YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Fine-Tuned WavLM Depression Classifier
A fine-tuned WavLM-based speech classification model developed for the BioCAS2024 Depression Detection Grand Challenge dataset.
Base Model
microsoft/wavlm-base-plus
Model Architecture
- WavLM Base Plus
- Last two WavLM encoder layers fine-tuned
- Mean pooling of hidden representations
- MLP classification head
- 4-class classification
Input
- Mono speech audio
- Sampling rate: 16 kHz
- Segment duration: 5 seconds
Validation Results
| Metric | Score |
|---|---|
| Macro F1 | 0.6451 |
| Balanced Accuracy | 0.6310 |
Important Note
The validation dataset was small and imbalanced, and Label 1 was not correctly classified in the reported validation evaluation.
This model is intended for educational and research purposes only.
It has not been clinically validated and should not be used for medical diagnosis.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support