Loading page
About this tool
Calculate audio lead or lag in milliseconds from a flash or clap recording.
An audio-video sync meter converts a clap or camera-flash slate into a signed lip-sync error in milliseconds: offset = audio spike time − (flash frame − 1) ÷ frame rate, where a positive result means the sound arrives after the picture. It then grades that offset against the published limits — ATSC IS-191 allows 15 ms of audio lead and 45 ms of lag, EBU R37 allows 40 ms and 60 ms, and ITU-R BT.1359-1 puts the detectability threshold at 45 ms lead or 125 ms lag. It is for editors and streamers who can see something is off but need the number before they can fix it.
Open Audio-Video Sync Meter on AltFTool — it loads instantly in your browser.
Enter the Frame rate (fps), the Frame of the flash / clap, the Audio spike time (seconds or HH:MM:SS.mmm), and whether the first frame is numbered 1 or 0.
Press "Add to average" to fold in a second or third slate, since reading one flash is only accurate to about half a frame.
Read the Audio offset in ms, the Offset in frames and the pass/fail row for ATSC IS-191, EBU R37 and ITU-R BT.1359-1, then press Copy report.
The frame of the flash and the timestamp of the audio spike are all you need — no plugin, no upload.
The verdict cites ATSC, EBU and ITU thresholds rather than a made-up 'good enough' figure.
Reading one flash is accurate to about half a frame; averaging two or three claps removes most of that error.
ITU-R BT.1359-1 puts detectability at 45 ms when audio leads the picture and 125 ms when it lags, and the point where viewers actively object at 90 ms lead and 185 ms lag. Ears are far less forgiving of sound arriving early because that never happens in nature — distant sound always arrives after the sight of the event.
ATSC IS-191 is the tightest in common use: at the point of emission, audio must not lead video by more than 15 ms and must not lag by more than 45 ms. EBU R37 covers programme production with 40 ms lead and 60 ms lag. At 25 fps, 45 ms is a little over one frame.
Feed the same file twice and offset one input's timestamps: ffmpeg -i input.mp4 -itsoffset -0.040 -i input.mp4 -map 0:v -map 1:a -c copy synced.mp4 pulls the audio 40 ms earlier while copying both streams untouched. The sign is inverted from the measurement: audio that is late needs a negative offset.
A camera flash lands inside a single frame, so the picture side of the measurement is unambiguous, while a clap has a visible hand movement spread over several frames. Audio is the opposite — a clap gives a sharp transient in the waveform. Using a flash for the picture and its own click or a clapperboard for the sound gives the cleanest pair of readings.