LiveKit voice agent worker for TurfVoice handles speech-to-text, text-to-speech.
63
LiveKit Agents worker that powers voice interaction for TurfVoice. Listens to the caller's microphone (STT), sends each completed turn to the turf-backend AI assistant, and speaks the reply back (TTS) — sharing the same conversation history as the text chat in turf-frontend.
turf-backend's /voice/tokenPOST /assistant/chat, so text and voice share one conversationdocker run \
-e LIVEKIT_URL=wss://your-project.livekit.cloud \
-e LIVEKIT_API_KEY=your-livekit-key \
-e LIVEKIT_API_SECRET=your-livekit-secret \
-e GROQ_API_KEY=your-groq-key \
-e OPENAI_API_KEY=your-openai-key \
-e MONGODB_URI=mongodb+srv://... \
-e BACKEND_INTERNAL_URL=http://turf-backend:8000 \
roboflowhnstesting01/turf-worker:latest
| Variable | Description |
|---|---|
LIVEKIT_URL / LIVEKIT_API_KEY / LIVEKIT_API_SECRET | LiveKit project credentials |
GROQ_API_KEY | Speech-to-text and (primary) text-to-speech |
OPENAI_API_KEY | Optional fallback text-to-speech provider |
MONGODB_URI | MongoDB connection, used to persist voice conversation turns |
BACKEND_INTERNAL_URL | Internal URL of turf-backend, used to call the shared assistant logic |
roboflowhnstesting01/turf-backend — FastAPI backend (mints voice tokens, hosts the shared assistant logic)roboflowhnstesting01/turf-frontend — Next.js frontend (starts voice sessions from the AI Assistant page)Content type
Image
Digest
sha256:09338ccc8…
Size
3.4 GB
Last updated
about 1 month ago
docker pull roboflowhnstesting01/turf-worker