Keep the name. Help the voice say it.

Choose a spelling you can read aloud.

Try separating a joined name, writing out a number, or spacing the letters of an abbreviation. The initial Remixure → Remixure and AI → A I entries are our original example. Replace them with wording appropriate to your script and chosen language. A spelling guides a text-to-speech model and cannot guarantee phonemes, tone or accent.

A term matches only at a word or phrase boundary. AI does not match inside “paid.” If two entries overlap, such as New York and York, New York wins at that position. Replacements are not scanned again, so a replacement cannot trigger a chain of other entries. Use the displayed match count to catch typos and phrases your script never uses.

Take the reviewed wording into your video.

1. Copy the narration draft, or download and copy the JSON text. Keep this tab open while you inspect the result.
2. Open the studio, choose your workspace, and save the video. Expand Create AI narration, then Import narration wording.
3. Paste the JSON and choose Review imported wording. Inspect both scripts, the voice and match count. Use imported wording replaces only that narration form. Undo wording import restores its previous wording while the imported form remains unchanged.
4. Check live voice and pronunciation availability. Choose Review spoken script when entries are present, confirm your rights and explicitly create a candidate. Listen before approving, then save the attached audio.

The portable draft includes no workspace, private media, job, rights approval or saved video ID. Import does not save scenes, reserve quota or generate audio. Saved scenes and existing candidates keep their original wording. Review generated captions separately; the spellings do not automatically rewrite them.

Follow the narration guide · Estimate spoken length · Rehearse the original script