Dubbing translates speech and speaks it back in another language, keeping the delivery. It works two ways. Attach a clip and the mouth is re-synced to the new speech. Attach an audio track and the audio alone comes back translated, with no picture involved.
Open Generate Click Generate in the top bar. You do not need a clip selected, because this starts from the file you attach rather than from what is on the timeline.
Pick "Dub into another language" Type "dub" in the search box and choose it from the Voice & lips group. One goal covers both the video and the audio-only jobs, so there is no second row to hunt for.
Attach a clip, or a track This is the important choice. Attach a video clip and you get lip-synced video back. Attach a voiceover, a recording or an MP3, anything with no picture, and the audio-only task runs instead and hands back a translated track.
Choose the language Thirteen are supported: Spanish, French, German, Italian, Portuguese, Dutch, Polish, Turkish, Hindi, Japanese, Korean, Chinese and English. Pick one and the cost for your exact clip length appears on the Generate button.
Generate, then use the result The finished dub lands in your media library as a new asset, and your original is never modified. Drop it onto a lane, or replace the original track and keep both versions of the project.
Tips
Audio-only dubbing is cheaper and faster than the video route, because nothing has to be re-rendered frame by frame. If the speaker is off camera, use it.
On some models the translated speech carries the original speaker's voice, so a narrator still sounds like themselves in Spanish.
Trim the clip before you dub: models bill the seconds you send them, and there is no point paying to translate a silent tail.
Dubbing and captions are separate passes: generate captions from the dubbed track if you want subtitles in the new language.