A Timestamped Pronunciation Decision Sheet Before the Final Narration Edit

Pronunciation questions are cheapest to settle before the final edit. Once a narration is trimmed, paced and mixed, a corrected name means re-recording or patching, and every patch risks a different tone. This proposed workflow, an example and not an observed client result, uses one timestamped decision sheet that the editor and the client complete before the final narration edit starts.

It is about the content of the decisions, not about how files are named or mixes are labelled.

Inputs

The rough narration recording or a clean read-through with timestamps.
A list of every proper name, place, brand, abbreviation and borrowed word the editor noticed or the script contains.
For each item, any public reference the client already trusts, such as the organisation's own page or a published recording, plus the client's own stated preference.
The name of one person on the client side who has final say.

The sheet

Use one row per occurrence group. Columns:

Timestamp of first occurrence, and further timestamps in the same cell.
Written form as in the script.
How it is currently spoken in the recording, written in plain syllables you invent for the sheet, such as KAIR-oh versus kuh-ROH.
Proposed decision.
Who decided and how, for example client email or call.
Status: open, decided, or needs new recording.

Syllable spelling is only a note for the team, not a claim about the correct sound.

Decision branches

The current reading matches the client's preference: mark decided, no action.
The reading differs and the item appears once: mark needs new recording, and group it with other single lines so one short pickup session covers them.
The reading differs and the item appears many times: ask the client whether a new read of those lines is acceptable before any editing time is spent.
Two acceptable readings exist and the client has no preference: pick one, record that the choice is editorial, and use it consistently.
The client is unsure: do not guess. Ask them to listen to the timestamped clip and choose. If the item concerns a person's own name, ask for that person's stated preference through the client, not from the editor's assumption.

Stop conditions

Pause the final edit if any row is still open on a name of a real person, an organisation, or any item the client marked as important. Do not start fine pacing work on a section whose terms are unresolved, because pickups change timing. If the client's decision maker is unreachable, do not pick a reading on their behalf; deliver a note listing the open rows and continue only with sections free of them.

Example

Imagine a fictional narration for a gardening society. The script mentions a founder surname and the word for a plant genus. At 03:12 the narrator says the surname with stress on the first syllable, and the sheet shows it again at 11:40 and 24:05. The client replies that the stress belongs on the second syllable. The row becomes needs new recording with three timestamps. The genus word appears once at 07:55 and the client has no preference, so the editor records an editorial choice and keeps it consistent across the episode. The editor books one pickup session for the three surname lines and does not touch the pacing of those lines until the new takes arrive.

Sending the sheet

Send the sheet with short audio clips for each unresolved row, not the whole file, so the client listens for seconds rather than minutes. Ask for answers in the sheet itself, not scattered across messages, so decisions live in one place.

Acceptance test

Try this on one short episode, around ten minutes.

Build the sheet from a rough read. Aim for every name and unusual term, even if the list is only a few rows.
Give the sheet to a colleague and ask them to open the clip for any three timestamps. Pass if each timestamp lands on the right word within a couple of seconds.
Ask which rows block the final edit. Pass if their answer matches the stop conditions above.
After the pickup, listen again at every timestamp listed. Pass if each occurrence matches the decision. Record any row that still differs and treat it as open.

Conclusion

A decision sheet turns scattered corrections into a short list that can be answered once. Keep the timestamps accurate, keep the decision maker named, and let the stop conditions protect the edit from rework. The small test shows quickly whether the sheet is usable before you rely on it for a larger project.