Audio drift problem solving on a mac

A

andrew00

Well-known member
Hey.

So I've been asked to edit some live music footage, shot recently. I've two main audio sources - one was a desk feed, recorded as an mp3 192kbps 44khz onto a usb stick, and the other a room mic recording into a digital recorder at 44khz 16 bit.

There's audio drift so the song goes out of sync by the end of the song, which is v. annoying.

How would people lining things up correctly? I'm editing on a Mac on FCP, but have a PC too with Vegas. Audio prog wise I can get access to Logic, STPro, Cubase and Soundforge through mates etc.

Currently I've tried importing both into Soundforge. There I added a compressor to get the volumes etc up and saved both as aiff's, but still drift.

Kinda don't know what to do so suggestions most welcome please!

Andy
 
Oh btw I've noticed from start to finish of the song the audio is out by FIVE frames - i.e. if sync'd at the start one source is 5 frames behind the other at the end. Can I do some software speed thing to fix this at all or is it just part of the inconsistency of things?
 
First thing is to check your sequence settings in FCP to make sure that the audio settings are matching.

How long is the piece from beginning to end?

You should be able to change the playback speed of the clip to get it to match up.

Best,
Andy
 
I've been working in Soundtrack Pro, and aside from a bug or two I like it a lot. I did see in there a feature which reportedly would stretch one audio clip into the time of another. I haven't tried it extensively, though I would certainly try it out.

http://manuals.info.apple.com/en/Soundtrack_Pro_User_Manual.pdf
http://manuals.info.apple.com/en/Soundtrack_Pro_2_User_Manual.pdf

Check out the Audio Stretching Tool (p. 210 for V2), or Time Stretch (p. 224 in V2). Either one should do what you're looking for.

My ONE word of warning with using Soundtrack Pro, however, is to make sure that you do an Export of the audio once in a while and make SURE that what you have is outputting as it should. There is a bug or two whereby if you make a change in the program, it will not nor never be exportable, thus rendering your work useless. I've only seen this, personally, in using the "feature" to select only one channel of audio for a particular clip, though your milage may vary. It's a bad thing to find out after working on something for a long time :)
 
rlaskey said:
My ONE word of warning with using Soundtrack Pro, however, is to make sure that you do an Export of the audio once in a while and make SURE that what you have is outputting as it should. There is a bug or two whereby if you make a change in the program, it will not nor never be exportable, thus rendering your work useless. I've only seen this, personally, in using the "feature" to select only one channel of audio for a particular clip, though your milage may vary. It's a bad thing to find out after working on something for a long time :)
Click to expand...

Interesting. I did quite a bit of work on STP and never came across that problem. You should report that to Apple. I suspect that the project is fixable and that you have a corrupt audio file but that is a total crap shoot. I know the Apple Pro Apps team would be very interested in a bug like your describing.
 
MP3 onto a memory stick... Your probably lucky a little drift is your biggest problem. Anyway time stretch as mentioned would be the way to go. I forget how STP implements it so I'm not sure if you can just type in five frames and have it stretch that amount. I would check that the drift is a steadily increasing drift. I would suspect that what you have is one or more dropped frames. In that case you don't want to stretch the whole file but cut it at the hiccups and fudge the transitions.

And FYI it's 44.1 not 44, might seem picky but there are applications where you could actually create a 44k file. And it would playback out of sync.
 
Interesting. I did quite a bit of work on STP and never came across that problem. You should report that to Apple. I suspect that the project is fixable and that you have a corrupt audio file but that is a total crap shoot. I know the Apple Pro Apps team would be very interested in a bug like your describing.
Click to expand...

Now that I look back, part of it is probably that my workflow was a bit strange. I had two channels of audio, one for each mic, and for a particular sequence I wanted only one channel and then the other, switching back and forth in order to swap between pieces of dialogue. Ergo, I used the "select channel" functionality for particular clips to have different mono channels.. but upon export it would not preserve these switching selections; it'd get the wrong channel, meaning one actor was almost silent and the other rather normal. I had looked it up and found a number of people online who experienced the exact same behavior, many of which had (supposedly) submitted bug reports. Since then, I changed to do most edits in straight two channel, no mono selections, and then made a second pass with the result to select which one I wanted at each part. I haven't had any troubles since. I only mention it because it was entirely disheartening to learn that what I wanted to do was not possible, and I had to think about it in a different way due to a problem with the program. Just a buyer (or user) beware. It's been entirely worth it (and wonderful) otherwise.

Anyway, didn't mean to drift too far off topic, so apologies for that.
 
Ah so you're saying if the drift is consistent then I can just stretch by what would amount to 5 frames and it should line up.

However, if it's not consistent, i.e. it's the odd random dropped frame like you suggested, then I'll need to cut the audio at the relevant parts when it goes out of sync - say by looking at the waveform and going 'oops there it slips', cutting there and moving the audio a frame so it's back into alignment.

With this in mind, should I cut the longer audio stream so it's in line with the shorter one, aka the one that has the dropped frames. Or should I cut the shorter clip and move the clip on a frame having 1 random frame of silence here and there (which would presumably be unnoticeable to the masses - I could even put a duplicate frame of the last one before the drop in there if needed?).

I'm thinking the latter option might be better because since it's syncing to the video I need to preserve the sync relative to the audio too.
 
You need to sync it to the video. I would bet that if there are dropouts it's with the MP3.
 
I'd bet the discrepancy is simply due to the two different devices' clocks disagreeing on the exact sample rate. It's quite common. If that's the case the 'stretch (or squeeze) to match' routine you mentioned should work. Just make sure to match to the version that plays in sync with the picture.
 
The same thing happened to me once and it was pretty much impossible to get the stretching spot on. Even if you figure out the math to do it, you will always have the two recordings drifting in and out of sync. Specially if one of your devices has a less than perfect sampling clock. In my case, I was using an H2 as room mic and Pro Tools recording close up mics. Now, having said that, my solution was, since the room mic was technically at a much greater distance than the close-up mics from the sound source, the sound arriving the at the H2 had a slight delay... If you do the stretch and use the sound recorded with the room mic at a much lower level relative to the board feed (to give the sound a bit of "reverb") you should be fine. At the end of each song, you can re-sync.

I hope this helps.
Pablo
 
You must log in or register to reply here.
Back
Top