A voice can be instructional source speech only after evidence proves who recorded it, that they are a verified native speaker, that consent covers the intended use, and that expert and media-integrity checks passed.
- Real recordingThe published voice is a human performance, not text-to-speech or a voice clone.
- Verified native speakerSpeaker and language evidence must meet the publication policy.
- Consent and provenanceUse is tied to traceable consent, dialect, register, and source identity.
- Expert and integrity reviewApproval, checksum, format, duration, and reachability evidence must pass.
Where AI stops
AI may assist draft transcripts, timings, translations, or exercises under review. It may not synthesize, clone, dub, or repair the published instructional voice. Assisted authored content remains draft until an authorized expert approves it.
Private consent documents and reviewer records are not exposed on this public website. The backend publication gate keeps that evidence auditable for authorized staff.