Topic: Time drift issue with video

[mmadd29]

I have an issue with time drift when put my audio to video.

I usually do matrix sound.  I use wavlabs, and do an audio montage.  I figure out the drift, and use r8brain, and a sheet to do the calculation I got from TS.  This always works perfectly.....


I then put the video into Adobe, and add the matrix file.  I sync the beginning.  I then turn off the camera audio and go to the end.  The end is slightly off.  I figured all I needed to do was to export the camera audio, figure the drift and reclock.  This doesn't work.

What is the correct way to sync audio to video?

Thanks in advance

[stevetoney]

I strip the audio from the video file and use that as the 'base'.  NEVER CHANGE THE BASE because that's of course synched perfectly with the video.  I then get the new audio that I want to master and 'parallel' that up to the base.  there are two ways you need to match the two files; a) synching for differences in audio drift, and b) synching for drop outs that may have occurred in the video. 

You already know how to correct for differences in audio drift to address a) above.

Regarding b) above, hopefully there are no drop outs on the video because then it's just a matter of linearly shrinking or stretching the new audio to match the base audio.  However, if there are drop outs or other issues with the base audio, then you have to break it down into smaller and smaller chunks to get the two audio tracks to exactly parallel each other. 

I've had some cases where I had to build the new audio minute-by-minute...a real PITA which can take many hours.  Usually, that's not the case though.  It's usually a case where there might be a slight video drop out every ten or twenty minutes.  In that case, I have to find where the drop out occurs and figure out, based on the differences in the two sources, what the length of time the drop out was by comparing the files and fixing the new file to exactly match the base.  OK, how to do this...

At the beginning of the recording, you make sure that the audio's start in exactly the same point.  So, you need to either add or subtract from the beginning of the new audio so it starts at the EXACT same nanosecond as the base.  If I need to add time, I just add silence to the start of my new audio track.  So, then the two audio's are exactly parallel with each other at time = 0 sec. 

Next, you adjust the new audio for time drift...you said you know how to do this.  Be careful here though because if there are drop outs on the base audio, your ratio will be off.  Make sure the new drift ratio is based on an apples-to-apples piece of the recording...IOW a length of audio where neither of the audio's have any drop-outs.

Next you have to seek out and correct the new audio to insert the same drop outs that may exist in the base audio.  Do this by comparing the time signatures of the two audios.  Obviously if that drum hit at 20.005 happens at 20.005 of both recordings (after drift adjustment), then there's no drop outs and you're OK through that point in the recording. 

However, if that same drum hit happens at 20.205 of the new recording, then you've got at least one drop out (maybe more) with a total duration of 0.200 seconds long.  This drop out happens at a point in the audio prior to 20.005.  Your challenge is to find it and then strip that 0.200 portion out of the new audio to make the new audio parallel up to the base audio.  Do this for the entire length of the new audio and you're finished building your new audio file.

Finally, mux the corrected master audio file to the video file and the two should match up perfectly, since your new master is exactly paralleled up to the audio that was stripped away from the video.

Hope this helps.

[TNJazz]

What is the correct way to sync audio to video?

The correct way is to use a word clock and SMPTE Time Code to keep both audio and video in sync.


(hey, you asked!)

[mmadd29]

What is the correct way to sync audio to video?

The correct way is to use a word clock and SMPTE Time Code to keep both audio and video in sync.


(hey, you asked!)

Yes I did 

I will be reading up on that...and will probably be crazy afterwards.....

[mmadd29]

I strip the audio from the video file and use that as the 'base'.  NEVER CHANGE THE BASE because that's of course synched perfectly with the video.  I then get the new audio that I want to master and 'parallel' that up to the base.  there are two ways you need to match the two files; a) synching for differences in audio drift, and b) synching for drop outs that may have occurred in the video. 

You already know how to correct for differences in audio drift to address a) above.

I have done the capture with WinDV.  There has never been any frame drops, always 0.  Now during recording on miniDV tape, can there be dropouts?

When I used to use Stoik there were a ton of drop outs.

Regarding b) above, hopefully there are no drop outs on the video because then it's just a matter of linearly shrinking or stretching the new audio to match the base audio.  However, if there are drop outs or other issues with the base audio, then you have to break it down into smaller and smaller chunks to get the two audio tracks to exactly parallel each other. 

I did get the shows working yesterday by slightly streching the audio and slightly shrinking the video.  I'm taking like .12% or so.

I've had some cases where I had to build the new audio minute-by-minute...a real PITA which can take many hours.  Usually, that's not the case though.  It's usually a case where there might be a slight video drop out every ten or twenty minutes.  In that case, I have to find where the drop out occurs and figure out, based on the differences in the two sources, what the length of time the drop out was by comparing the files and fixing the new file to exactly match the base.  OK, how to do this...

At the beginning of the recording, you make sure that the audio's start in exactly the same point.  So, you need to either add or subtract from the beginning of the new audio so it starts at the EXACT same nanosecond as the base.  If I need to add time, I just add silence to the start of my new audio track.  So, then the two audio's are exactly parallel with each other at time = 0 sec. 

