YOUR VOICE, TREATED WITH CARE

Privacy at StarlearnAI Voice

We ask for your consent before recording. Your microphone is active only while recording, for up to 60 seconds.

Your recording is uploaded to the local inference service, temporarily decoded to mono audio, and deleted when processing finishes or fails. Audio is not permanently saved by default. Browser preview audio is held in memory until you send it, discard it or leave the page.

We store the transcript, any edits, the AI explanation, optional feedback, a random session identifier, an anonymous browser-profile tester ID, consent version, transcript confirmation and response provider, subject, language, timestamps, model identifiers, processing times, coarse browser/device type and errors. We do not ask for your name, phone number, email or precise location. Avoid mentioning personal details in your question or feedback.

This information helps us evaluate voice-first learning and document real use for NAIC. Only the protected administrator dashboard exposes validation records. Development, validation and production records are kept separate in reporting.

The MVP stores records on the operator’s local machine. There is no automatic deletion schedule yet; the operator must manage retention and delete records when no longer needed. To request deletion during a testing session, show the operator your session ID below.

Your session ID: No session started

A hosted response provider, if configured later, receives the confirmed question and limited learning context; the development guide does not send questions to an external service. AI explanations may be inaccurate. Please verify important facts with a teacher or textbook. Participation is optional; for school or child testing, the operator should obtain the appropriate guardian and school consent.