Dub each speaker in their own voice

Applies to 0.11Updated 6 Nov 20262 min read

In Dub from Captions, press Detect Speakers, name each person, then Clone their voice. Every phrase is dubbed in the voice of the person who says it.

Interviews, podcasts and dialogue sound right when each person keeps a voice of their own.

1. Translate the captions (see Translate captions into another language).
2. Choose Sequence › Languages › Dub from Captions….
3. In the Speakers row, press Detect Speakers. Choose Any number or tell it how many people talk (2 to 6). Press Detect Again to retry.
4. Type a name for each speaker. Giving two speakers the same name joins them into one person.
5. Press Clone next to a speaker. The app builds the voice from that person's clearest lines, up to 25 seconds of them.
6. Start the dub.

Detect Speakers listens only to the track in the Dub from row. With no track chosen, it listens to the tracks you can hear (the solo ones, if any). Each speaker has two listen buttons: their own clearest lines, and the voice they will get. Cloning needs at least 4 seconds of that person speaking.

To set the speaker of a single caption yourself, use Properties › Caption › Speaker; Nobody leaves the caption without a speaker.

When the captions are a little off

If the voice and the captions don't line up, press Line Up Captions. It fixes them in one step, which one Undo reverses.

Phrases that are too long

A translation is often longer than the original. The Timing options decide what happens:

• Speed limit: how much faster a phrase may be spoken.
• Use the pauses between phrases and Let phrases shift up to a few seconds: give a phrase room from the silence around it.
• Fill caption time, slowest: slows down short phrases without changing the pitch (off at first).
• Still too long: by default Place it and flag it, so you can review it.

When the dub is done, Dub Again: Too Long redoes only the flagged phrases, Dub the Rest finishes any that were skipped, and Again redoes a single phrase.

Included in Pro and in the 3-day free trial. On Creator it carries the PRO badge. Not sure what your plan includes? Open Settings › Licence in the app.

Detecting and cloning speakers uses the voice component. If it is missing, see Voice, speech or AI components didn't finish installing.

If it still doesn't work: open a ticket in Account › Support and we'll help.