Sync Offset Calculator

Audio-video sync offset is the signed time difference between a sound and the picture it belongs to, and it is expressed in three interchangeable units. One frame lasts 1000 divided by the frame rate in milliseconds, so a frame is 41.667 ms at 24 fps, 40 ms at 25 fps, 33.367 ms at 29.97 fps and 16.667 ms at 60 fps. Multiplying an offset in seconds by the audio sample rate gives it in samples, the finest unit available: at 48 kHz one sample is 20.83 microseconds. The net error in a real signal chain is the video path delay minus the audio path delay plus any offset already baked into the file, which is why frame synchronizers, scalers and codecs each have to be accounted for separately. Human tolerance for this error is markedly asymmetric, because in the physical world sound always arrives after light. ITU-R BT.1359 gives detectability thresholds of about plus 45 ms for audio leading video and minus 125 ms for audio lagging, with acceptability limits wider still, and EBU R37 recommends production stay within roughly plus 40 to minus 60 ms. Sound travelling 15 m in air already lags by about 44 ms, which is why a distant loudspeaker seems natural while early audio does not. The important distinction in diagnosis is between a fixed offset, which a single slip corrects, and drift, where the error grows because audio and video clocks or nominal rates disagree. Sample rate mismatch, variable frame rate capture from phones and screen recorders, and 3:2 pulldown between 24 fps and 29.97 fps all produce drift that no constant shift can fix; those need resampling or conforming instead.

The problem is as old as synchronized sound. Early sound-on-disc systems such as Warner Brothers Vitaphone, introduced with Don Juan in 1926, mechanically linked a 33 1/3 rpm disc to the projector, and any slip or repaired splice destroyed sync permanently. Sound-on-film optical tracks, developed through the 1920s in the Movietone and RCA Photophone systems, solved that by printing the audio on the same strip of film, with a standard offset between the sound head and the picture gate. The clapperboard, in use from the same era, provided the sharp simultaneous flash and crack that let editors align separately recorded material by hand. Electronic production reintroduced the problem in new form. SMPTE time and control code, first standardized in 1969 and now maintained as SMPTE ST 12, gave every frame an address so that separate audio and video machines could be locked together, and genlock distribution of blackburst or tri-level reference kept device clocks aligned in studios. Digital processing then added variable, path-dependent latency: frame stores, standards converters, MPEG encoders and consumer displays each hold picture for tens of milliseconds while audio takes a shorter route. The EBU issued Recommendation R37 to bound the resulting error in broadcast chains, the ITU published BT.1359 in 1998 on the relative timing of sound and vision, and streaming protocols such as HLS and MPEG-DASH now rely on presentation timestamps to re-align the two streams at the player.