Use AI Video Translation to translate a video to English in LipSync Studio. Upload your source video, choose English, select a translation mode, and generate an English version. Then review the meaning of the dialogue, the spoken English, and the visible mouth movement separately. A convincing mouth performance cannot tell you whether the translation is correct.
This workflow is useful for an on-camera tutorial, a product explanation, or a short course segment that you want English-speaking viewers to follow. Start with a video file you own or have permission to adapt. If it already appears on YouTube, upload your original local export. The workspace does not import YouTube URLs directly.
Compare an existing multilingual example
The AI Video Translation page includes this source clip and its Spanish and Japanese versions. Play the source first, then compare the translated speech, pauses, and visible mouth movement. The translated clips below use Spanish and Japanese. To make your own English version, select English as the target language in the same workspace.
These are existing examples from the AI Video Translation page. The phone-stand script later in this guide is a separate practice exercise you can record and translate yourself.
Decide what the English version needs to change
English subtitles, English dubbing, and lip sync solve different parts of the viewing experience.
| Part of the video | What it changes | What to check |
|---|---|---|
| English subtitles | The words viewers read | Meaning, spelling, reading time, and line breaks |
| English dubbing | The words viewers hear | Meaning, pronunciation, pacing, and speaker identity |
| Lip sync | The visible mouth performance | Whether the mouth movement follows the English recording |
LipSync Studio's translation workflow provides translated speech with lip sync and an option to generate captions. It is a good starting point when the presenter remains on screen. A screen recording with narration may put more weight on speech and captions, while a close-up interview makes mouth timing easier to notice.
Prepare a short source excerpt and a reference translation
Choose an excerpt that contains the things you need to verify. A product demonstration might include a name, a number, and a physical action. An interview might include a speaker change. Keep the original recording available so you can compare it with the English output.
For a first pass, use clear speech with a visible face, as little overlapping dialogue as possible, and enough space around the beginning and end of each line. If you can export from the original project, use a version without burned-in subtitles. Otherwise, the old words may remain on screen underneath the English version.
Here is an original practice script you can record in Spanish, followed by a reference English translation. It is a preparation exercise, rather than a generated example.
Hola, soy Ana. Este es nuestro soporte plegable para teléfono. Presiona el botón azul y espera cinco segundos antes de abrirlo.
Hello, I'm Ana. This is our folding phone stand. Press the blue button and wait five seconds before opening it.
Before generating, write down the details that must survive translation. In this example, the speaker is Ana, the object is a folding phone stand, the button is blue, and the wait is five seconds. A fluent sentence that changes any of those details needs correction.
Keep a small reference list for your own review. Product names should stay consistent throughout a real video. Numbers, units, and instructions deserve particular attention because a small change can alter what the viewer does.
Upload the video and choose English
Open AI Video Translation and sign in. Under Upload Video, add the prepared source clip. Wait for the upload to finish, then choose English under Target language.
Play the source once more before submitting. Check that it is the intended edit and that the last sentence is complete. Check the current input requirements and credit estimate shown for the selected workflow, especially before processing a longer clip.
For the practice script, the first result should let you judge all three sentences. For a longer tutorial, choose an excerpt that includes your hardest terminology. A simple greeting at the beginning provides little evidence about how the tool handles a later product explanation.
Set the translation controls for the clip
The current interface offers Fast - Lip Sync and Advanced - Lip Sync. Fast provides standard lip sync. Advanced is a paid option described in the interface as providing stronger lip-sync precision. Choose according to the requirements of the shot and the estimate shown, then judge the result on that shot. Review the translated words whichever mode you use.
The Speakers control offers Auto and counts from 1 to 5. Leave Auto for a straightforward first pass. For a clip with several distinct speakers, choose the matching count. Listen for speaker changes in the output as well as watching their faces.
