MiniMax Voice — Cloud STT and TTS with Local Fallback

MiniMax Voice — Cloud STT and TTS with Local Fallback A architecture diagram generated by Archify. POST /transcribe contract preserved speak_direct.sh contract preserved ESP32-S3 · wake word + mic · Architecture component ESP32-S3 wake word + mic voice_bridge · OpenWakeWord (local) · POST /transcribe contract preserved · buffers command voice_bridge OpenWakeWord (local) buffers command voice_stt_shim · POST /transcribe :5001 · POST /transcribe contract preserved · fallback chain voice_stt_shim POST /transcribe :5001 fallback chain Cloud STT · OpenRouter whisper-1 / MiniMax asr-1.0 · POST /transcribe contract preserved Cloud STT OpenRouter whisper-1 / MiniMax asr-1.0 voice_whisper_fallback · faster-whisper base.en :5002 · POST /transcribe contract preserved voice_whisper_fallback faster-whisper base.en :5002 voice-agent (Jervis) · LLM tool-calling :8501 · Architecture component voice-agent (Jervis) LLM tool-calling :8501 Home Assistant · lights, timers, calendar :8123 · Architecture component Home Assistant lights, timers, calendar :8123 tts_router · MiniMax T2A v2 · speak_direct.sh contract preserved · speak_direct.sh tts_router MiniMax T2A v2 speak_direct.sh piper_tts · local fallback · speak_direct.sh contract preserved piper_tts local fallback Snapcast · 48 kHz stereo :4953 · Architecture component Snapcast 48 kHz stereo :4953 House speakers · MainAudio group · Architecture component House speakers MainAudio group MQTT audio POST /transcribe primary fallback transcript tool calls speak_direct.sh cloud failure reply audio Legend Backend Cloud External