Video translator and AI dubbing
A video translator with AI dubbing built in: give it a link from TikTok, Instagram, YouTube, Facebook, X, Reddit or any other supported platform, or upload a file. The speech is written down with timings, carried over line by line, then voiced again in the language you chose, so the clip plays back in it. Free AI video translation, translate a video to another language, add subtitles or dub it with AI voices.
Paste a video link
Dub into
Voice
Delivery
One voice reads the whole video and lips will not match, lines are placed in their own slots instead. Clips up to ten minutes.
Only subtitles? Use the transcript generator.
See the dubbing flow, step by step
A walkthrough of one video being spoken back in another language.
8%
Ready to play or save
- Dubbed video, Spanish voice track
- Translated subtitle file, .srt
- Voice only audio, .mp3
Why people use it
Files stream straight to your device. We keep no copy of the video or the link.
No sign up, no email, no tracking pixel. Paste a link and save the file.
Every tool on the site is free to use, with no daily download cap.
We pick the highest stream the source offers, up to 4K where it exists.
One paste, one click. No queues, no waiting rooms, no forced ads between steps.
Runs in any modern browser on phone, tablet and desktop. Nothing to install.
How to dub a video
Pick dubbed video, then paste the link or choose a file. If you already know what language is spoken, pick it so the speech model has a head start.
Choose the language you want it dubbed into and a voice, then press dub. You can keep the original audio quietly underneath so music and ambience survive.
The dubbed clip plays right on the page. Download it as MP4, take the new voice track on its own as MP3, or grab the subtitles as SRT, VTT or plain text.
Three ways to use it
Dubbed video replaces the soundtrack with an AI voice reading the translation in time with the picture.
Voice track gives you the new audio as an MP3, useful when you want to mix it yourself in an editor.
Subtitles is the original behaviour, translated captions that keep the timings of the source.
Subtitles that stay in sync
The translation is done one caption at a time rather than as a wall of text, so every line keeps the start and end time it had in the original.
That means an SRT file dropped into CapCut, Premiere or a player shows the right words at the right moment without any manual fixing.
Honest limits
One voice reads the whole video, so separate speakers are not given separate voices.
Lip movement is not matched. Lines are placed in their own slots, and a dense line is sped up slightly rather than running into the next one.
The rebuild happens in your browser, so clips are capped around ten minutes.
When a link will not open
Private accounts, deleted posts and anything behind a login cannot be read, because nothing here signs in anywhere.
In that case download the video first, then upload the file here and dub it from the file instead.
Three steps, not one
Translating a video means transcribing the speech, translating that text, then producing the new audio or captions.
Each step depends on the one before it. A mumbled line transcribes badly, and a bad transcript translates into confident nonsense.
Clear speech with little background music gives by far the best result, which is why interviews and talking heads work better than club footage.
Where machine translation slips
Names, brands, in jokes and slang are the usual casualties, because they have no dictionary equivalent.
Idioms translate literally more often than not, so a line that sounded natural becomes strange in the target language.
Reading the translated text once before publishing catches most of it in a minute or two.
Dubbed audio versus subtitles
Subtitles keep the original voice and let viewers hear tone, which matters for interviews and personal content.
Dubbing suits instructional footage where the viewer needs their hands and eyes on what is being shown.
Timing rarely matches exactly between languages, so dubbed lines can run slightly long against the picture.
Good to know
Speech is translated, not the picture
The translator works on what people say. Text that appears inside the video itself stays in its original language.
Video transcript generatorCheck names and technical words
Names, brands and specialist terms are the most likely to come out wrong. Read those parts before you publish the translation.
Download YouTube subtitlesClear speech translates best
One person speaking clearly gives the most accurate result. Overlapping voices or loud background music make any translation weaker.
Questions people ask
What does the video translator do?
It translates what is said in a video and gives it back three ways, as a dubbed video with an AI voice, as the voice track on its own, or as translated subtitles that keep the original timings.
How do I translate a video into another language?
Paste the link or upload the file, pick the language you want, then press dub. The clip plays back in that language on the page and you can download it.
Is the video translator free?
Yes. Dubbing, the voice track and the subtitle files are all free, with no account and no watermark.
How long can the video be?
Around ten minutes, because the video is rebuilt in your own browser rather than on a server.
What is AI video dubbing?
It transcribes what is said, translates it, then speaks the translation with a synthetic voice placed on the original timings, so the video plays in the new language.
Is the video dubbing free?
Yes. No account, no payment step and no watermark on anything you download.
Which languages can it dub into?
Every widely used language is in the list, and the speech side recognises dozens more automatically.
Does it match lip movement?
No. Lines land in the right slot in time, but mouths will not agree with the new audio.
Can I keep the original music?
Yes. Leave the option on and the original audio stays quietly underneath the new voice.
Can it dub a video I have on my phone?
Yes, upload the file instead of pasting a link and it runs the same way.
Do you store the video or the transcript?
No. The audio is read once to produce the text and nothing is written to storage.
Does the speaker's voice stay the same?
Dubbed audio is generated speech, so it will not sound like the original speaker. Subtitles are the option that keeps the real voice.
How accurate is it?
Clear single speaker audio comes out close. Crosstalk, heavy accents and loud music all lower accuracy noticeably.
Can I edit the translation?
Yes. The text is produced first, so you can correct names and terms before anything is finalised.