EpiCast HeAR Mobile (ONNX)
ONNX-exported HeAR embedding model + trained sklearn classifier for cough audio classification.
Base Model
google/health-acoustic-representations (Google Health AI Developer Foundations)
Details
- HeAR ViT encoder exported to ONNX (1.2 GB)
- Trained on COUGHVID dataset (25,000+ cough recordings)
- 3-class classification: healthy, symptomatic, COVID-19
- 512-dim embeddings + sklearn Gradient Boosting classifier
- Ready for on-device deployment via ONNX Runtime + CoreML
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support