Quick Answer
To dub a video with AI in Percify, open the Playground, upload the MP4 or MOV to ElevenLabs Dubbing or HeyGen Video Translate, pick the target language and run it; there is no script step. ElevenLabs Dubbing is 1 credit a second across 33 languages and keeps the original voice; HeyGen Video Translate is 4 credits a second (8 in precision mode) with 178 options. For a new video, make the audio in the target language and lip sync a photo to it from 2 credits a second.
Applicability: This guide applies to content creators, marketers, educators, and businesses looking to localize video content efficiently. It does NOT apply to users seeking basic video editing or voice-over recording without AI avatar generation.
Dub a video with AI in Percify: ElevenLabs Dubbing (1 credit a second, 33 languages) or HeyGen Video Translate, or lip sync a new video in any language.
There are two ways to put another language on a video with Percify, and they suit different jobs. If the video already exists, you dub it: upload it to a dubbing model in the Playground, pick the target language, and get the same video back speaking that language. If you are making a new talking video, you record or generate the audio in the target language and lip sync a photo to it. This guide walks through both, with the price of each.
Which way to go
| You have | Use | What you get | Price |
|---|---|---|---|
| A finished video with speech | ElevenLabs Dubbing | The same video, translated and dubbed, keeping the original speaker's voice | 1 credit a second |
| A finished video with speech | HeyGen Video Translate | The same video translated, with regional options | 4 credits a second, 8 in precision mode |
| A photo and a script | Lip sync | A new talking video in the language of your audio | From 2 credits a second |
Dub an existing video, step by step
Dubbing runs in the Playground, which is in the desktop web app at app.percify.io ↗, under Playground in the sidebar.
- Get the video ready. An MP4 or MOV file with speech in it. Trim it to the part you need first, because both models charge for the whole length.
- Open a dubbing model. In the Playground, open ElevenLabs Dubbing or HeyGen Video Translate.
- Upload the video. There is no script to type: the model listens to the speech and translates it itself.
- Pick the language. ElevenLabs Dubbing has 33 target languages; you can also name the original language from 57, or leave it on Auto. HeyGen Video Translate has 178 language and dialect options, such as Spanish (Mexico) or Portuguese (Brazil).
- Run it and review. Watch the result before you publish, ideally with a native speaker. If a run fails, or is still unfinished after 15 minutes, the credits are refunded.
If the video will not upload, re-encode it to a standard MP4. The message "Could not read the video duration" means the server could not measure its length.
What dubbing costs
The server measures the length of the uploaded video and charges from that.
| Video length | ElevenLabs Dubbing | HeyGen Video Translate, speed | HeyGen Video Translate, precision |
|---|---|---|---|
| 30 seconds | 30 | 120 | 240 |
| 1 minute | 60 | 240 | 480 |
| 2 minutes | 120 | 480 | 960 |
Test with a short clip on ElevenLabs Dubbing, the cheaper of the two, before you send a long video. Creating an account is free and needs no card; dubbing takes credits from a plan, from $12 a month. Subscribers can top up with one-time credit packs. The pricing page has the current plans.
Make a new talking video in another language
When there is no video yet, dubbing is the wrong tool. Make the audio in the target language first, then lip sync a face to it.
- Get the audio. Record it yourself in that language, or generate it. To speak it in your own voice, clone your voice from a sample of at least 6 seconds (15 to 30 is better); XTTS-v2 in the Playground speaks a cloned voice in 16 languages. Saving a cloned voice is free.
- Pick the face. One clear photo where the face fills about a third of the frame. Your own photo works as is.
- Lip sync it. The mouth, jaw and head move to the audio. The fast model is 2 credits a second at 480p, so a 20 second clip is 40 credits. See lip sync a photo to audio.
Before you publish
- Check names, numbers and product terms in the translation; these are where machine translation slips most.
- Translated speech often runs longer or shorter than the original. Recut the video if a line now overruns a scene.
- Say that the video is AI dubbed where your platform or market expects it.
For the reference version of these steps, see the docs: Dub a video into another language ↗.
Ready to Create Your Own AI Avatar?
Make a talking video of yourself from one photo, in your own voice. No watermark on any plan.
Cancel anytime from your profile
Got questions?
Frequently asked
Open the Playground in the desktop web app, choose ElevenLabs Dubbing or HeyGen Video Translate, upload the MP4 or MOV, pick the target language and run it. There is no script to type: the model translates the speech itself.
ElevenLabs Dubbing is 1 credit a second, so a minute is 60 credits. HeyGen Video Translate is 4 credits a second in speed mode and 8 in precision mode. Failed runs are refunded. Creating an account is free; dubbing takes credits from a plan, from $12 a month.
ElevenLabs Dubbing has 33 target languages and can detect the original language automatically. HeyGen Video Translate has 178 language and dialect options.
ElevenLabs Dubbing translates and dubs the video while keeping the original speaker's voice. For a new video in your own voice, clone it (saving the clone is free) and speak the script in the target language with XTTS-v2, which covers 16 languages.
Yes. Record or generate the audio in that language, then lip sync a photo to it. The fast lip sync model is 2 credits a second at 480p, so a 20 second clip is 40 credits.