Next, you adjust the new audio for time drift...you said you know how to do this.  Be careful here though because if there are drop outs on the base audio, your ratio will be off.  Make sure the new drift ratio is based on an apples-to-apples piece of the recording...IOW a length of audio where neither of the audio's have any drop-outs.

Next you have to seek out and correct the new audio to insert the same drop outs that may exist in the base audio.  Do this by comparing the time signatures of the two audios.  Obviously if that drum hit at 20.005 happens at 20.005 of both recordings (after drift adjustment), then there's no drop outs and you're OK through that point in the recording. 

However, if that same drum hit happens at 20.205 of the new recording, then you've got at least one drop out (maybe more) with a total duration of 0.200 seconds long.  This drop out happens at a point in the audio prior to 20.005.  Your challenge is to find it and then strip that 0.200 portion out of the new audio to make the new audio parallel up to the base audio.  Do this for the entire length of the new audio and you're finished building your new audio file.

Finally, mux the corrected master audio file to the video file and the two should match up perfectly, since your new master is exactly paralleled up to the audio that was stripped away from the video.

Hope this helps.

Thanks for that tip.  I have printed this out for future reference.  Hopefully I don't have to do that..

Thanks for all the tips

[stevetoney]

Sounds like you got it, but shrinking video isn't the way to go IMHO because you end up losing more video.

I strip the audio from the video file and use that as the 'base'.  NEVER CHANGE THE BASE because that's of course synched perfectly with the video.  I then get the new audio that I want to master and 'parallel' that up to the base.  there are two ways you need to match the two files; a) synching for differences in audio drift, and b) synching for drop outs that may have occurred in the video. 

You already know how to correct for differences in audio drift to address a) above.

I have done the capture with WinDV.  There has never been any frame drops, always 0.  Now during recording on miniDV tape, can there be dropouts?

When I used to use Stoik there were a ton of drop outs.

There are drop outs all the time on my videos (Sony PC100 > miniDV).  I think it probably depends on the video camera's tape transport and perhaps how well the miniDV tape has been packed/unpacked just prior to recording.

BTW, I'm NOT talking about frame drops in the process of capturing the video onto your hard drive.  I'm talking about where there's small sections of video that might not have made it onto tape, causing your video to by mis-synched with your audio.  Dropped frames during the video capture is 100% different thing.

Regarding b) above, hopefully there are no drop outs on the video because then it's just a matter of linearly shrinking or stretching the new audio to match the base audio.  However, if there are drop outs or other issues with the base audio, then you have to break it down into smaller and smaller chunks to get the two audio tracks to exactly parallel each other. 

I did get the shows working yesterday by slightly streching the audio and slightly shrinking the video.  I'm taking like .12% or so.

This isn't the way to go, IMHO.  Going this way ends up messing up more video when you don't have to.  I leave my video intact and synch the new audio to match.  IMHO, the video is the most important aspect of...well a video tape, not the audio. 

BTW, I understand that the way your program shrinks video is that it removes frames which per your first paragraph is something that you're concerned about.  That's legitimate...as I said, don't let the program shrink or stretch your video because it can end up being intrusive to the end result whereas adjusting audio is far less so.

I've had some cases where I had to build the new audio minute-by-minute...a real PITA which can take many hours.  Usually, that's not the case though.  It's usually a case where there might be a slight video drop out every ten or twenty minutes.  In that case, I have to find where the drop out occurs and figure out, based on the differences in the two sources, what the length of time the drop out was by comparing the files and fixing the new file to exactly match the base.  OK, how to do this...

At the beginning of the recording, you make sure that the audio's start in exactly the same point.  So, you need to either add or subtract from the beginning of the new audio so it starts at the EXACT same nanosecond as the base.  If I need to add time, I just add silence to the start of my new audio track.  So, then the two audio's are exactly parallel with each other at time = 0 sec. 

Next, you adjust the new audio for time drift...you said you know how to do this.  Be careful here though because if there are drop outs on the base audio, your ratio will be off.  Make sure the new drift ratio is based on an apples-to-apples piece of the recording...IOW a length of audio where neither of the audio's have any drop-outs.

Next you have to seek out and correct the new audio to insert the same drop outs that may exist in the base audio.  Do this by comparing the time signatures of the two audios.  Obviously if that drum hit at 20.005 happens at 20.005 of both recordings (after drift adjustment), then there's no drop outs and you're OK through that point in the recording. 

However, if that same drum hit happens at 20.205 of the new recording, then you've got at least one drop out (maybe more) with a total duration of 0.200 seconds long.  This drop out happens at a point in the audio prior to 20.005.  Your challenge is to find it and then strip that 0.200 portion out of the new audio to make the new audio parallel up to the base audio.  Do this for the entire length of the new audio and you're finished building your new audio file.

Finally, mux the corrected master audio file to the video file and the two should match up perfectly, since your new master is exactly paralleled up to the audio that was stripped away from the video.

Hope this helps.

Thanks for that tip.  I have printed this out for future reference.  Hopefully I don't have to do that..

