A long phone MP4 starts with perfect lip sync, but by the end of a 20-minute recording the voice is visibly late. Sliding the audio can fix one end but ruins the other. The original looks normal on the phone. Is this the classic variable-frame-rate problem?
That pattern is much more like a timing mismatch than a simple offset. Phones often record variable frame rate, while an editing timeline may be working at a fixed cadence. The error can accumulate gradually.
Editing workflows
Inspect the source for VFR and test a short constant-frame-rate conversion. Check sync at the beginning, middle and end; do not repair gradual drift with dozens of manual audio cuts.
Also note the audio sample rate, but do not blame 44.1 versus 48 kHz automatically. A competent editor can resample audio; irregular video timestamps are often the bigger clue in phone recordings.
Action Video Lab Editorial Team
The file is VFR, so I finally have a concrete thing to test. I will make a CFR copy of a short section and verify end sync before touching the full edit.
Was the phone recording made with the normal camera app or a third-party app? I ask because some apps produce much more irregular frame pacing when the phone gets hot.
Normal camera app. MediaInfo reports VFR and the minimum/maximum frame rates are farther apart than I expected. The middle is already a few frames off, so it is definitely gradual drift.
Then I would leave the AAC alone for the first test. Make one CFR copy while preserving 48 kHz audio, cut the same three points, and compare. Changing audio and video at the same time makes it harder to know what fixed it.
CFR test stays aligned at the start, middle and end. That fixed the practical problem. I am keeping the phone original, but I will edit this long recording from the conformed copy.
Useful result. VFR is not inherently invalid, but this is exactly the case where normalizing timing is justified: there was a repeatable editing/export symptom, not just a warning label.
Codecs & export