Audio drift and the solutions

I

Inprogress

Well-known member
Just when I thought I knew what portable recorder I want, I read about audio drift. Ok, given, in most my future projects I don't forsee it being an issue, however, the opportunity that has present itself to me to get into the media industry is the recording of a monthly conference - long recording times!

If I sync my video and audio with a movie slate in my NLE, will audio drift still be an issue? Does it depend on the frame rates?

All that said, thankfully the edited product will not be one laborious one angle shot, but cut between two angles and possibly three. So I suppose audio drift will not be an issue, but the question still remains since thus far my Google searches only showed posts and blogs related to software specific problems that users found.

Thank you all for entertaining a audio greenhorn. Please do point me at a website/book to read up on it if it is a larger issue than explainable on a forum.

P.S. Recording via Panasonic GH2. Audio from lineout of pro-audio setup for the conference thus requires audio recording from the audio deck to a recorder - possibly then record straight into one of the stationary GH2's? Maybe via a BeachTek for XLR input?
 
I'm thinking a GH2 is not a good choice for this... sounds like an indoor event, and lighting is rarely optimal. Setting up your own lighting for something like this is unlikely.

I think I'd go with broadcast-type cameras for events like this (like my Canon XH A1, or similar from whatever company).


Regardless, it may be better in a case like this to record the audio independently and match it up in post, particularly if you are sticking with the GH2s.

Remember that you'll need to match up the cameras to each other too -- if you are capturing different angles from different cameras, you would still only have the "good" recording on one of them, so you'll be syncing to all of the other cameras as you cut between them anyway.

FCPX has a nice feature that tries to match up the soundtrack from a separate recording based on the lower-quality recording of the built-in mic (or whatever) that is in the video footage, so if you let the built-in cameras record too, and happen to be using FCPX, it might be able to link everything up for you. It can even sync the cameras to each other using the soundtracks in most cases.

http://provideocoalition.com/index....ut_pro_x_multicam_webinar_questions_answered/
 
Inprogress said:
Just when I thought I knew what portable recorder I want, I read about audio drift. Ok, given, in most my future projects I don't forsee it being an issue, however, the opportunity that has present itself to me to get into the media industry is the recording of a monthly conference - long recording times!

If I sync my video and audio with a movie slate in my NLE, will audio drift still be an issue? Does it depend on the frame rates?

All that said, thankfully the edited product will not be one laborious one angle shot, but cut between two angles and possibly three. So I suppose audio drift will not be an issue, but the question still remains since thus far my Google searches only showed posts and blogs related to software specific problems that users found.

Thank you all for entertaining a audio greenhorn. Please do point me at a website/book to read up on it if it is a larger issue than explainable on a forum.

P.S. Recording via Panasonic GH2. Audio from lineout of pro-audio setup for the conference thus requires audio recording from the audio deck to a recorder - possibly then record straight into one of the stationary GH2's? Maybe via a BeachTek for XLR input?
Click to expand...

There are several source of 'drift', and perhaps in some cases 'drift' is not the correct term... but anyway...

1) Different clocking circuits. This the primary source of clock differences that will add up to 'different times', after some long recording. The way that 'real' pros deal with is, is to have a 'master clock' and everything is synchronized to that master clock. In the case of Camera and Recorder, the two need to be joined together and one declared the 'master' and the other the 'slave'... The basic reason for 'clock' differences is that the devices were made to different tolerances of components, and so, they will just be 'different'... This holds for equipment made by different manufacturers, and it will hold for equipment made by the same manufacturer... of course one reason for the 'expensive' equipment matching up better is the use of higher quality components... and so there will be less variant from the 'expensive' product line... or so we would hope...

2) Even with a single device, given the specifications of the components, temperature variations will cause 'drift' relative to some absolute time reference... for example the Buro of Standards Cesium clock buried in the middle of Colorado... or somewhere...
This is less of a problem if one is in an air conditioned area, and the temperature variation is minimal... A shoot that is on location may have significant temperature variation, and so there may be more drift in such a situation.

3) idiosyncratic 'drift'... s*%t happens... can't figure out why...

4) You may be the luck lotto winner and all the various component differences add up to 'no drift'... for that day... that hour...