You're welcome, however, my response to your last sentence is that your original post asked for the correct way to synch audio and video.  If you synch your audio to video the best/correct way, you'll have to use the process I've outlined above (or something similar) 100% of the time!   

[mmadd29]

There are drop outs all the time on my videos (Sony PC100 > miniDV).  That's why I mentioned it.    I think it probably depends on the video camera's tape transport and perhaps how well the miniDV tape has been packed/unpacked just prior to recording.

BTW, I'm NOT talking about frame drops in the process of capturing the video onto your hard drive.  I'm talking about where there's small sections of video that might not have made it onto tape, causing your video to by mis-synched with your audio.  Dropped frames during the video capture is 100% different thing.

I thought you were talking about the recording but wasn't sure.  Basically this means there will always be sync issues, which would explain the reason it is always there.

This isn't the way to go, IMHO.  Going this way ends up messing up more video when you don't have to.  I leave my video intact and synch the new audio to match.  IMHO, the video is the most important aspect of...well a video tape, not the audio. 

BTW, I understand that the way your program shrinks video is that it removes frames which per your first paragraph is something that you don't want.

What is the actual effect of doing this?  The rendered video looks correct throughout, although I'm sure the way below is much better.  Per the Adobe manual it says it only strech/shrink, doesn't remove video/audio

LOL.  If you synch your audio to video the correct way, you'll have to do this 100% of the time!   


I'm sorry if the above process seems cumbersome to you, but you asked the question and I gave you the response for the best way to synch audio and video when they aren't clocked together.   

I was referring to the audio portion not he video portion.  The reason I printed this out, is so that I can refer to it while trying this technique.  One of my issues, other than still learning Adobe Premiere, is that I don't like the timeline, more correctly would be I'm not totally familiar with the timeline.  I haven't figured out how to tell where I am down to the milli second.  It is much easier in wavelab, or Cool Edit, and that is probably due to I have been using them for years....

I really do appreciate you taking the time to explain this

[Gordon]

I use vegas 8 and have never had to do any type of drift.  I know I've been lucky.   working with digital video and digital audio many times there is no drift.  that doesn't mean that it's always easy to sync

[stevetoney]

BTW, I understand that the way your program shrinks video is that it removes frames which per your first paragraph is something that you don't want.

What is the actual effect of doing this?  The rendered video looks correct throughout, although I'm sure the way below is much better.  Per the Adobe manual it says it only strech/shrink, doesn't remove video/audio

I'm not really sure.  I think I read a long time ago that when a program like Vegas invokes a shrinking, that it does a calculation of how many total frames need to be dropped and then uniformly deletes that number of frames from the duration so that the net effect isn't noticeable.

Similarly, it seems like I read that when it needed to be stretched it somehow adds frames (not sure if they're duplicated frames inserted linearly or perhaps a blank frame (you know, such that say one black frame every so often wouldn't be noticeable.

Take care and good luck!

[stevetoney]

I use vegas 8 and have never had to do any type of drift.  I know I've been lucky.   working with digital video and digital audio many times there is no drift.  that doesn't mean that it's always easy to sync

Wow that IS lucky.  In my experience, if I want to synch an external audio to a video, they rarely synch up perfectly and, of course, the longer the video is, the more of a chance there is that the new audio with drift apart from the original video.

[stantheman1976]

I always have some drift with longer recordings but I use Vegas also and it's very easy to correct.  I even came up with a little trick to make sure the amount I correct for is just right.  I go to the end of the tracks after the beginning is lined up and find a distinctive spot on both.  I make a loop region between the beginnings of each point so I can see exactly how much the drift is.  I drag the loop region to the end of my new track and use it to stretch/shrink exactly that amount.  Works like a charm.

[Gordon]

how long of recordings?  working with hour or hour and a half I honestly haven't noticed anything.  I match at the beginning and end and check in the middle etc (using drum hits for example).

[beatkilla]

Maybe i'm a bit more picky about drift,but in my experience with using minidisc and mini dv the drift happens around every 15mins.If i use 2 minidisc recorders one himd in hisp and a sharp legacy md in lp2 there is no drift at all between the md sources.I also notice that if one mini dv is in lp and the other in sp that the lp tape around 15mins in will be ahead 1 frame compared to the sp tape.But thats my experience i'm am very picky about drift though.Since most of my shoots are 3 to 7 cams i do not stretch anything i keep the md audio as my master source(and i only use the md sound in final mix)i check for drift and since i have multiple cams i just split where necessary and nudge back or forwards a frame on all cams.i just pick spots where i can use another cam angle or if you have only 1 cam and like this method you can copy 1 frame from the gap and you'll never see the split.but if your cams are in different areas like 1 in back of hall and 1 up close audio synching doesn't work.You can use camera flashes or other tricks to synch just the videos together.

[willndmb]

if your audio is 44.1 and NOT 48 you are more likely to have drift issues too

[guitard]

I will echo the comments of some of the others here - I almost always experience a little bit of drift - even with two modern digital recordings.

I think it's gear specific to some extent.  I say this because I get recordings to author to DVD from a particular filmer and the drift differential is almost always the same for this one cam and audio recorder he uses.  However, when I use an alternate audio source, it's slightly different.