Nothing breaks a video faster than sound that does not match the picture. The words arrive a fraction of a second late, the lips move first, and suddenly a polished recording feels like a badly dubbed movie — viewers notice, even when they cannot name what is wrong. The usual culprits are the recording chain itself: screen recorders, capture cards and Bluetooth audio each add their own latency, because the picture takes one path through the hardware and the sound takes another, and they arrive a few hundred milliseconds apart. Our free Audio Sync Fix corrects exactly that: slide the audio offset until it locks in, A/B test original against fixed live, then export a corrected video — free, private, with no watermark, and your video never leaves your device. This guide explains how to find the offset, which direction to push it, and the one case this tool honestly cannot fix.
Step one: load the drifting clip. Drop in the video whose audio lags or leads — MP4 or WebM. The picture plays on screen while a separate audio track plays with your offset applied, so you hear exactly what the export will sound like. Step two: dial in the offset. Drag the slider or nudge in 50 ms steps for fine tuning until the lips lock with the words. The sign rule is the whole trick: sound after the lips move means the audio is late, so go negative (audio plays earlier); sound before the lips move means the audio is early, so go positive (audio plays later). Flip between Original and Fixed with one tap while it plays — the difference is usually obvious on plosives like b, p and m, and on claps. Step three: export the corrected file. Click export and the video plays through in real time while the shifted audio is captured alongside the picture, producing a corrected WebM you can upload anywhere. Keep this tab focused until it finishes: export takes as long as the clip, because it is a real-time re-encode.
Human ears notice sync errors above roughly 40 milliseconds, which is both good news and bad: good, because you do not need frame-perfect precision, and bad, because a 100 ms error that looks small on a slider feels obvious on a face. The tool page's test trick is the fastest way to settle an offset: pause on a frame where the mouth is clearly open on a "b" or "p" sound, then flip between Original and Fixed. The wrong one will feel like a badly dubbed movie; the right one snaps into place. Test on the hardest passage, not the easiest — a talking head in a quiet room hides errors that a fast argument or a clap reveals instantly. And trust the A/B flip over the number: the slider tells you the offset in milliseconds, but your eyes and ears tell you whether the lips match, and the flip lets you compare the two states back to back without re-remembering what the original sounded like. If two nearby offsets both look plausible, pick the one that feels right on plosives and claps, since those sharp sounds expose timing errors that vowels hide.
Most lip-sync problems are a constant offset — the same delay from start to finish — because the delay comes from the recording chain, and that delay does not change during the take. Shift the audio by minus the delay and every lip matches again, which is the case this tool handles, and it covers the large majority of "my recording is out of sync" complaints. The harder case is drift: in sync at the start, noticeably off ten minutes later. That happens when the audio and video clocks run at slightly different rates — say the audio was captured at 44.1 kHz but stamped as 48 kHz — so the gap grows linearly with time. No single offset can fix drift: correct the start and the end is still off, correct the end and the start breaks. Drift needs re-time-stretching the audio, which is what desktop tools sell and what VLC deliberately does not do, since VLC only shifts playback and never the file. If your video drifts, the honest fix is to note the offset at the start and at the end, compute the rate difference, and re-encode with ffmpeg's atempo filter — a desktop job this browser tool does not attempt, and it says so plainly rather than pretending an offset would work.
Pushing the slider the wrong way. Sound after the lips means negative; sound before the lips means positive. If the fix makes it worse, you have the sign backwards — flip it and test again. Trying to fix drift with an offset. If the sync is fine at the start and off by the end, no offset will save it; that is a clock-rate problem, not a delay problem. Judging on an easy passage. Vowels hide timing errors; test on plosives, claps and fast speech, where the error shows. Letting the tab lose focus during export. Export is a real-time pass — a crashed or backgrounded tab loses the whole run, so keep it focused until it finishes. Exporting a very long clip in one pass. For clips over 15 minutes, consider splitting first: a failed pass near the end costs the whole clip's runtime. Re-encoding the export again and again. The video track is re-compressed once during export — a generation of quality loss, usually invisible at these bitrates — but every further re-encode adds another, so keep the first export as your master.
The honest limits are four. First, the tool fixes a constant offset only; drift from mismatched audio and video clocks needs time-stretching on the desktop. Second, the export is WebM, because browsers can only write WebM — a real-time re-encode that takes as long as the clip, with the video track re-compressed once and the audio captured straight from the shifted playback with no extra processing. Third, export requires the tab to stay focused, and for clips over 15 minutes a split-first workflow is safer than one long pass. Fourth, everything runs on your device, which is what keeps your video private, but also means export speed and reliability depend on your machine and browser rather than a server farm. Within those limits, the tool does one job completely: it takes a video whose sound arrives late or early and returns one whose lips match the words, with the fix confirmed by your own ears before you commit. The companion guides are our audio speed guide, for stretching or tightening a track to fit the cut once the sync is right, our background noise removal guide, for cleaning the track before you judge its timing, and our voice recording guide, for capturing clean audio in the first place.
Stretch or tighten a track to fit the cut once the sync is right.
Clean the track before you judge its timing — noise makes sync harder to hear.
Capture clean audio in the first place, so there is less to fix afterwards.