-------------

The easiest way around this is to keep the shot times short... the drift will most likely be small in modern circuits. If shot is not short... cutting to other cameras, will allow each camera to be sync'd to the audio of that moment, and so minimize the effect over the long term.

Since I've not had to do 'long takes'... and further don't have software that 'automagically' matches up externally captured audio to the camera reference audio... I've not really gotten into this sort of thing...

I will note that this sort of application is probably where I would recommend an all-in-one video solution, rather than the dual system solution... if the video camera maker can't make the internally captured audio sync up with the video... get a better camera...
 
Last edited:
Very interesting, thank you all. I must confess I am not sure what the duration of the conference is. This new thing I just learned (audio drift for a lack of better description) is something that I still need to wrap my head around especially considering the cutting between footage of different camera's. It seems the foot in the door of the media industry is now causing additional expenses where most other applications, the GH2 will be perfect. My research will have to continue of the suitability of specifically the GH2. It seems rental at this point in time might be the better solution - an all in one prosumer camera for the application - until such a time a two units can be purchased when the frequency of use of that camera increases.

Again I thank you all, and hope more experiences (hopefully GH2 specifc) will come out of the woodwork.
 
The short answer is that unless you are running sync to every device there will be drift. How much depends on the devices and that goes down to production run differences.

The good news is that even the cheapest recorders and cameras today have incredibly accurate clocks compared to a decade or two ago so the drift is going to be small. Depending on length of take and how much drift there is and how critical sync is, it may not be off enough to worry about.

The only way to know is to run some tests with THE SAME equipment you plan on using and seeing how far off they get.

Since you are going to do some editing you are probably going to be OK.


Now if you are serious about working in the field go get those books and start reading. If you plan to work in the field you better know the "long" explanations because you won't have time to get the short ones and still keep a job. And the info is changing all the time. The above (RE: drift) is only true in the last few years. Before the new generation of file based recorders you could have some serious drift. A little before that and with out some sort of sync control you would not only have drift but varying speeds so essentially no possible way to get sync back in post except through a lot of editing.
 
If I sync my video and audio with a movie slate in my NLE, will audio drift still be an issue? Does it depend on the frame rates?

Conflicting frame rates are usually a problem, but the issue here is clock time. Each device has a clock. They may or may not be close enough to remain the same for long periods.

Science! Go Figure.

Regards,

Ty Ford
 
I predict that there will be next-generation novel solutions to this sync problem. Once we start getting DSLRs that will reliably shoot "long-form" video there will be a bigger demand for sync audio recording solutions. (Because I also predict that the DSLR people will NEVER give audio its proper priority in their designs.)

One obvious (and rather simple) solution is to design an audio recorder that "samples" the video monitor out of the camera to establish a frame-rate and "genlocks" (using software phase-lock loop) the audio sample rate to the frame-rate of the camera. In previous generations of production equipment, this would have been prohibitively expensive to do for a "popular-price" product. But modern digital hardware and software solutions make this pretty easy to do. I wouldn't be surprised if we saw more than one product like this within the year. I have no insider knowledge of the likely sources, but the technology is just lying there begging to be deployed. It would make sense to see this as the next generation of a modern product like the Foxtex DC-R302, for example.
 
One "save your butt" thing to do is to always record audio on the cameras as a guide track. That way you have a sync reference if you need it and you can quickly tell if you are drifting out of sync since you will hear the audio start to "phase" (start sounding strange and echoey).
 
Many folks use 'Plural Eyes or Dual Eyes' to sync two separate audio tracks.. however.. AFAIK, that won't correct a drift problem. Most just make a break in the file and re-sync, before it becomes noticeable.
Another option, some NLEs and DAWs can time-stretch/compress the audio track, thus matching the tail-to-tail of the file.
 
Noiz2 said:
One "save your butt" thing to do is to always record audio on the cameras as a guide track. That way you have a sync reference if you need it and you can quickly tell if you are drifting out of sync since you will hear the audio start to "phase" (start sounding strange and echoey).
Click to expand...

IMHO, this is MUCH more than simply a "save your butt" concept. It is ESSENTIAL when using modern consumer and pro-sumer gear. Even if only using the built-in microphone on the camera. It is more than presumptuous, it is FOLLY to think you can get away with establishing sync and tracking unless you are recording a guide track on the camera.
 