Choose output options deliberately.
| Option | When to consider it | Review afterward |
|---|---|---|
| Generate captions | Viewers will read along or watch without sound | Names, numbers, line breaks, and timing |
| Dynamic Duration | The translated speech needs more or less room than the source | Sentence pacing, pauses, and timing against actions |
| Remove background music | Music in the source competes with the spoken voice | Whether the speech is clearer and the remaining sound feels appropriate |
| Speech Enhancement | The source voice is quiet or noisy | Whether quiet syllables and names remain understandable |
Dynamic Duration lets translated speech timing stretch or shorten. If a presenter points to an object or presses a button, check the relationship between that action and the English sentence. Natural speech matters, but so does hearing the instruction at the right moment.
Generate the excerpt after these settings are ready. You can return to My Creations to find the task and inspect the output.
Review the English version in three passes
Start with meaning. Compare the source, your reference translation, and the generated English. For the practice script, check “blue,” “five seconds,” and “before opening it” explicitly. A reviewer who understands both languages should assess any translation that you cannot confidently check yourself.
Next, listen without studying the picture. Does the English sound complete? Are names intelligible? Is the pace comfortable? A voice that rushes through the final words may need a different timing choice or a shorter, manually prepared version of the line.
Finally, watch and listen together. Look at the starts and ends of sentences, the pauses, and any speaker changes. Examine head turns and moments when a hand or microphone crosses the mouth. Also watch the overall face and the objects near it, so you can catch a visual problem that a close focus on the lips might miss.
If captions are enabled and the result includes a Caption file / SRT download, save that file and open it with the finished video in your editor or a player that supports subtitle files. Check the text against the spoken words; do not assume it is burned into the video. Captions and narration must agree about the details that matter. Read them on a phone-sized view too, particularly if you plan to publish a vertical crop later.
Fix the cause of a weak result
| What you notice | What to check next |
|---|---|
| A name or instruction is wrong | Compare the source pronunciation with the reference text. Use the reviewed-recording route below when the English wording must be exact. |
| The English is rushed | Review Dynamic Duration and the available space around the sentence. If you prepare the dub manually, shorten the phrasing while preserving its meaning. |
| A person appears to have the wrong voice | Verify the source's speaker changes and the selected Speakers count. Try a cleaner excerpt with less overlapping speech. |
| Music competes with the dialogue | Consider Remove background music. Keep the original music separately if you will rebuild the mix in an editor. |
| Mouth movement struggles during a turn | Compare a more front-facing excerpt from the same footage. Review the result at normal speed before deciding whether that shot is usable. |
| English captions sit over old subtitles | Return to a clean export of the source project when available, then regenerate from that version. |
Change one input or setting at a time. Otherwise, an improved second version may leave you unsure whether the source edit, audio processing, or timing change helped.
When the English wording must be exact
The translation form does not provide an editable translated script. Have the English wording reviewed, then record it or create it with text to speech. For this redubbing route, pair the finished English audio with your original footage in Video Lip Sync. Starting from a finished recording gives you control of the exact words in that separate English version.
Prepare the recording with the footage in view, leaving enough time for the English line and its pauses. Check that instructions still coincide with on-screen actions. The existing tutorial explains how to prepare these inputs and handle mismatched lengths. Video lip sync guide.
Add English captions separately in your video editor using the reviewed English script. Check their text and timing against the actual final English recording, since the spoken words and pauses may differ from the written draft. The Generate captions option belongs to the translation workflow. Captions from an earlier translation may differ from your new recording. Use a clean source export if the original footage contains burned-in subtitles.
Check the delivered file before sharing
Watch the entire finished version, including its first and last seconds. Check each item on the file you will actually upload or hand over.
- The English preserves names, numbers, units, and instructions from the source.
- Every sentence is audible and complete, with understandable pronunciation and comfortable pacing.
- The correct person speaks at each turn, and their mouth follows the English audio.
- Instructions line up with actions, pointing, and product demonstrations.
- Captions agree with the final recording, remain readable at the intended viewing size, and avoid old burned-in subtitles.
- The final frame and final word both finish cleanly.
Choose a representative source excerpt and open AI Video Translation to make the English version with the settings you have reviewed.