Check synchronization at several points before moving the audio. If the gap grows as the recording plays, a single offset cannot repair it. Investigate source timestamps, editor interpretation, speed changes and the audio recording clock. Variable frame rate is one possible factor, not a diagnosis established by “recorded on a phone.”
Record the shape of the error
Choose a visible and audible event near the start, middle and end: a spoken consonant, clap or object contact. Compare the event in the untouched source player, the editing timeline and a short export.
| Pattern | What to investigate |
|---|---|
| Same delay everywhere | Initial alignment or playback offset |
| Delay grows continuously | Timing interpretation, speed or recording clocks |
| Delay jumps at one point | A cut, dropped section or damaged recording |
| Source is synchronized, export is not | Editing and export path |
Use meaningful events rather than trying to infer a precise lip position from a blurred frame. If the source contains no audible matching event, state that synchronization cannot be checked in that way.
Inspect timestamps, not only the fps label
An average frame-rate label can hide changing intervals between frames. ffprobe's documentation describes stream, frame and packet inspection. If that tool is already available, an editor or technician can examine the actual timing information; there is no need to upload private vows to a public diagnostic site.
Also check whether the clip was slowed down, trimmed or reinterpreted earlier. A source with variable timestamps can play correctly in a compatible pipeline. Do not convert every phone recording merely because its report says VFR.
If a separate audio recorder was used, its timing can differ from the camera's. A video-only conversion will not necessarily repair two independently recorded clocks.
Make a working copy when justified
HandBrake's frame-rate reference distinguishes VFR, constant frame rate and peak-limited output. CFR is a conversion choice, not just a new label on the original file.
Where an editing workflow requires CFR, make a separately named working copy that keeps the intended real-time duration. Record the chosen rate and compare its audio-event alignment with the source. Do not reinterpret the same frames at another rate if the goal is normal-speed dialogue; that can change duration rather than preserve timing.
A conversion may repeat or omit visual frames to meet a regular output clock. Inspect fast hand movements and keep the untouched source as the evidence reference.
A fictional long-phone-take diagnosis
A planned interview contains an audible hand clap near the beginning and a spoken response near the end. The untouched phone file plays in sync, while the editor's version drifts. In this fictional case, the next step is to test that editor's ingestion and a suitable working copy, not to slide the final audio until the ending fits.
If the independent export still drifts after conversion, the conversion is not accepted as a cure. Return to the measured points and investigate another timing cause. This is a diagnostic plan, not a successful repair claim.
Recheck before using the clip as a reference
Watch the same three events in the new file and verify that the complete action and sound tails remain. For dance references, preserve the source's real-time movement before comparing generated choreography or syncing music.
Use reviewed sources when planning shots in Imagild Film Studio. Timing preparation belongs in an external media workflow. A constant-rate file opening successfully does not prove its audio is synchronized or its actions match the source.