Working with a recording studio

Generated voice-over is often temporary — a way to hear the whole script before real actors record it. This page is the round trip to a studio: a script out, recordings back, each landing on the line it was recorded for. It needs no AI at all: the script works from the casting alone.

All three actions are in the toolbar of the Lines page (producing the lines).

Export a recording script

1. In Lines, press Export…. The dialog is Export recording script.

2. Choose what goes in:

   Option	What it does
   Languages	Which languages get rows
   What to include	Everything, Only lines with no approved audio, or Only lines whose text changed after recording
   Characters	Untick a voice to leave its lines out
   Only the board open in Lines	One board instead of the whole project

3. Press Export N lines.

You get a tab-separated file, recording-script-«date».tsv — tab-separated because dialogue contains commas, and it opens in a spreadsheet as it is. Its columns:

Column	What it holds
Line ID	The line’s stable identity. The recordings must come back with a filename that starts with it.
Character	Who speaks
Voice	The voice they are cast to — the casting map
Cue · Context · Inflection · Effect	Always present, and empty for now, so a studio template expecting them still lines up
Text	The line
Filename	The name the recording should come back as. A suggestion — see below.
Status	The line’s current take: none when it has no audio — the studio’s cue that this one is theirs to record — placeholder for an AI draft, delivered, or final for an approved take

If any of the exported lines use a name from the pronunciation dictionary, a second file comes with it, pronunciations-«date».tsv, with Term, Pronunciation and Note — only the names those lines actually use.

The filename column is a suggestion. A studio may rename files; as long as a name still starts with the line id, the import finds its line — «line id»_v2.wav included.

Exporting also remembers what each line said when it went out, so an import can tell you which lines were rewritten after the studio booked them.

The recording script needs the Pro plan or above. Below that, the export button carries the plan’s badge and explains the gate when pressed.

Report which lines ship AI audio

AI audio report, in the same dialog, downloads ai-audio-report-«date».tsv: every line in the chosen languages and board, with Line ID, Language, Character, Voice, Text and AI generated.

It counts the take that currently plays, not the history — so it answers what the build ships today. That is the answer a storefront’s AI-disclosure question needs. It is free on every plan.

Import the recordings

1. Switch the project header to the language the recordings are in. The dialog shows it — Language: it — and every file is imported into that language.

2. In Lines, open the board and press Import recordings.

3. Drop the folder the studio delivered, or click to browse for it. If the studio sent back the recording script with its Filename column filled in, drop it in with the audio: Using «sheet» — N lines mapped by the sheet. Files it does not mention are matched by name.

4. Read the review. Nothing is written yet.

   Group	What happens
   Matched	Imported. Nothing is in the way.
   Already has audio	Skipped unless you tick them. The take they replace is kept in the line’s history.
   Text changed since you exported	The line was rewritten after the script went out, so the recording is of the older words. Tick any you still want.
   Two files for one line	Never picked for you. Ticking both keeps both, as takes.
   Matched no line	No filename starts with a line id — Did the studio rename them?
   Expected a recording and got none	The lines that are still owed: the pickup list

5. Press Import N recordings.

Each recording is uploaded to Media under Voiceovers / «board name» and becomes its line’s current take, as delivered. It is never marked as AI audio, costs no credits, and from then on is out of reach of Generate and of a re-cast. Stop leaves the recordings already imported in place.

The upload follows the normal Media limits: your plan’s per-file limit and the project’s storage.

Match clips already in Media

If the files are already in the media library, Match from library in the Lines toolbar attaches them without a dialog: any clip whose filename starts with a line id goes on that line, in the current language. Lines that already have audio are skipped.

It reports Matched N lines by filename. — or, when nothing matches, that the match only works on files named against an exported recording script. It never guesses from a human-chosen name; attach those by hand on the row.

Pickups

When lines change after a session, export again with Only lines whose text changed after recording, or Only lines with no approved audio, and send just those. In Lines, the text changed badge marks the same lines, and Only AI audio shows which lines still play generated audio.