audio drift problem at DVinfo.net

Matt Bishop · April 3rd, 2007, 11:36 AM

I am currently using an Olympus WS-320M to record as a backup lavalier or podium mic during weddings. I am using the following

sony vx2100s
panasonic dvx100b on some other projects
sony vegas

My problem is that the audio i'm getting from the recorder is a bit faster than any of my camera audio and therefore they don't match up after about a minute or so. I've manually moved each bit to match up when I needed it to, but obviously that is very time consuming and frustrating.

Any idea why they would be different speeds or what settings could cause something like this? I have also tried this in adobe premiere just for troubleshooting and got the same result. All audio is set to 16 bit. All cameras are running on SP mode. I'm at a loss at what could cause this.

I was told that resampling this might fix it. Does vegas do this automatically based on project settings? Along those same lines, I guess i'm curious as to what my vegas projects should be set at for audio. 48Khz? It's currently (and always) been set at 44.1 by default.

Thanks!

Colin Willsher · April 3rd, 2007, 12:50 PM

Hi Matt,

First step would be to check the record settings on the Olympus. What sampling rate is that recording at and is it WAV or MP3? I wouldn't use MP3 as the file lengths do change a little with the compression. Different sample rates would cause a much greater difference of a few seconds every minute so unless they are way out, that is unlikely to be the cause.

The normal sample rate for DV audio is 48Khz which is what the VX2100 uses (along with 32Khz. I don't think 44.1Khz is an option), so make sure you change your default settings in Vegas. You could then also try re-importing the audio into a new Vegas project already set to 48Khz - could be something to do with the sample rate conversion, I'm not sure. But at least there are a few things to check and try out while you wait for more help.

Good luck!

C

Roger Shore · April 23rd, 2007, 06:14 AM

I use the Olympus WS200S, which I believe is identical to the WS320-M in the recording department, and I have found a drift in the opposite direction -my tracks are too long!

I found a workaround, which I posted here: http://www.mfbb.net/myvideoproblems/...yvideoproblems

Seems to work OK, so it might help your problem

Steve House · April 23rd, 2007, 06:40 AM

You have several issues here. The most basic one is the recorder is sampling at the CD normal rate of 44.1 while video is 48kHz. When you playback the 44.1 file at 48, the sound that plays in one second has the samples that were recorded in 1.1 seconds of time originally.

Another issue is the accuracy of the clocks in little portable recorders like that is questionable. Successful double system sync sound requires a highly stable and accurate timebase in both the audio and video recorders and I just don't believe those mini dictation recorders have it in 'em.

Roger Shore · April 23rd, 2007, 07:19 AM

If you convert the sample type from 44.1KHz to 48KHz ,as described in the workaround procedure linked to above, then the length of file should remain the same as the original, so that the 'stretch' procedure can be matched exactly to the 48KHz track from the camera audio track, as both tracks are now sampled at 48KHz.

Steve House · April 23rd, 2007, 07:42 AM

Quote:

Originally Posted by Roger Shore

If you convert the sample type from 44.1KHz to 48KHz ,as described in the workaround procedure linked to above, then the length of file should remain the same as the original, so that the 'stretch' procedure can be matched exactly to the 48KHz track from the camera audio track, as both tracks are now sampled at 48KHz.

That assumes that it really is 44.1 to start with and not 44.2 or 44.0. For that matter, I saw somewhere that that Olympus recorder can also record at 32kHz, making it even worse. Clock accuracy still matters.

Roger Shore · April 23rd, 2007, 08:26 AM

Quote:

Originally Posted by Steve House

Clock accuracy still matters.

Absolutely! -- the whole purpose of my correction procedure!

The Olympus recorders record in WMA format, and although the sampling frequency for the 'HQ' mode is quoted as '44KHz', the files open as 44.1KHz, 16bit, when imported directly to Cool Edit.

Now while the actual crystal frequency may not be that accurate to start with, it certainly doesn't seem to 'drift' much in actual operation, and I have found it possible to correct the offset in an audio editor, as I describe, and line up the 'stretched' (or squashed!) track with satisfactory accuracy to the camera recorded audio for a 30 minute single 'take'.
I can't detect any 'drift' in the corrected sequence.

I would be interesting if the original poster,Matt, finds the procedure of any benefit in his case, where the clock offset is in the opposite direction to mine!

April 3rd, 2007, 11:36 AM	#1
Matt Bishop Major Player Join Date: Nov 2005 Location: East Moline, IL Posts: 231	audio drift problem I am currently using an Olympus WS-320M to record as a backup lavalier or podium mic during weddings. I am using the following sony vx2100s panasonic dvx100b on some other projects sony vegas My problem is that the audio i'm getting from the recorder is a bit faster than any of my camera audio and therefore they don't match up after about a minute or so. I've manually moved each bit to match up when I needed it to, but obviously that is very time consuming and frustrating. Any idea why they would be different speeds or what settings could cause something like this? I have also tried this in adobe premiere just for troubleshooting and got the same result. All audio is set to 16 bit. All cameras are running on SP mode. I'm at a loss at what could cause this. I was told that resampling this might fix it. Does vegas do this automatically based on project settings? Along those same lines, I guess i'm curious as to what my vegas projects should be set at for audio. 48Khz? It's currently (and always) been set at 44.1 by default. Thanks!

April 3rd, 2007, 12:50 PM	#2
Colin Willsher Regular Crew Join Date: Jun 2005 Location: Southampton, England, UK Posts: 70	Hi Matt, First step would be to check the record settings on the Olympus. What sampling rate is that recording at and is it WAV or MP3? I wouldn't use MP3 as the file lengths do change a little with the compression. Different sample rates would cause a much greater difference of a few seconds every minute so unless they are way out, that is unlikely to be the cause. The normal sample rate for DV audio is 48Khz which is what the VX2100 uses (along with 32Khz. I don't think 44.1Khz is an option), so make sure you change your default settings in Vegas. You could then also try re-importing the audio into a new Vegas project already set to 48Khz - could be something to do with the sample rate conversion, I'm not sure. But at least there are a few things to check and try out while you wait for more help. Good luck! C __________________ 2b Media Services

April 23rd, 2007, 06:40 AM	#4
Steve House Inner Circle Join Date: Mar 2005 Location: Hamilton, Ontario, Canada Posts: 5,742	You have several issues here. The most basic one is the recorder is sampling at the CD normal rate of 44.1 while video is 48kHz. When you playback the 44.1 file at 48, the sound that plays in one second has the samples that were recorded in 1.1 seconds of time originally. Another issue is the accuracy of the clocks in little portable recorders like that is questionable. Successful double system sync sound requires a highly stable and accurate timebase in both the audio and video recorders and I just don't believe those mini dictation recorders have it in 'em. __________________ Good news, Cousins! This week's chocolate ration is 15 grams!

April 23rd, 2007, 06:14 AM	#3
Roger Shore Regular Crew Join Date: Apr 2007 Location: UK Posts: 113	I use the Olympus WS200S, which I believe is identical to the WS320-M in the recording department, and I have found a drift in the opposite direction -my tracks are too long! I found a workaround, which I posted here: http://www.mfbb.net/myvideoproblems/...yvideoproblems Seems to work OK, so it might help your problem

April 23rd, 2007, 07:19 AM	#5
Roger Shore Regular Crew Join Date: Apr 2007 Location: UK Posts: 113	If you convert the sample type from 44.1KHz to 48KHz ,as described in the workaround procedure linked to above, then the length of file should remain the same as the original, so that the 'stretch' procedure can be matched exactly to the 48KHz track from the camera audio track, as both tracks are now sampled at 48KHz.