Moonlight Reader is privacy-first. We store voices, not texts.
Your text is processed in-memory only to synthesize audio. Logs redact text and only keep counts and timings.
Uploads are stored in an uploads/ bucket with 1–24h TTL and deleted after voice creation. Outputs are delivered via pre-signed URLs with short expiries and auto-purged lifecycle rules.
Creating a voice requires explicit consent. You can revoke a voice at any time, which disables further TTS and revokes it with the provider.