Why does my audio drift only on long takes?

Because drift accumulates. A tiny clock difference is invisible over ten seconds and obvious over ten minutes.

Can I fix drift by stretching the audio?

Yes, a small speed conform is the standard fix, but confirm the exact ratio first, otherwise you replace one drift with another.

How do I check the real sample rate of a file?

Open it in the audio inspector here. It reads the WAV header locally and reports rate, depth, channels and duration.

6 min read

Audio not synced with video: the five causes, and how to tell them apart

Nudging the clip until the clap looks right only works for one of the five causes. The other four come back later in the timeline.

Sync problems all look the same on screen: lips move, sound arrives late. Under the hood there are five very different causes, and the fix for one makes the others worse. Diagnose before you drag.

1. Constant offset

The sound is late or early by the same amount from the first frame to the last. This is the friendly one. It comes from a delayed record start, a wrong timecode offset or a monitoring latency baked into a bounce. Slide the clip once and it stays in sync.

2. Progressive drift

The clap is perfect, and ten minutes later the sound is two frames out. That is a clock difference between the camera and the recorder, or between a 29.97 and 30 fps interpretation. Sliding the clip only moves the point where it happens to be right. You need to resample or conform the audio to the correct rate.

3. Variable frame rate footage

Phone and screen recordings often use VFR. Editors assume a constant rate, so the picture stretches unevenly against a perfectly steady audio track. Transcode the video to a constant frame rate before you cut, and the drift disappears.

4. Sample rate mismatch

A 44.1 kHz file interpreted as 48 kHz plays about 8.8 percent fast, which is roughly two and a half seconds of drift per thirty seconds. Check the real rate of the file rather than trusting the filename. The inspector on this site reads the header and shows you the true sample rate, bit depth and duration.

5. The wrong take

It happens more than anyone admits. Two takes with similar performance, one linked to the wrong sound file. If the waveform shape and the picture never match anywhere, stop nudging and check the scene and take metadata in the file.

A quick decision list

Same offset at head and tail: slide once, done.
Offset grows over time: clock or frame rate problem, conform the audio.
Screen or phone source: transcode to constant frame rate first.
Audio plays fast or slow and pitch sounds wrong: sample rate misread.
Nothing lines up anywhere: wrong file, check metadata.