

Choose the video, say if the sound is early or late, and download the fixed file.
The sound is

Lip sync made from your audio starts in sync, so there is nothing to fix.
A sharp, front facing photo. Yours, or one you have the rights to.
The mouth, jaw and head are generated from the sound itself.
A 30 second clip is 60 credits, and the cost shows before you make it.


It runs in your browser. Nothing is uploaded.
Drop it on the tool or pick a file, up to 250 MB and 10 minutes.
Watch a clap or a hard consonant such as p or b. If you hear it before you see it, the sound is early; after, it is late.
The picture and sound are copied, not re encoded, so it takes seconds. If it is still off, try the next step.

Check the first and last few seconds before choosing. This tool fixes an offset: the whole soundtrack moves by the amount you pick.

FFmpeg reads the file twice and delays one copy with its itsoffset option: the sound when it is early, the picture when the sound is late. The two streams are then copied into one file without re encoding, so nothing loses quality.

Every setting is a URL parameter, so an assistant can send a link with the options already chosen.
Sound is 0.2 seconds early: https://percify.io/tool/fix-audio-video-sync?shift=0.2
Sound is 0.3 seconds late: https://percify.io/tool/fix-audio-video-sync?shift=-0.3
Every free tool and its parameters, as JSON: percify.io/tools.json. Platform facts checked on 17 September 2026 against each platform’s own help pages.

Choose a video, pick how much louder, and download it. Only the sound changes; the picture is copied as it is.
OpenChoose a video, pick a shape and a fit, and download the result. Your file never leaves your device.
OpenAdd two videos, pick a transition and its length, and download one MP4 with the picture and sound blending between them.
OpenIf the sound is early or late by the same amount all the way through, shift it: choose the video, pick Early or Late and the amount, and download. If it drifts further out over time, that is a frame rate problem instead.
Watch a clap, a door closing or a p, b or m sound. If you hear it before you see it, the sound is early. If you see it first, the sound is late.
No. The picture and sound are copied, not re encoded; only their timing changes.
Usually the audio was trimmed, sped up or replaced after rendering. Rendering again from the final audio is the cleanest fix; for a constant offset, this tool shifts it.
No. It is processed on your device with FFmpeg running in WebAssembly.
One photo and a script is all it takes.
Cancel anytime