L'électron rare
491e6ad8a4
chantier-1+2: voice cloning XTTS-v2 + génération musicale /compose
## Chantier 1 — Voice Cloning
- scripts/xtts_clone.py: Coqui XTTS-v2 voice cloning (6s sample → voix)
- ws-chat.ts: synthesizeTTS avec détection sample → XTTS ou Piper
- PersonaDetail.tsx: upload/delete sample vocal
- app.ts: endpoints voice-sample (POST/GET/DELETE)
## Chantier 2 — Génération Musicale
- scripts/compose_music.py: ACE-Step fallback MusicGen
- /compose command (5min timeout, broadcast audio base64)
- Chat.tsx: player <audio controls> inline
- Eno persona (musique générative/ambient)
- Pharmacius: routing @Eno
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>