The take that became four languages
A man sits in his study and says four sentences to camera. That one take is the original below, and it is the only thing the Spanish, German and Chinese versions were made from. Nothing was filmed twice.
One take, four languages. Watch the original, then read the other three.
Sound on, or this one makes no sense. The whole demo is what you hear.
Part of Showcase.
What does an AI dubbed video actually look like?
It looks like the clip above, with the language changed underneath it. The face stays put, the voice stays the same, and the mouth is re-timed to fit the new words. Nothing is re-shot and nobody reads a translation. Here you dub a video you already finished, one language per run.
Who it is for
People already making videos here who sell into more than one language.
What you need
A finished video in your library, a saved voice, and the line written in the language you want.
What it costs
80 credits a second of the original on the Plus plan and above, with a floor of 120 credits a run, plus 3 credits a second for the voice take that feeds it.
What it does not do
It will not dub a file you upload from your computer. It does one language per run. And it hands you no subtitle file at the end.
The door itself is dubbing, where the four takes play back to back with a language switcher.
What you are watching
A man sits in his study and says four sentences to camera. One take, and the take never changes. What changes is the language coming out of it.
In the dubbed versions he carries on from where he was, in Spanish, with his mouth moving to the Spanish words. Then the same thing in German, and again in Chinese. Nobody was hired to read a translation and no subtitle is doing the work.
That is what dubbing means here: the video you already have, saying the same thing to someone who does not speak your language.
How this one was made
Four steps, and three of them are the product doing its normal job.
A still, approved first
Before anything moved we made a single picture of the presenter in his study and looked at it properly. Nothing gets rendered until the still is right, because a still costs a fraction of a video and a face is much easier to fix before it starts talking.
The English take
That still became the video above, with its narration already inside it. This is the original, and it is the only one of the four that was not dubbed.
Three dubs, three runs
The line was written out in Spanish, German and Chinese, read in the same voice, and handed back to the English video. The mouth was re-timed to the new words each time. Three languages, three runs, one source video.
The captions came last
The caption tracks were written from the finished audio afterwards, with the transcript tool in the Studio, and saved by hand. A dub does not produce them for you.
Every step except the first is something you can do yourself, on a video of your own.
The same four sentences, four times
The English is what you hear above. The other three are what the dubs say, word for word.
| Language | What he says |
|---|---|
| English | What he saysEvery customer hears you in their own language. Same face, same voice, same message. Record your video once and it speaks to the whole world. Go ahead, switch the language and watch it carry on. |
| Spanish | What he saysCada cliente te escucha en su propio idioma. La misma cara, la misma voz, el mismo mensaje. Graba tu video una vez y le habla al mundo entero. Adelante, cambia el idioma y mira cómo sigue hablando. |
| German | What he saysJeder Kunde hört Sie in seiner eigenen Sprache. Gleiches Gesicht, gleiche Stimme, gleiche Botschaft. Nehmen Sie Ihr Video einmal auf und es spricht zur ganzen Welt. Wechseln Sie einfach die Sprache und es spricht nahtlos weiter. |
| Chinese | What he says每一位客户都能用自己的语言听到你。同样的面孔，同样的声音，同样的信息。视频只需录一次就能对全世界说话。来切换一下语言，看他继续往下讲。 |
The four takes are not the same length
Here is the part nobody shows you. Languages do not take the same amount of time to say the same thing, so the four takes all run to slightly different lengths and the dub follows the words rather than the clock.
Spanish is the long one, by about a second and a half over the English. German lands a little over it. Chinese comes in a shade under.
Plan for that. If your video has to land on an exact runtime in every market, the copy has to be written to that runtime in each language before anything is dubbed.
What this demo does not prove
It is not a file we uploaded
There is no door here that takes a video off your computer and dubs it. The source has to be a video you finished in Kyndrify.
It is one language at a time
Four languages was four separate runs. There is no button that does all ten at once, and three markets is three runs from the same original.
The presenter is not a real person
He is an Avatar we built for this demo, from a picture we approved. He carries no consent record because there is nobody to consent. Your own face doing this is a Digital Twin, and a Digital Twin needs a consent recording before it renders anything.
Your video is not this video
A short piece to camera in good light is the easy case. A long recording with two people talking over each other is not what this door is for.
- AI dubbing
- The finished video kept as it is, with the narration replaced in another language and the mouth re-timed to the new words. Not a subtitle, and not a second shoot.
The transcript
The English take, in the order you hear it.
Read the transcript
0:00Every customer hears you in their own language. Same face, same voice, same message.
0:06Record your video once and it speaks to the whole world.
0:09Go ahead, switch the language and watch it carry on.
Questions about this demo
- Is that the same voice in all four takes?
- Yes. The voice does not change from take to take, which is the point of the demo. What changes is the language it is reading and the mouth movement that goes with it. In your own Workspace the voice is whichever saved voice you hand the dub, including your own.
- Can I dub a video I filmed on my phone?
- No. Dubbing runs on a video you finished in Kyndrify, not on a file from your computer. If you want a video of yourself to dub, you make it here first, from your own face and voice, and then dub that.
- Which languages can I dub into?
- Speech and dubbing work in ten languages: English, Spanish, Portuguese, French, German, Italian, Japanese, Korean, Russian and Chinese. Four of them are on this page.
- Do I need a paid plan for this?
- For the dub, yes: it opens on the Plus plan and above. Making the source video does not. On the Free plan a talking-head video runs to one minute, so you can make something first and decide about dubbing later.
- Do I get subtitles as well?
- No. A dub gives you a video, not a subtitle file. The captions on this page were written from the finished audio afterwards, with the transcript tool in the Studio, and saved by hand.
- Is the man in the video real?
- No. He is an Avatar we made for this demo, so nobody's likeness is being used to sell you something. If you want this done with a real face, that is a Digital Twin: a real person, their own voice, and a consent recording on file before anything renders.
The tools that do this
The door that plays all four takes, and the tools that feed it.
Make the video first, then send it everywhere
Start free and make something worth dubbing. The Free plan begins with 8,400 credits, once. Dubbing opens on the Plus plan, one language per run, on the videos you have already made.