AI Lip Sync: Match a Portrait to Your Own Audio
AI lip sync on MarketVideo takes one portrait and an audio clip you recorded, up to 20 seconds long, and returns a video where the mouth follows your words.
Lip-sync a portraitReady in minutes ยท Free templates every day
It runs on LTX-2.3 LipSync, a model on the Create video page. There is no built-in voice and no typed script: the sound always comes from a file you bring, such as a greeting, a product line or a short toast recorded on your phone.
Record the audio first
Start with the sound, because it decides everything else. Keep the recording under 20 seconds, speak clearly and trim long silences at the start and end. A clean voice track is easier to follow than one buried under music or street noise. If the message is longer, split it into two or three clips and run each one separately.
Pick a portrait the model can read
A front-facing photo with the whole mouth, chin and cheeks in view works best. Avoid hands near the face, masks, heavy shadow across one side and profile shots. If your only selfie is dim or cluttered, one of the portrait templates below can turn it into a clean close-up first, and that still becomes the face you lip-sync.
Run it on Create video
The run itself happens on Create video: pick LTX-2.3 LipSync, attach the portrait and the audio file, then start it. The star cost is shown before you confirm. A few minutes later, sometimes longer when the queue is busy, the clip appears in your history and can be saved to your device like any other video.
What it will not do
There is no voice generator here: typing a script will not produce speech, and the model will not change how your voice sounds. Each run covers one portrait and one clip of up to 20 seconds. Results depend on the photo and the recording, so mouth shapes are not guaranteed to be perfect; a second try with cleaner audio or a sharper photo is the usual fix.
Ready-made templates
FAQ
Can I use AI lip sync for free?
Every new account gets 5 stars after the email is confirmed, and a lip-sync run is paid from that balance if its cost fits. The Free today label appears on selected catalog templates, so check the cost shown for LipSync before running. Plans and star packs for more runs are on the Pricing page.
How long can the audio clip be?
Up to 20 seconds per run. That is enough for a greeting, a short intro or one punchline. For a longer message, cut the recording into parts, generate a clip for each and join them in any video editor you already use.
Can it make the portrait say text I type?
No. The model listens to a recording, not to written words, and MarketVideo has no built-in voices. Record the line on your phone or computer, save it as an audio file and upload it together with the photo.
Can I lip sync someone else's photo?
Only if that person has agreed to it. Your own face, a partner or a friend who said yes are fine; celebrities, public figures and strangers are not, and neither is putting words in someone's mouth to pass as them. Photos of children belong in a family context only.