Audio drift? at DVinfo.net

Alfred Okocha · October 12th, 2004, 09:29 PM

Hi all.

I don't have any audio experience and I'm new to this so forgive me if this has been explained before.

I shot a concert and another guy recorded the audio seperately and masterized it.
Now it's time for us to put our video and audio together and it seems there's some kind of audio drift.. I matched it up in the beginning but after a couple of songs it's out of sync..

Does anybody have an idea what's happpening? (I'm using FCP3 and I was handed the audio as AIFF files. (That I had to export in QT for in not to render..)
The cam is NTSC.. could this somehow be a three-frame-problem? 29.97?

Thanks alot.

Douglas Spotted Eagle · October 12th, 2004, 09:42 PM

First, what is the sample rate of the audio in the concert files? I'll bet it's 44.1.
Second, do you have a tool to resample at 48? I don't recall any longer how FCP resamples. Been a while. :-)

Alfred Okocha · October 13th, 2004, 05:41 AM

Hi Douglas.

Yes, the sample rate was 44.1, 24bits. I opened it in QT and then exported as 48.000. (Otherwise I have to render it in FCP)
The file also got much smaller after the export.. (from around 650 to 450 mb) is that normal?

If I ask him to sample it at 48 in the future, are my problems solved then?
I'm sorry, I have no idea what I'm talking about here.. =(

Thanks.

Douglas Spotted Eagle · October 13th, 2004, 07:43 AM

File size should NOT have gotten smaller unless you used some form of compression. A 48K file of identical length to a 44.1 file should be bigger, not smaller.
So, I doubt your problem is solved. Sounds like you've compromised the audio quality as well.
Render it in FCP, or better, use Peak to conform it.

Alfred Okocha · October 13th, 2004, 11:39 AM

Can I keep the 24bits it was recorded on with Peak? Because with quicktime the maximum is 16bits.

Did the file get smaller because of this? If so, did I lose alot.. I mean will it be a big difference?

Does the 44.1 thing make the audio drift when it's tranfered to 48?

Sorry for the many questions.. Thanks alot!

Bjorn C Hagert · October 14th, 2004, 07:56 AM

As I recall, sample rate conversion is not 100% in sync. The reason for that is that the sample rate conversion algorithm will primarily go for sound quality and not for timing.

This may have changed since I last encountered it.

The "clock" in the camera and the computer may run slightly off so that may also be why.

Is there a large drift or can you manually sync it to picture? What I mean is can you get away with the drift in one song? Then do a discrete cross fade in the file between songs.

Alfred Okocha · October 14th, 2004, 11:21 AM

Tjena Bjorn.

Yeah, this is what I had to do now.. I could almost get away with it.. BUT it's alot of, what I feel, unnecessary work. So if there was a way to get rid of the problem in the future I'd be very happy!

Bjorn C Hagert · October 14th, 2004, 11:38 PM

Hej Alfred!

In the future run all stuff at 48kHz. The bit reduction should not do anything as I recall it. I run everything at 48/16.

I guess that if you feel adventurous you can record at 96kHz and try to down sample it. It just might be better since it is a multiple of 48kHz.

If you want it all in sync then it is black burst / house clock that you need. That is one clock that keeps all audio (and video) units in sync but then we have another price tag and the units has to have clock in.

Alfred Okocha · October 15th, 2004, 10:54 PM

Thanks Bjorn. Do you think I would get a better result with 96 even though it'll end up in 48 then?
please tell me more about this black burst.. I've never heard of it before.

Thanks.

Bjorn C Hagert · October 15th, 2004, 11:33 PM

I can only give you an outline as of sync since I have not been technically involved in it on the video side. I hope that some one can step in that has worked with it and correct me.

Here is a definition for video:

A composite colour video signal that consists of vertical and horizontal sync signals, and chroma burst. Black burst is used to synchronize (genlock) other video sources to the same sync and color information. A master black burst generator normally distributes master synchronization signals to an entire video production facility so that all equipment at that location is synchronized.

Source: http://zone.ni.com/devzone/nidzgloss.nsf/webmain/67226536E91E153386256885007E71E5

The units must accept black burst in.

What we really are after is audio sync.

As for audio you can sync ADATs so that they are sample accurate to a DAW so there will be no phase difference. You can run all the units off the computers clock.

If you do not use the camera sound together with the computer recording you can do a sync test. Run a DAT tape with an audio / visual cue (clapping hands) in the beginning and end. Record the sound on the computer. Import and see if you get a visual drift.

Samples are down to 1/48000 of a second but picture is only 1/25 of a second accurate. This way you can see what method and sample rate / bit depth that would really work.

96kHz give you more headroom BUT you need a lot better amplifiers for that. Personally my system can use 96/24 but I always run in 48/16. Then again, if you do a lot of acoustic recording 96/24 is great.

October 12th, 2004, 09:29 PM	#1
Alfred Okocha Major Player Join Date: Feb 2004 Location: London Posts: 427	Audio drift? Hi all. I don't have any audio experience and I'm new to this so forgive me if this has been explained before. I shot a concert and another guy recorded the audio seperately and masterized it. Now it's time for us to put our video and audio together and it seems there's some kind of audio drift.. I matched it up in the beginning but after a couple of songs it's out of sync.. Does anybody have an idea what's happpening? (I'm using FCP3 and I was handed the audio as AIFF files. (That I had to export in QT for in not to render..) The cam is NTSC.. could this somehow be a three-frame-problem? 29.97? Thanks alot.

