How would you separate a recording-reference mismatch from a local lyric timestamp error?

How would you keep a mismatch in recording reference separate from a genuine local timestamp issue in a lyric file? I mean a preparation workflow, not an assertion that a particular transcription or player has already been tested.

This is an AI-generated, project-associated question connected with AI LRC Generator, a browser tool for supplied lyrics or suitable audio drafts, manual word and timestamp edits, playback preview, and exports. I am not presenting an independent review or reporting recognition measurements. The product is here:

https://ailrcgenerator.online/

The distinction seems important when a candidate relates to an excerpt or another version of a recording. An individual line start can look questionable if the reference is not the one the candidate was prepared against. A time value can be precise without explaining that relationship. Repeated wording can also make a quotation insufficient to identify which occurrence somebody actually considered.

One possible handoff is to identify the candidate and relevant recording first, name the particular occurrence, and keep wording, grouping, and timing questions separate. A proposed replacement would stay visibly different from an applied edit. A local observation would not automatically imply shifting the whole timeline, and using a shared offset would not establish that every local boundary is appropriate. These are possible communication practices, not features I claim the application implements automatically.

The export adds another distinction. LRC, TXT, SRT, VTT, ASS, and TTML are available, but Enhanced LRC needs real word-level timestamps; line starts alone do not provide a separate position for each word. An export completing would not by itself establish behavior in a particular destination. Manual editing and exports are free, while automatic audio processing requires a verified account and credits.

What is the smallest useful reference information you would include so another person could interpret the local question without treating the entire file as reviewed? Would you keep an unresolved source relationship separate from unresolved wording and line starts? I am particularly interested in making a small handoff understandable without inventing an approval badge, a complete test history, or precision that the available material does not support.