PAL to NTSC Audio Sync + Drift

Hi All,

Using FCP 5.1.4.

I have a music concert shot in PAL. I have edited it all together no problem in a DV PAL 48k timeline.

The audio was also recorded separately (to DAT i believe) and that is what was used for mixing/overdubs/etc. So at this point we have a rough cut of PAL video (with matching camera audio) and fully mixed external audio files (protools I believe) of the performance.

When I take the protools exported audio (48k/16bit) and put it into my PAL FCP timeline, over time it drifts out of sync with the video. After 6 1/2 minutes it is out of sync by maybe 2 seconds. My questions are:

1) How should the audio be exported to match up to my PAL timeline (and why doesn't it just sync up)?

2) When we convert to NTSC, I will use the Nattress filters for video, but what do i need to be concerned about re: the audio?

Thanks a ton,
Jason

<<<I will use the Nattress filters for video>>>

You use Nattress for both. Typically, he works on performance duration independent of framerate. If your show started out 30.00 minutes and it's still 30.00 minutes, the audio will still sync.

<<<After 6 1/2 minutes it is out of sync by maybe 2 seconds.>>>

Without taking off my shoes and socks for calculations, somebody picked the wrong framerate for import/export. Might that be the difference between 25/29.97? I still don't quite get how you mixed the sound, but we assume at one time, it was perfect, so the DAT transfer was probably correct.

Koz

Kozikowski Wrote:
-------------------------------------------------------

> You use Nattress for both. Typically, he works on
> performance duration independent of framerate. If
> your show started out 30.00 minutes and it's still
> 30.00 minutes, the audio will still sync.

Here's the problem - until I get the PAL and audio to sync, i can't create the final edit that will then be transferred to NTSC... So I can't use the filter yet.



> Without taking off my shoes and socks for
> calculations, somebody picked the wrong framerate
> for import/export. Might that be the difference
> between 25/29.97? I still don't quite get how you
> mixed the sound, but we assume at one time, it was
> perfect, so the DAT transfer was probably
> correct.


Turns out it was recorded directly into ProTools at 96/24. So what we're trying to figure out is what frame rate the session was recorded at, etc.... And how that might affect matching to PAL video. Any ideas or suggestions would be very helpful!!

Thanks,
Jason

I think the issue is simply that there is no sync system between the video and audio recordings. You can't expect it to line up if they weren't locked to a common sync source, or at least one to the other.

Audio doesn't have a "frame rate" because it doesn't have frames. You can have SMPTE or EBU time code attached to audio and think of it in terms of video frames. But that code would be meaningless outside of a sync lock situation or a code chase on the audio side.

You'll need to adjust the audio to fit the video; once that's done PAL and NTSC conversion should hold sync within FCP - but that depends on how and whether 30 frames > 29.97 frames was dealt with.

<<<You can't expect it to line up if they weren't locked to a common sync source, or at least one to the other. >>>

Actually, you can. If both systems were crystal controlled (like on a movie set) they should hold sync for way longer than the poster's work is doing. His error is sooo big that it has to be something broken or the wrong timecode was chosen, or a timecode tape played free-wheeling with *no* sync or something like that.

<<<6 1/2 minutes it is out of sync by maybe 2 seconds.>>>

At the worst possible "normal" error drift rates, I would expect that sound to be maybe seven or eight frames off. Not two seconds.

There's still a little fuzziness in my head. I'm almost not sure how to ask this. Can you describe who had what parts of the show? What were the DAT people watching when they did their sound mix? What were you listening to when you cut your picture?

Koz

Koz, I agree if you're talking about drift between cameras, or between video and video-aware audio. The way I understand it they recorded the sound wild, direct to ProTools on a computer. No mention of any video anything, or of the computer getting any reference or time code of any kind. I think it's perfectly plausible to see the .4% drift. It's better than using a cassette recorder...

In any case it clearly wasn't planned or produced as a video project.

<<<It's better than using a cassette recorder... >>>

But not much.

<<<I think it's perfectly plausible to see the .4% drift.>>>

I'm still not comfortable with that. A .4% error means your 48 KHz sample frequency drifts to 46. I don't think any sound service ever gets that far off unless it's actually broken.

<<<In any case it clearly wasn't planned or produced as a video project.>>>

Well, that's pretty clear. Still, you can't cut sound in a vacuum (if for no other reason than it makes your ears pop). What were the sound people watching?

Koz

I say gently: .4% from 48kHz is 47.8kHz (or 48.2).

I like the idea of cutting sound in a vacuum. Sync problems just wouldn't exist!

<<<I say gently: .4% from 48kHz is 47.8kHz (or 48.2). >>>

Oh, right. Caffeine deprivation.

Koz