynamics via Self-Supervised Learning for Enhancing Speech-Driven Depression Detection

• Published in: IEEE journal of biomedical and health informatics Vol. PP; pp. 1 - 12
• Main Authors: Wang, Hao, Ye, Jiayu, Yu, Yanhong, Lu, Lin, Yuan, Lin, Wang, Qingxiang
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.08.2025
• Subjects: Acoustics; Autoencoders; Bioinformatics; Contrastive learning; Data mining; Depression; emotion word reading experiment; Ethics; Feature extraction; Interference; self supervised learning; Sensitivity; Speech based depression detection

Acoustic features are crucial behavioral indicators for depression detection. However, prior speech-based depression detection methods often overlook the variability of emotional patterns across samples, leading to interference from speaker identity and hindering the effective extraction of emotiona...