AI dubbing for video ads: one voiceover, many languages
How to turn one video ad into several language versions with AI dubbing: when you need lip sync, how to write the script, what to check before launch.
By Olena, me20 Studio · 3 min read
AI dubbing lets you turn one finished video ad into versions in dozens of languages, in the original speaker's voice, without recasting or reshooting. The key decision is whether a face speaks on screen: a voiceover-only ad needs dubbing, while a presenter or actor on camera also needs lip sync. Here is how we would plan it.
Does your ad need lip sync?
If nobody speaks on camera, voice dubbing alone is enough. Product films, motion graphics and most e-commerce reels only need the new audio track. ElevenLabs Dubbing translates speech into 90+ languages, but its own docs say lip sync is not part of dubbing.
If a person or an AI presenter talks to camera, the mouth has to match the new words. HeyGen video translation covers 175+ languages and dialects with voice cloning and lip sync, and you can pick up to 10 target languages at once.
Write the script for translation, not only for English
A script written with translation in mind gives cleaner dubs in every language. Keep sentences short, avoid puns and idioms that do not travel, and leave a little breathing room in the edit, because translated lines are often longer than the original.
Keep on-screen text out of the voiceover script. Titles, prices and offers are easier to swap in the edit than to fix in a dubbed voice.
Keep the voice, not just the words
Modern dubbing models try to carry the original performance, not only the text. ElevenLabs says Dubbing v2, launched in May 2026, conditions on the original recording, so tone, emotion and pacing carry across languages, and its sync-aware translation lines up where speech starts and stops. It also tells regional variants apart, such as Castilian and Latin American Spanish.
There is a trade-off: the "Speaker similarity" setting runs from 0 to 10 (default 7). Higher values sound more like the original voice but, according to the docs, can sound less natural in a very different language. For new voiceovers rather than dubs, Eleven v4 keeps one voice across 90+ languages and takes delivery directions in plain words, as we covered in our October tools roundup.
Check before you launch
AI translation still needs a native speaker's ear. Both tools let you review the translated script before the final version — HeyGen through Edit & Review, ElevenLabs by editing the transcript in the interface.
Plan the final mix separately. ElevenLabs notes that its dubbed video is a preview, not a production master, so keep music and effects on their own tracks and mix the new voice into your master in the editor. Free plans add a watermark you cannot remove, and only clone voices you have permission to use.
A quick plan
- Decide: voice only, or voice plus lip sync.
- Write a short, translation-friendly script and keep on-screen text separate.
- Test one line per language, then dub the full ad.
- Get a native speaker to check every version.
- Mix the new voices into your master edit.
Want your ad in several languages? Tell us about your project.
P.S. The caption under the cover image is a ready-to-use prompt. Copy it into Midjourney, Higgsfield or any image generator and make your own version.