Multicam tracks are aligned, but playback not in sync

I ran the Gling fcpxml (v1.9) and the FCP-generated fcpxml (v1.13) through Claude and described the audio sync problem with instructions to inspect them for meaningful differences. Of course Claude can be wrong and/or BS us, but here's the response:

"After carefully analyzing both FCPXML files, I've identified the root cause of the 18-frame audio sync issue. The problem lies in how the 3rd party tool calculated the audio offset and timeline synchronization.

The main issues are:

Missing Gap Compensation: The working Info.fcpxml uses a gap element (<gap name="Gap" offset="0s" start="86400314/24000s" duration="18018/24000s"/>) to properly align the audio timeline before the audio clip starts. The broken Moose 29.fcpxml completely omits this gap.

Incorrect Offset Calculation:

• • Info.fcpxml: Uses 36036/48000s = 0.75075s for audio offset
  • Moose 29.fcpxml: Uses 17017/24000s = 0.709041667s for audio offset
  • The difference combined with the missing gap creates a timing shortfall of approximately 0.79 seconds, which equals about 19 frames at 24fps - very close to your reported 18 frames.

Sample Rate Inconsistency: The Moose 29.fcpxml uses a 24000Hz denominator for the audio offset timing, but the audio is actually 48kHz. This fundamental mismatch in how time is calculated for audio vs video elements causes the synchronization to break.

The fix would be for the 3rd party tool to:

• Include the proper gap element before the audio clip
• Calculate the audio offset using the correct sample rate (48kHz)
• Properly account for the timeline synchronization needed between video and audio-only angles

This explains why the audio waveforms appear incorrectly positioned and why the playhead doesn't snap properly - the timeline calculations are fundamentally flawed in the generated FCPXML."

I then asked Claude to double-check the audio asset definitions between the two fcpxmls in case they are referencing the audio differently.

"The audio asset definitions are functionally identical in terms of timing - both start at 3600 seconds and have the same duration (1998848/48000s). The synchronization issue is NOT caused by differences in the asset definitions themselves.

This confirms that the problem lies entirely in how the multicam angles are constructed and synchronized, specifically:

1. The missing gap element in the Moose 29.fcpxml
2. The incorrect offset calculation (17017/24000s vs 36036/48000s)
3. The improper timeline alignment between video and audio-only angles

The asset definitions are correct - it's the multicam implementation that's broken in the 3rd party generated file."

In the Angle editor, select Angle 2 so it's highlighted, then click its popup and choose "Sync selection to monitoring angle." If it's a long clip, does the sync start out okay and then drift more and more out of sync? That would point to a sample rate problem.

I brought the assets in and manually created a multicam, then exported that out as an fcpxml and compared it to the Gling fcpxml. The audio asset durations and start times differ between the two fcpxmls.

In FCP, the imported Gling multicam audio is 18 frames off. The audio angle also behaves oddly where the playhead doesn't snap to the beginning, like it's confused about where the clip actually starts. There appears to be a phantom snapping point at about 18 frames into the gap area before the audio clip starts. Also, the audio angle waveform is off by around 18 frames, meaning the visual display doesn't match the actual audio.

I'm thinking there's a miscalculation of the audio asset, but I didn't have time to dig into it further. Maybe tomorrow.

In the dropdown menu to the right of "Angle 2" in the Angle Editor, if you select Sync to Monitoring Angle, does the audio move slightly and play back in sync with Angle 1?

I'm experiencing this issue as well. I've made a Loom that shows the problem in action. The imported xml file is visually showing the files in sync, but audibly they are not. Even when I replace the audio in the angle editor and sync to monitoring angle again, it is still out of sync. Only when I delete the angle entirely, add a new angle, and sync the audio to monitoring angle does it finally sync. https://www.loom.com/share/17689855873c4fc29b9875672e1bab11?sid=83887c56-a74d-41c4-93eb-67e708650163

I think the multicam xml is broken coming out of gling. I'd send the error screen to their devs and the mulitcam xml. You could also post a link to the xml here and we could have a look.

Yes @terryb, the beta did fix my sync issues.

The screenshot seems perfectly aligned, and I cannot see anything untoward in the snippet of the xml file.

How far out of sync do you hear it when playing back?

And are you just checking there in the Angle Editor, or have you added this multicam to a project?

I've been talking with their support team for a couple of weeks on this, which lead me to here. I think they're trying to figure out of this is a Gling issue or a FCP issue. Their CTO is the author of this thread. Here's a link to the XML file and the original media if that happens to be helpful too. https://drive.google.com/drive/folders/1vwgUzn_yoBLGP0gEPodpfwmzUfV-JsWZ?usp=sharing

@yonatan166, I received notification via email that you have a beta update of Gling's fcpxml export. What ended up being the fix?

@reynoldsm89 - with this beta are you able to successfully import a properly sync'd multicam fcpxml?

“In both the out of sync is in a couple of seconds”

This happens when the audio sample rates between clips differ.