The short answer
Open the editor's Voice tab. Install the voice engine once (about 4 GB). Under My voice, confirm This is my own voice and choose Record a sample to read a short script aloud. Then, under Translate video, pick a Language, choose who translates under Translated by, keep Voice on, and click Translate to your language. Creavit Studio replaces the speech with your own voice in that language, timed to the original, and can switch the captions to the new language too.
The voice engine is Chatterbox Multilingual from Resemble AI (MIT license). It runs on your Mac, using the GPU where it can and the CPU as a fallback, and it requires a Mac with Apple silicon (on other Macs, only the captions can be translated). It clones your voice from a 6 to 15 second sample, so there is no long training step.
What it does, and who it is for
Dubbing in your own voice lets one recording reach viewers in other languages without hiring a voice actor or recording every version again. A product demo in English can become a Spanish, German and Japanese version that still sounds like you.
It suits creators, founders and educators who speak in their own videos: tutorials, product walkthroughs, course lessons and short social clips. It is built for your own voice only. The app asks you to confirm that the sample is your voice before it learns from it.
Supported languages
The engine speaks 23 languages: Arabic, Danish, German, Greek, English, Spanish, Finnish, French, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Dutch, Norwegian, Polish, Portuguese, Russian, Swedish, Swahili, Turkish and Chinese.
Step by step
1. Finish the edit first
Dub the final cut. If you remove silences, trim mistakes or rearrange sections after dubbing, the new speech no longer lines up with the picture. A clean edit also gives the translation clearer sentences to work with. See how to remove silence from a screen recording for that step.
2. Install the voice engine
Open the Voice tab and click Install voice engine. It is a one-time download of about 4 GB, and you can keep editing while it installs. When it finishes, the panel shows Ready and notes that the engine runs on this Mac.
The voice engine needs a Mac with Apple silicon. On an Intel Mac the panel says so, and you can still translate the captions.
If you need the disk space back later, Remove deletes the engine. Your voice sample is kept.
3. Create your voice sample
Under My voice, tick This is my own voice. The note below it reads "Only your own voice. The sample stays on this Mac."
Then choose one of two options:
- Record a sample: the panel shows a short script to read and a timer. Read it aloud in a quiet room, at your normal pace, with the microphone you use for your videos. The app needs at least 8 seconds and suggests about 15 to 20; it keeps a clean clip of 6 to 15 seconds from your reading. Click Stop when you finish.
- From this video: if the speaker in the current video is you, Creavit Studio can learn your voice from the recording itself.
A good sample matters more than anything else. Avoid background music, echo and whispering. You can play the sample back, and record a new sample if it does not sound like you.
4. Choose the language and translator
Under Translate video, choose the target Language. Then choose who translates under Translated by:
- The built-in translator runs on this Mac and works offline. It is an m2m100 model (about 2 GB). The panel describes it honestly: "Runs on this Mac, more literal than an AI." It needs the voice engine installed first.
- An AI agent (Claude Code, Codex or Local AI) usually gives more natural phrasing and handles idioms and product terms better.
5. Set the output options
Three toggles control the result:
- Voice: "Your voice, timed to the original." Leave it on to dub. Turn it off if you only want translated captions.
- Keep original quietly: keeps the original speech underneath at a low level. Off mutes the original sound.
- Captions: switches the captions to the new language.
6. Translate
Click Translate to followed by the language name, for example Translate to Spanish. The panel shows progress for each stage, and you can cancel.
When it finishes, a summary line tells you how many voice takes were made and how many were sped up to fit the original timing. Open Translated lines to compare each original sentence with its translation. Translated captions are timed to the dubbed speech. If you do not like the result, press ⌘Z to undo the whole run.
7. Review before you export
Watch the whole dubbed video with sound. Check names, product terms and numbers, which translators often get wrong. Check the places where takes were sped up: if a line sounds rushed, shorten the original sentence or ask the AI translator for a more concise version, then translate again.
Limits and responsible use
- Apple silicon only for the voice engine. Caption translation works without it.
- Your own voice only. Do not use a sample of someone else's voice. The confirmation checkbox exists for this reason.
- Watermarked output. Speech made by the engine carries Resemble AI's Perth watermark. It is inaudible, and it allows the audio to be identified as generated.
- Timing is approximate. Languages differ in length. When a translated line runs long, the app speeds that take up to fit, which can sound hurried.
- Literal translation. The built-in translator is more literal than an AI agent. For anything customer-facing, use an agent or have a native speaker review the lines.
- Accent and emotion. The clone follows your voice, but delivery in a language you do not speak may not match how a native speaker would say it.
When to dub, and when to caption
Dubbing is worth it when the viewer needs to listen while watching: tutorials where eyes stay on the screen, talking-head videos, and courses. For short social clips viewed with sound off, translated captions alone may be enough, and they are faster to check. You can do both in one pass.
A pre-export checklist
- The edit is final before dubbing.
- The voice sample is clean and sounds like you.
- The target language and translator are the ones you intended.
- Names, numbers and product terms are correct in the translated lines.
- Sped-up takes still sound natural.
- Captions match the dubbed speech.
Common questions
Does my voice leave my Mac?
The sample stays on your Mac, and the voice engine runs locally. If you choose an AI agent for translation, the text to translate goes to that agent (Claude Code and Codex use their own services; Local AI stays on your Mac).
How long does the sample need to be?
Read for at least 8 seconds; about 15 to 20 seconds is better. The app keeps a clip of 6 to 15 seconds as your voice sample.
Which Macs can dub?
The voice engine needs a Mac with Apple silicon. On other Macs, Creavit Studio can still translate the captions, but it cannot make the dubbed voice.
Can I dub a video in which someone else speaks?
No. The feature is for your own voice, and the app asks you to confirm that before creating a sample.
Keep going
The AI video dubbing page gives an overview. Before dubbing, read how to add captions to a screen recording. After dubbing, you can schedule each language version to YouTube, TikTok, Instagram and X. Plans are on the pricing page.