Listening field note
How should a timecoded interview index work?
Each transcript segment may carry a start and optional end time, while each selected passage inherits the boundary locators and can seek authorized local media to the first timed segment.
Validate shape and order
Reject impossible timecodes and end-before-start ranges; surface reversed order, overlaps, long gaps, and missing timecodes for review. Write the decision in plain language and retain the original source separately so the index can be audited against something outside the browser project.
Time is a locator, not truth
A timestamp does not prove the transcript is synchronized, the speaker is correctly labelled, or the correct recording is linked. Review the result as a retrieval aid rather than a conclusion. Human-authored fields must remain visibly distinct from exact source-derived text.
Keep no-timecode text usable
Untimed transcripts can still support exact passage indexing, but media retrieval and timed handoff remain incomplete and should be labelled. Carry the limitation into the frozen edition and passive handoff so a recipient can understand what was checked, what was not checked, and what remains their responsibility.