Non-uniform Speaker Disentanglement For Depression Detection From Raw Speech Signals
Authors:
Preprint
Abstract
While speech-based depression detection methods that use speaker-identity features, such as speaker embeddings, are popular, they often compromise patient privacy. To address this issue, we propose a speaker disentanglement method that utilizes a non-uniform mechanism of adversarial SID loss maximization. This is achieved by varying the adversarial weight between different layers of a model during training. We find that a greater adversarial weight for the initial layers leads to performance improvement. Our approach using the ECAPA-TDNN model achieves an F1-score of 0.7349 (a 3.7% improvement over audio-only SOTA) on the DAIC-WoZ dataset, while simultaneously reducing the speaker-identification accuracy by 50%. Our findings suggest that identifying depression through speech signals can be accomplished without placing undue reliance on a speaker's identity, paving the way for privacy-preserving approaches of depression detection.Keywords
All Related Versions
This publication has 31 references indexed in Scilit:
- Toward Knowledge-Driven Speech-Based Models of Depression: Leveraging Spectrotemporal Variations in Speech Vowels2022
- MFCC-based Recurrent Neural Network for automatic clinical depression recognition and assessment from speechBiomedical Signal Processing and Control, 2022
- Quantifying the separability of data classes in neural networksNeural Networks, 2021
- Using i-vectors from voice features to identify major depressive disorderJournal of Affective Disorders, 2021
- Digital privacy in mental healthcare: current issues and recommendations for technology useCurrent Opinion in Psychology, 2020
- Automated assessment of psychiatric disorders using speech: A systematic reviewLaryngoscope Investigative Otolaryngology, 2020
- Feature Augmenting Networks for Improving Depression Severity Estimation From Speech SignalsIEEE Access, 2020
- Automated speech-based screening of depression using deep convolutional neural networksProcedia Computer Science, 2019
- A review of depression and suicide risk assessment using speech analysisSpeech Communication, 2015
- Linguistic Analysis of Speech in Affective DisordersArchives of General Psychiatry, 1976