# Non-uniform Speaker Disentanglement For Depression Detection From Raw Speech Signals

arXiv: https://arxiv.org/abs/2306.01861 · HTML: https://mlsanity.com/paper/2306.01861/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Jun 2023.

## Built on by (most cited first, 4 of 4)

- [Robust and Explainable Depression Identification from Speech Using Vowel-Based Ensemble Learning Approaches](https://mlsanity.com/paper/2410.18298/) (2024)
- [Personality-guided Public-Private Domain Disentangled Hypergraph-Former Network for Multimodal Depression Detection](https://mlsanity.com/paper/2511.12460/) (2025)
- [DepFlow: Disentangled Speech Generation to Mitigate Semantic Bias in Depression Detection](https://mlsanity.com/paper/2601.00303/) (2026)
- [SNAP: Speaker Nulling for Artifact Projection in Speech Deepfake Detection](https://mlsanity.com/paper/2603.20686/) (2026)

## Abstract (arXiv)

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.