Morrigan

Event-driven prototype: record → STT → LLM → Piper TTS → playback (no Realtime, no WebRTC).

Checking health…

Choose a voice, then record. List comes from the server.

Transcript

Assistant

Timings (ms)

Reply audio