Language differences. I would argue that "Save your Butt" and "Essential" are not mutually exclusive.

Since folks new to filmmaking are used to ignoring "essentials", and often getting away with it, "Saving their butts" might have more impact.

But if you want to get picky about it the film industry has established sync for 80-90 years now with out recording sound in camera so if it's FOLLY it's a folly that has been working pretty well.
 
Noiz2 said:
Language differences. I would argue that "Save your Butt" and "Essential" are not mutually exclusive.

Since folks new to filmmaking are used to ignoring "essentials", and often getting away with it, "Saving their butts" might have more impact.

But if you want to get picky about it the film industry has established sync for 80-90 years now with out recording sound in camera so if it's FOLLY it's a folly that has been working pretty well.
Click to expand...


A) I wasn't trying to CONTRADICT what you were saying, I was trying to REINFORCE it.
2) Those 80-90 years of Hollywood cameras without sound WERE using one of several forms of hard-sync. They started with hard-wired 3-phase power generated and distributed throughout the lot. The same power was used to power synchronous motors which drove both the cameras and the sound recorders. The motors coming up to lock RPM was the origin of the response "Speed!" as the notification that sound was rolling. Just before the era of SMPTE timecode, we were using "Pilot-tone" where the camera produced a signal (typically matching the mains frequency) that was recorded on the sound recorder on a separate track to form "virtual sprocket holes". Those old systems produced much better results than anything you can get from even professional video equipment without employing similar hard-sync measures (timecode, wordclock, etc.)
 
LOL, you guys are arguing about being in agreement with each other. Sheesh.

There is another "work-around" as well......... Have cut-aways. Or a second camera set up. This allows for you to edit between camera footage and "correct" for any audio drift by picking up any loss using a different angle. Jumps in either the audio or video, when it is supposed to be constant can be jarring. But using a second camera or the cut-aways, will clean those "jumps" up.
 
Doc Bernard said:
LOL, you guys are arguing about being in agreement with each other. Sheesh.

There is another "work-around" as well......... Have cut-aways. Or a second camera set up. This allows for you to edit between camera footage and "correct" for any audio drift by picking up any loss using a different angle. Jumps in either the audio or video, when it is supposed to be constant can be jarring. But using a second camera or the cut-aways, will clean those "jumps" up.
Click to expand...
Only problem with that is it only allows you to "adjust" sync in full-frame or greater increments since you can't cut between shots in the middle of a frame. Broadcast standards can be tighter than that, such as zero tolerance for audio to lead picture, 1/4 frame for audio to lag picture as being the maximum deviations from dead-sync allowed at any point in the program. That being said I see an awful lot of sloppy work making it to air these days with even network programming and national ads all too often running clearly out of sync. Don't know if the problem is in the original production or in the fact that audio and video can be transmitted on different satellite channels out to the outlets or if they go through separate chains at the cable head-end and there's delay in one of the chains but it really looks amateurish, yet another example of slipping standards and dumbing down.
 
Last edited:
Steve House said:
Only problem with that is it only allows you to "adjust" sync in full-frame or greater increments since you can't cut between shots in the middle of a frame.
Click to expand...

Um, this is a sound section so I'm kind of assuming we can use a DAW for sound post??? So make that adjust by a single sample. Doing your sound in a NLE is like painting a portrait with a four inch wall brush. It can be done with effort but the results are likely to be a bit crude and it's just the wrong tool.


As to our violent agreement ;~) Nagra's and cameras were also using crystal sync which did not require a connection.

But we were talking about syncing up sound to picture, or at least I was, not drift. Having a guide track on your video doesn't fix drift. And syncing the head of the take with a clapper is the way it's been done for decades, in film. Video is a bit of a different beast since they went to time code since they didn't have sprocket holes. Film though hardly ever used Time Code. We used to use it in post to sync to the video we were using for editing but the film stayed in F+F. All notes were in F+F and we were mixing to mag till the late 90's. All sync was via pops and start marks.

There are really two issues which are getting your tracks in sync and drift. With todays recorders and cameras any "normal" narrative shot, as in under 4 min, drift is generally a non issue. Most of these new toys will be within a quarter of a frame for more than a half hour and that is as close as you need to be.

IF your recording music with multiple devices then you need to worry.

