DeepFilterNet speech enhancement HTTP API for the Vocence platform.
Upload noisy audio → get clean, noise-reduced audio back. Uses DeepFilterNet3 (48 kHz full-band noise suppression).
| Method | Path | Auth | Description |
|---|---|---|---|
| POST | /enhance |
none | Upload audio file → enhanced WAV |
| GET | /health |
none | Basic health check |
| GET | /healthz |
bearer | Ops dashboard health probe |
| GET | /metrics |
bearer | Ops dashboard metrics scrape |
docker build -t vocence/voice-dubbing:latest .
docker run -d --gpus all -p 8116:8116 -e DUBBING_API_KEY=mykey vocence/voice-dubbing:latest
curl -F audio=@noisy.wav http://localhost:8116/enhance -o clean.wav| Var | Default | Description |
|---|---|---|
PORT |
8116 |
HTTP port |
HOST |
0.0.0.0 |
Bind address |
DUBBING_API_KEY |
(none) | Bearer token for /healthz + /metrics |
DUBBING_CAP |
4 |
Max concurrent enhance requests |
DUBBING_MODEL |
DeepFilterNet3 |
Model variant |