Appearance
Transcript Correction
An audio transcription artifact starts with ASR segments. A user edits segment 12. The product records artifact.update_part, creates a new draft snapshot, then finalizes version 2.
The original ASR output stays immutable. The correction operation links to the user, timestamp, segment id, previous version, new version, and evidence refs.