FCPX tiny sync drift with very long shots

Hi everyone,
I'm a teacher and I'm recording my lessons so my students can watch them later.

I have my iPhone 6S that is recording video, a Rode smartlav+ mic to record my voice, and I record my Macbook screen with Openbroadcaster Studio.

Everything is set on 1080p 30fps.

But the videos are VERY LONG : they last 3-5 hours straight, in one shot.

I have a sync issue : it seems like the screen recording and the audio mic are drifting from the iPhone main video footage, by 3 frames (100ms) per hour exactly.

It is very annoying to split the files afterward to sync it back, because 100ms per hour means it's impossible to adjust speed directly in Final Cut, because the software doesn't take "99.99%" as a valid speed correction.

Is there a way to avoid this sync drift,


I imagine a solution where I can put a marker in sync on a first clapper at the beginning of the recording, and another marker in sync at the end of the recording, and Final Cut will match the beginning markers and the end markers.

Is there anyone with similar problems, and solutions ?

Thank you. Sorry for my bad english, I'm not a native speaker

That is a problem with digital, you have to understand that video and audio run on clocks, 30 frames per second for video and 48 000 cycles for audio, also known as 48kHz. As they are different, there is a drift over a long period.

I may be wrong, but I thought that 30 frames is actually 29,97 frames per second. This may contribute to the drift, it is a well known problem, and so far I have not heard of any way to solve it, other than using a different recorder, and adding the audio to the video after, using the on camera audio as a sync.

Maybe changing the frame rate to something like 25 frames, which is a whole round number instead of 30 frames,[29,97] might help the drift.

Try separating the audio and video, and adding a 1 frame gap clip and connect the audio to the gap clip at the head of the track, so the audio and video are not linked, they are separate assets.

Try stopping and doing shorter clips, maybe blade the clips in the timeline and drag the audio back 1 frame to regain sync. After all, it was never really designed for 5 hours of a single take, that is digital video that is... :blush:

MsJustine wrote: ...I thought that 30 frames is actually 29,97 frames per second. This may contribute to the drift, it is a well known problem, and so far I have not heard of any way to solve it, other than using a different recorder, and adding the audio to the video after, using the on camera audio as a sync....

This is commonly stated but I don't think that is the source of the problem. Material recorded at 29.97 is also played back at 29.97, so the elapsed "wall clock" time should be the same. Likewise audio sampled at different rates is played back in the same elapsed time as it was recorded.

However -- each recording device is ultimately driven by an oscillator, just like a quartz wristwatch. Oscillators don't have perfect frequency stability, which is why two quartz watches will gradually drift apart. The oscillators inside consumer electronic devices are like an expensive watch.

One manufacturer of a quartz watch specified their precision as +/- 15 sec per month. That means two different watches could drift 30 sec apart in one month, or 1 sec per day. That was a spec. Actual surveys show some quartz watches drift by 2 sec per day, so two watches with opposite drift could move apart at 4 sec per day.

The OP case was drifting 100 milliseconds per hour, which is very roughly 2.4 sec per day -- about equal to some typical inexpensive quartz watches.

The solution is for camera and audio recorders to have a low drift, precise timebase. This could be achieved by using a temperature-compensated crystal oscillator, and/or an oscillator which is individually calibrated at manufacturing time, or continuous adjustment via some known time standard such as a cell phone which gets accurate time from the cellular network or GPS.

In this case one of the recording devices was a cell phone, but I doubt the video recording rate was was being fine tuned by cellular or GPS timing data.

Thank you for your answers

Joema is right : it isn't a 29.97 / 30 fps problem, because the difference can be corrected by playing the video at exactly 99.90% speed, and I've done it before.

The drift I experience is a lot smaller than that, more like 0.001%

And indeed, I think that it's precisely the difference between the CPU quartz of each device.

So... no easy solution. "Consumer-lever digital video equipement was not designed to take two hours long shots, period."

But I discovered that in Final Cut I can use the "two claps" method to precisely and quickly sync all drifts : by dragging the speed bar with the mouse cursor, I can adjust the length of a shot with precision, smaller than 1 frame. So that's what I'm doing now, because it's quick, precise, and good enough overall (no re-encoding).