'Keep voice clips' setting (default off). When on, /api/voice/transcribe
saves the audio (0600) to the owner-only ZFS subvol at /var/lib/void/
voice-clips (CT 311 mp0, replicated to Z3) + a voice_clips row (migration
029, transcript+metadata in void-db). New clips list/play/delete API +
Settings UI. Storage path is configurable (VOICE_CLIPS_DIR).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
faster-whisper (small.en, GPU+CPU fallback) on CT 102 → POST
/api/voice/transcribe (multer→whisper client) → mic in the bubble
records (MediaRecorder), uploads, drops the transcript into the input
to review-and-send. Infra scripts in deploy/whisper/. Retention (P2b)
next. NOTE: mic needs a secure context (the https domain), not the LAN IP.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
