Solve the Mystery! Bizarre audio drift issue on a feature doc.

[David St. Juskow]

I've been cutting professionally with FCP for a long time, but have never seen this. Here's the setup:

Feature documentary started years ago on FCP 6. Because of the multiple parties involved, we're still on 6.0.5- no problems in the 3 years cutting. Finally have pic lock. I export an OMF of our 82 minute sequence for the sound mixer. We're at 44k, 24-bit audio, drop-frame, 29.97. The footage is a combination of 720/60 HD (EX-1) and up-scaled DV (PD-150.) Plus a score given to me as 48k, 16 bit wav files. He's on ProTools- the music and effects sync up perfectly to the reference quicktime (with TC burnt in) but some (not all) of the SOT's are slightly out of sync. We can't figure that one out, but time is short, so he slips all the weird SOT's into sync manually. Not the best solution, I know, but we are short on time.

He sends me a temp / test mix embedded in the reference quicktime. I extract the audio from that reference video and convert it into a 48k .aif file. When I cut it into the sequence, it starts fine lined up to the 2-pop, but is 4 seconds and 2 frames out of sync by the end of the film. Huh? I try converting into different formats, nothing helps- the sound track is 4 seconds longer somehow. However, when I import the reference video on it's own, THAT remains in sync when I lay it over my sequence- both video and audio (even though that's the same audio track I created the .aif file from- the one that drifts!) When I cut JUST the audio track from that quicktime (which has been compressed into an mp4) it is perfectly in sync.

Again, time is short but it seems like there must be some odd setting that is off. When he sends me the final mix as both a stereo AIF file and as 6 surround mono tracks, everything is drifting again by 4 seconds, no matter what I try.

We're about to give up when we try importing the same exact files into the same exact project, but on someone's laptop. And, guess what- that same audio file cut into the same sequence is now in sync. On my quad processor tower, out of sync. On his, in sync.

Now, yes, these are different machines- I'm on a G5 still, he's on an intel laptop, much newer. But still- why would that matter even in the slightest? Neither of us have any external device connected that might affect sync- like a audio slave clock, digital deck, etc. Same software, same project, even same external drive- but on one computer, it works, and the other is off.

Any ideas? We thought it might be a drop-frame / non-drop thing at first, or some weird mismatch... but it all seems irrelevant when we have 2 contradictory machines. Web searches so far reveal nothing.

thanks!

[Tim Dashwood]

It sounds to me like your sound mixer's framerate was set to 30 instead of 29.97 when he started the mix. Maybe there was a bug in OMF export from FCPv6 that I don't know about, but way back when on the Avids and ProTools systems we used to have to flip the switch on the Digidesign audio interface to 1x or 0.999x depending on whether we were working on a feature film for print (sound speed 24fps) or a NTSC project.

The film is exactly 82 minutes long? Divide the exact length of the film minus (or plus) the drift amount in seconds by the expected total run time in seconds and see if the the result is .999 or 1.001 give or take a few ten thousandths. If it is then likely the problem is related to the mix being written out at 30fps vs NTSC 29.97fps.
You can either inform your mixer and have him export it again or you can adjust the rate yourself from 30fps to 29.97fps in Soundtrack Pro (47.952Khz) and re-render the aiff tracks back to 48Khz.


I did the math to check:

If your film is supposed to be exactly 82 minutes long you would get the runtime in seconds by multiplying 82x60=4920

Then you would subtract the 4 second drift at the end of the film from 4920 to get 4916.

4916/4920 = 0.999....

This is because 29.97/30 = 0.999

If the soundtrack is longer than the picture track then 4924/4920 = 1.001

30/29.97 = 1.001001001001001....

I have no good explanation for why you got a different result on a different system. Maybe the laptop had a newer version of FCP and there is a bug in v6.0.4 on your G5?