Finds the offset for you
Two recordings of the same moment are matched by the shape of their sound, to well under a millisecond on clean audio. One video on its own is checked for claps and hits, and every answer comes with a confidence figure.
Free · runs in your browser
Drop one video to find and fix its audio delay, or drop a video with its separate recordings to line them up by their sound. Check it in slow motion, then save a new file with the picture untouched. Nothing is uploaded.
Two recordings of the same moment are matched by the shape of their sound, to well under a millisecond on clean audio. One video on its own is checked for claps and hits, and every answer comes with a confidence figure.
The picture is always copied as it is. When only the timing changes, the sound is copied too and the file is written in seconds. Drift fixes and mixes re-encode the sound alone.
Videos, recordings, songs and subtitles are processed in this browser tab and never uploaded. Free, with no account and no watermark.
How it works
Drop one video to fix its own sound, or a video together with a separate recording (or several camera clips) to line them up. MP4, MOV, MKV, WebM, WAV, MP3 and M4A all open.
Separate recordings are matched by their sound straight away. For a single video press Detect automatically, or Jump to the clap and pick the frame where the hands meet.
Play it back, slow it to ¼× around a clap, and nudge the sound Earlier or Later in 10 ms or 1 ms steps until lips and words meet.
Press Export video and download the new file. The picture is copied untouched, so nothing is lost and nothing leaves your device.
Four ways to sync
Under the hood
Sound and picture drift apart for two different reasons, and they need different fixes. A fixed offset means every sound arrives the same few milliseconds early or late: a camera that delays its audio, an export setting, a Bluetooth microphone. Drift means the gap grows as the video plays, because two clocks ran at slightly different speeds.
When a camera and a separate recorder both heard the room, their waveforms have the same shape at different times. sync.video turns each recording into an outline of its loud attacks, slides one past the other and measures how well they match at every position (cross-correlation, done with fast Fourier transforms in a background worker). The best position is then refined on the raw sound with a method called GCC-PHAT, which ignores differences in microphone colour and level and pins the match to a fraction of a millisecond. The same measurement repeated across the recording shows whether the two clocks drift apart, and by how many parts per million.
With only one file there is nothing to compare the sound against except the picture. Detect automatically scans a stretch of the video at low resolution, notes where the picture changes sharply (hands meeting, a slate closing, a cut) and where the sound has sharp attacks, and finds the delay that lines the two up best. It is honest about how sure it is. When it can’t tell, the clap method is quick: Jump to the clap, step frame by frame to the moment of contact and press This frame is the clap.
Video files store the time each frame and each slice of sound should play. Moving the sound only changes those times, so sync.video copies the compressed picture and sound into a new file with the sound’s times shifted, and records the shift in the file’s edit list. Nothing is decoded, nothing is lost and a long video is written in seconds. Stretching the sound to cure drift, mixing two recordings or fading music does need the sound to be decoded and encoded again; the picture is still copied.
Guides
Everything runs in your browser, so these answers are about offsets, drift, formats and devices rather than uploads.
Open the video here, find how far out the sound is (Detect automatically, the clap method, or by ear with Earlier and Later), check it in slow motion, then press Export video. The new file has the sound moved by that amount and the picture copied as it was.
Yes, in two ways. When you have two recordings of the same moment (a camera and a separate mic), their waveforms are compared and slid until they match, which is very reliable when both heard the same sounds. For a single video it looks for sharp sounds such as claps and matches them to sudden changes in the picture; that works well with a clap or a slate and less well with talking alone, and it tells you how sure it is.
Drop the video and the recording together. They are matched by their sound, so the camera’s own microphone needs to have heard the same thing (its scratch sound is enough). Choose Replace the camera sound and export: you get the camera’s picture with the clean recording in place.
Yes. Add up to eight files at once: cameras and recorders. Each is matched to the first, drift between their clocks is measured, and you can export synced copies or download an FCPXML or Premiere XML timeline with every file already in place.
Drift happens when the sound and picture were timed by clocks running at slightly different speeds, which is common with separate recorders, long phone recordings and screen captures with a variable frame rate. Set a sync point near the start and one near the end, or let it measure them, and the sound is stretched evenly to stay in step.
No. Every file is read and written by your own browser; nothing about it is sent to a server. The only things that ever travel are, in the sync player, small control and clock messages between the devices in a room.
The picture is never re-encoded: its compressed frames are copied into the new file. When you only move the sound, the sound is copied too. Correcting drift, mixing or fading means the sound is decoded and encoded again (AAC for MP4 and MOV), at a high bitrate.
Milliseconds. You can move the sound in 1 ms steps, and matching two recordings is precise to a fraction of a millisecond when both are clean. At 30 frames per second one frame is 33 ms, so sync to the frame is easy to reach.
MP4, MOV, MKV and WebM videos and WAV, MP3, M4A, AAC, FLAC and Ogg sound files open directly in modern browsers. Files the browser can’t read, such as AVI, WMV or camcorder MTS, are converted on your device first.
Yes, in a recent Safari or Chrome. Pick the clip from your photos or files, fix it, and use Share to save the result back to Photos. Long 4K files are easier on a computer.
Open the Audio delay test from the tools. It plays a flash and a beep you can line up by eye, or times a chirp with the microphone, and gives a number in milliseconds that you can type into the TV’s or app’s sync setting.
Yes, with the Sync player. Everyone opens their own copy of the same file, one device starts a room, the others join with its six-character code, and play, pause and seek stay in step. The video itself never leaves each device.
Yes. Beat sync finds the tempo and beats of the song, cuts your clips or photos on beats or bars, and exports a 9:16, 1:1 or 16:9 video. The BPM finder and Tap tempo tools help when you just need the number.
Then the file is fine and the TV (or its soundbar or Bluetooth speaker) is adding delay to the sound. Measure that delay with the Audio delay test and set the TV’s audio sync or lip-sync setting rather than changing the file.