Upload a video and a new audio track, and the AI redraws the mouth so the lips match — for dubbing, dialogue fixes or music videos. 7 lip-sync engines, clips up to 10 minutes, free plan.
Talking head
Podcast host
He leans into the microphone and says to the camera: "Welcome back to the show, today we're talking about sleep." Clear speech, natural expressions
Quick answer
AI lip sync redraws a speaker's mouth and lower face so the lips match a new audio track. Dubbing swaps the language, talking photos animate a still — lip sync is the step that makes either believable. In Kenerate's Avatar Talk (video mode) you upload a video and an audio file and pick one of 7 engines: VEED Lipsync 2 (up to 5 min), Sync Lipsync 3 and LatentSync (5 min), Sync React-1 (15 s, with emotion control) and InfiniteTalk (up to 10 min, including a two-speaker version). It works on the free plan. The demo below: an English podcast clip, dubbed to Spanish, then lip-synced.
Updated 27 September 2026 · specs read from Kenerate's code
Real demo
Switch versions and turn the sound on. The Spanish voice alone doesn't match the lips — lip sync fixes that.
All three are unedited outputs of Kenerate's tools.
Lip sync your videoShort answers
AI that redraws a speaker's mouth and lower face so the lips match a new audio track — for dubbing, dialogue fixes or music.
Dubbing replaces the voice with another language; lip sync changes the mouth to match audio. The demo shows both: the Spanish dub alone, then the dub with lip sync.
Originally a one-take music video where a crowd lip syncs to a song; today it also means AI dubbing with the lips re-synced.
Yes — it follows the sounds in the audio, so any spoken language works.
Real clips
Real 5-second Wan 3 clips with speech — the kind of footage you'd lip sync to a new line, a new language or a song.
Real Wan 3 clips (5 s, 720p) from Seedream 5.0 Pro stills. Open one to watch with sound and copy both prompts.
How it works
Keep the performance; replace the words.
01
Seedream 5.0 Pro at 2K — or upload your own photo.
02
He leans into the microphone and says to the camera: "Welcome back to the show, today we're talking about sleep." Clear speech, natural expressions
03
Wan 3 · 5 s · 720p · with sound. The clip on the left is the real result.
Tips
From VEED's, Kling's and Sync Labs' guides and our tests.
Step by step
Engines
From Avatar Talk's video-mode engine list.
VEED
VEED Lipsync 2Default
Everyday dubbing and fixes
Sync Labs
Sync Lipsync 3
Control when audio and video lengths differ
Sync Labs
Sync React-1
Emotion-controlled performances
ByteDance
LatentSync
Open diffusion lip sync
WaveSpeed
InfiniteTalk Fast · Video
Long talks and lessons
WaveSpeed
InfiniteTalk Duo · Video2 speakers
Two-person conversations
The example clips were made with Wan 3; the demo above ran on ElevenLabs dubbing and VEED Lipsync 2.
Every clip on this page started as one still image. Make yours, then animate it.
In depth
What the models can do, how to prompt them, and what isn't allowed.
The model reads the audio — ByteDance's LatentSync uses Whisper audio features — and regenerates the mouth region frame by frame to match, keeping the rest of the face and the scene untouched.
For a translated video, dub the audio first (Kenerate's dubbing tool does 33 languages), then lip sync the video to the new track. The demo on this page is exactly that pipeline.
FAQ
7 questions
AI that redraws a speaker's mouth and lower face so the lips match a new audio track — for dubbing, dialogue fixes or music.
Dubbing replaces the voice with another language; lip sync changes the mouth to match audio. The demo shows both: the Spanish dub alone, then the dub with lip sync.
Originally a one-take music video where a crowd lip syncs to a song; today it also means AI dubbing with the lips re-synced.
Yes — it follows the sounds in the audio, so any spoken language works.
Up to 10 minutes on InfiniteTalk, 5 minutes on VEED Lipsync 2, Sync Lipsync 3 and LatentSync, 15 seconds on React-1.
Yes — that's a talking photo. Use Avatar Talk in photo mode (19 engines).
It works on the free plan with sign-up credits; longer videos use one-time credit packs that never expire.
Explore
Generators, tools, models and guides — one studio, one credit balance.
Written by the Kenerate AI team
We build and run Kenerate's video studio. Every clip here is a real, unedited Wan 3 output (5 s, 720p) and every start image a real Seedream 5.0 Pro output; model specs and limits were read from the app's code; facts about other tools link to their own pages. Last updated 27 September 2026. Corrections: hello@kenerateai.com.
Ready when you are
One studio for images and video.
