whisper.cpp (STT) and Kokoro (TTS) are implemented and tested, but enabling them still involves the command line. A user should switch voice on from the Model Manager: download runtime, download voice/model, test mic, done. Builds on #63 (voice settings UI), #217 (voice modes), #385 (voice invite deferred out of first-run).
Acceptance
- Model Manager offers STT and TTS cards with install/enable states matching LLM cards
- Mic check and TTS preview run inside the UI
- First voice conversation achievable with zero terminal use, fully offline
From docs/post-alpha-issues.md item 9 (Milestone 2 — voice polish).
whisper.cpp (STT) and Kokoro (TTS) are implemented and tested, but enabling them still involves the command line. A user should switch voice on from the Model Manager: download runtime, download voice/model, test mic, done. Builds on #63 (voice settings UI), #217 (voice modes), #385 (voice invite deferred out of first-run).
Acceptance
From docs/post-alpha-issues.md item 9 (Milestone 2 — voice polish).