BAYMAX-X Persian voice server (lite)

Small xiaozhi-v1-compatible server for the BAYMAX-X robot (ESP32 client, robot-phone mic/speaker). Groq Whisper (fa) → Groq LLM (qwen3.8-27b, gpt-oss fallbacks) with device MCP tools → Edge neural fa-IR TTS → PCM16 16 kHz. Runs in ~60 MB RAM (aiohttp + edge-tts + miniaudio, no torch/ffmpeg).

Env: GROQ_API_KEY, BAYMAX_SECRET (required); PORT, LLM_MODELS, ASR_MODEL, TTS_VOICE, AUTH_ENABLED; optional GOOGLE_ASR_API_KEY (host secret only) and AUTH_MAX_AGE_SECONDS (default 900). Endpoints: wss://<host>/xiaozhi/v1/ (HMAC bearer auth), GET /health. No secrets are stored in this repository.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support