README.md

---
title: FutureCafe Voice Core (Private)
emoji: ☎️
colorFrom: indigo
colorTo: blue
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
---

# FutureCafe Voice Core (Private)

This Space runs the **full** Gradio app (voice + SMS). It’s **private** and will be called by a public wrapper Space via `gradio_client`.

## Run

- Uses **Piper** TTS model at `models/piper/en_US-amy-medium.onnx`
- Uses **faster-whisper** (tiny) for ASR

### Environment variables (set in Space → Settings → Secrets)
- `BACKEND_LLM=openai` (or `groq`)
- If `openai`: `OPENAI_API_KEY=<your-key>`
- If `groq`: `GROQ_API_KEY=<your-key>`
- `TTS_ENGINE=piper`
- `PIPER_MODEL=models/piper/en_US-amy-medium.onnx`

This project writes generated audio files into `runtime/audio`.