---
title: STT Model
emoji: 🎤
colorFrom: indigo
colorTo: blue
sdk: docker
app_port: 7860
pinned: false
license: mit
---

# RealtimeSTT

FastAPI server running faster-whisper for speech-to-text.

**Endpoints:**
- `POST /transcribe` — upload audio → `{"text": "..."}`
- `GET /` — browser web UI with mic recording
- `GET /health` — health check

**Env vars:**
- `STT_MODEL` — model size (tiny/base/small/medium/large, default: tiny)
- `STT_DEVICE` — cpu or cuda (default: cpu)
- `STT_LANGUAGE` — language code (default: en)