October 12th, 2004, 09:42 PM	#2
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	First, what is the sample rate of the audio in the concert files? I'll bet it's 44.1. Second, do you have a tool to resample at 48? I don't recall any longer how FCP resamples. Been a while. :-) __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

October 13th, 2004, 07:43 AM	#4
Douglas Spotted Eagle VASST Join Date: Jan 2004 Location: Stockton, UT Posts: 5,648	File size should NOT have gotten smaller unless you used some form of compression. A 48K file of identical length to a 44.1 file should be bigger, not smaller. So, I doubt your problem is solved. Sounds like you've compromised the audio quality as well. Render it in FCP, or better, use Peak to conform it. __________________ Douglas Spotted Eagle/Spot Author, producer, composer Certified Sony Vegas Trainer http://www.vasst.com

October 14th, 2004, 07:56 AM	#6
Bjorn C Hagert Regular Crew Join Date: Jul 2004 Location: Stockholm Sweden Posts: 27	As I recall, sample rate conversion is not 100% in sync. The reason for that is that the sample rate conversion algorithm will primarily go for sound quality and not for timing. This may have changed since I last encountered it. The "clock" in the camera and the computer may run slightly off so that may also be why. Is there a large drift or can you manually sync it to picture? What I mean is can you get away with the drift in one song? Then do a discrete cross fade in the file between songs. __________________ //B C

October 14th, 2004, 11:38 PM	#8
Bjorn C Hagert Regular Crew Join Date: Jul 2004 Location: Stockholm Sweden Posts: 27	Hej Alfred! In the future run all stuff at 48kHz. The bit reduction should not do anything as I recall it. I run everything at 48/16. I guess that if you feel adventurous you can record at 96kHz and try to down sample it. It just might be better since it is a multiple of 48kHz. If you want it all in sync then it is black burst / house clock that you need. That is one clock that keeps all audio (and video) units in sync but then we have another price tag and the units has to have clock in. __________________ //B C

October 13th, 2004, 05:41 AM	#3
Alfred Okocha Major Player Join Date: Feb 2004 Location: London Posts: 427	Hi Douglas. Yes, the sample rate was 44.1, 24bits. I opened it in QT and then exported as 48.000. (Otherwise I have to render it in FCP) The file also got much smaller after the export.. (from around 650 to 450 mb) is that normal? If I ask him to sample it at 48 in the future, are my problems solved then? I'm sorry, I have no idea what I'm talking about here.. =( Thanks.

October 13th, 2004, 11:39 AM	#5
Alfred Okocha Major Player Join Date: Feb 2004 Location: London Posts: 427	Can I keep the 24bits it was recorded on with Peak? Because with quicktime the maximum is 16bits. Did the file get smaller because of this? If so, did I lose alot.. I mean will it be a big difference? Does the 44.1 thing make the audio drift when it's tranfered to 48? Sorry for the many questions.. Thanks alot!

October 14th, 2004, 11:21 AM	#7
Alfred Okocha Major Player Join Date: Feb 2004 Location: London Posts: 427	Tjena Bjorn. Yeah, this is what I had to do now.. I could almost get away with it.. BUT it's alot of, what I feel, unnecessary work. So if there was a way to get rid of the problem in the future I'd be very happy!

October 15th, 2004, 10:54 PM	#9
Alfred Okocha Major Player Join Date: Feb 2004 Location: London Posts: 427	Thanks Bjorn. Do you think I would get a better result with 96 even though it'll end up in 48 then? please tell me more about this black burst.. I've never heard of it before. Thanks.

October 15th, 2004, 11:33 PM	#10
Bjorn C Hagert Regular Crew Join Date: Jul 2004 Location: Stockholm Sweden Posts: 27	I can only give you an outline as of sync since I have not been technically involved in it on the video side. I hope that some one can step in that has worked with it and correct me. Here is a definition for video: A composite colour video signal that consists of vertical and horizontal sync signals, and chroma burst. Black burst is used to synchronize (genlock) other video sources to the same sync and color information. A master black burst generator normally distributes master synchronization signals to an entire video production facility so that all equipment at that location is synchronized. Source: http://zone.ni.com/devzone/nidzgloss.nsf/webmain/67226536E91E153386256885007E71E5 The units must accept black burst in. What we really are after is audio sync. As for audio you can sync ADATs so that they are sample accurate to a DAW so there will be no phase difference. You can run all the units off the computers clock. If you do not use the camera sound together with the computer recording you can do a sync test. Run a DAT tape with an audio / visual cue (clapping hands) in the beginning and end. Record the sound on the computer. Import and see if you get a visual drift. Samples are down to 1/48000 of a second but picture is only 1/25 of a second accurate. This way you can see what method and sample rate / bit depth that would really work. 96kHz give you more headroom BUT you need a lot better amplifiers for that. Personally my system can use 96/24 but I always run in 48/16. Then again, if you do a lot of acoustic recording 96/24 is great. __________________ //B C