What this feature does, in short
Lip-Sync makes the speaker’s mouth match the translated speech, so your video looks natural, authentic, and ready to share. When Lip-Sync is enabled, CHAMELAION aligns mouth movements to the newly generated audio in your cloned voice. You focus on the message, we handle the visuals 😉.
How it works, in simple terms
We track the speaker’s face, locate the mouth, and use our AI model to adjust lip movements to match the words in the selected language. Behind the scenes, we have already transcribed the source, translated it, and synthesised the new audio in your cloned voice. If Lip-Sync is on, we then align the lips to that audio.
Token use and plans
Lip-Sync is an optional add-on that uses tokens per processed second, per target language, in addition to base Video Translation. You can take a look at our Pricing site for a detailed breakdown. Just do not worry, we don't round up like a parking lot😉.
To estimate the right plan for your volume, open the Plans page and try the calculator. It considers whether Lip-Sync is on or off and suggests a plan that best fits your needs 👍.
How to turn it on
Before translation
After you upload your video, confirm your source language and select your target language or languages.
Switch on the Lip-Sync toggle before you click Translate. You can leave it off and add it later; however, we recommend leaving it off for the first translation. Find out why down below 😉.
In the Dubbing Studio
When you click Generate, a small window lets you re-select options for the render. Here, you can choose whether to include Background Sounds, and you can also switch Lip-Sync on or off for that render.
Recommended workflow to save tokens
For the first pass, translate into one language with Lip-Sync off.
Open the result in the Dubbing Studio, play the source and the target, and fix any transcript issues first, for example, classic misreads like “Haus” or “Maus” that could turn into “mouse” instead of “house.”
Tweak timing and pacing if needed, then click Generate.
Only now, enable Lip-Sync and generate again.
This avoids spending tokens on lip alignment for lines you would change anyway. If the transcript is wrong, the translation is wrong, then the cloned voice will say the wrong word, and Lip-Sync will faithfully mimic that wrong word's mouth shape. Fix this first, and activate Lip-Sync through the Dubbing Studio 😄.
Tips for best results
Record the speaker facing the camera; a frontal view yields the most accurate lip movements.
Done!
That is it. Decide when to add Lip-Sync, fix text and timing first, then sync for a flawless, real-looking result.