- Remove duplicate TTS_MODE/TTS_AUTO_MODE — single TTS_AUTO_MODE now
defaults to 'always' and accepts TTS_MODE as backward compat alias
(on_voice_input → inbound)
- Fix default voice: en-US-JennyMultilingualNeural does not exist,
changed to en-US-EmmaMultilingualNeural (verified working)
- Remove dead --format arg from edge-tts CLI spawn (not a valid flag,
format is determined by file extension of --write-media)
- Remove node-edge-tts dynamic import dead code (package removed)
- Fix bin/edge-tts wrapper: remove hardcoded /home/mevy path, use
PROJECT_ROOT-relative venv with system edge-tts as fallback
- TTS_PROVIDER now defaults to 'edge-cli' instead of empty string
---
Build: pass | Tests: pass — 19 passed (1 file) (Sam & Claude)