Syncing takes can easily be done the way it has in film with a slate.

The "save your but" part for me is that on low budget films slates can be "forgotten", editors sometimes don't sync and marry the second system sound before they cut, or let them slip during the edit. In all those cases some guide track will save a LOT of time when sound post has to "fix it".

Strictly speaking though it isn't necessary IF everyone does their job and you keep a good workflow.

That is our slight divergence. I would recommend always recording sound on all cameras. But if it were MY shoot and I was running it all I wouldn't sweat not having sound on all cameras. I would do it if it were easy but I wouldn't go to a lot of extra effort. If I were in charge of the shoot but not involved in post I would insist on sound on all cameras.
 
Steve House said:
...
That being said I see an awful lot of sloppy work making it to air these days with even network programming and national ads all too often running clearly out of sync. Don't know if the problem is in the original production or in the fact that audio and video can be transmitted on different satellite channels out to the outlets or if they go through separate chains at the cable head-end and there's delay in one of the chains but it really looks amateurish, yet another example of slipping standards and dumbing down.
Click to expand...

Ironically I think this has more to do with "advancing" standards ;~) The digital transmission and display is full of places where delay can slip in. Most of these devices put a priority on getting the picture intact and let sound fend for itself. Plasma displays have a noticeable video processing lag. Depending on how you are getting your audio it may or may not go through the built in delay to get it back in sync with the picture. Generally if you are not using the built in speakers you need to install a delay line.

Some of those shows might be mixed to displays with a lag that has not been compensated for but I doubt any major show has. Post sound folks are as a rule quite anal about such things.

But a fun fact I like to mention in these threads is that there is about a three frame sync difference from the front row to the back row of any good sized movie theatre. Two thirds back is where the mixers were sitting so that is the only place in the house that a film is actually "in sync".

For some reason I don't really understand we (the audience) compensates and generally doesn't notice this. But take that same film and move it a frame out of sync and everyone will think the sync is off, even though for most of the house the "sync" was no more off than it was with the "in sync" version.

The brain I assume has learned what "sync" should be at various distances from the screen and adjusts the playback.
 
Noiz2 said:
Ironically I think this has more to do with "advancing" standards ;~) The digital transmission and display is full of places where delay can slip in. Most of these devices put a priority on getting the picture intact and let sound fend for itself. Plasma displays have a noticeable video processing lag. Depending on how you are getting your audio it may or may not go through the built in delay to get it back in sync with the picture. Generally if you are not using the built in speakers you need to install a delay line.
Click to expand...

Indeed. I think it is a combination of newer digital transmission artifacts (as Noiz2 says) plus the explosion of channels. We no longer have the luxury of a dedicated human monitoring the origination point of a network feed. We have hundreds of channels and each likely has various sources and transmission paths. And then there are dynamically allocated network transmission paths, all of which add up to a/v sync all over the map by the time it gets to the viewer. Even prime-time shows from the mainline networks are not immune as they pass through local affiliates. So we end up putting lipsync into the hands of the end-user if they are motivated enough to find the controls buried in the setup menu. :(
 
Um, this is a sound section so I'm kind of assuming we can use a DAW for sound post??? So make that adjust by a single sample. Doing your sound in a NLE is like painting a portrait with a four inch wall brush. It can be done with effort but the results are likely to be a bit crude and it's just the wrong tool.
Click to expand...
Painful. This is an audio section and here you are advocating us do the work to fit the video, rather than video fit the sound.

LOL, just playing around. But us audio guys have to make sure we take a stand with our work. Needs to be re-synced and there is drift? Cool, cut the vid to match.

Actually, it can be easier to edit the vid to match the audio, though.
 
The sound slip problem is something we had to deal with every edit when we were using sepmag with real film and a Nagra. Even with the clever crystal controlled locks and clever systems there was still drift. Now we edit digitally, even without plural eyes plug ins and that kind of thing, it's rarely a big issue. You simply find a suitable sync moment where sound and vision events are obvious, and then check on a similar event later and see if they align - if they do, move on. When you find a problem - fix it! Your sound is quantised to 1/25th (or 1/30th) anyway - so absolute sync is never going to happen unless you are really keen and go out of the editor to an audio editor to do it perfectly. There's usually some silence or continuous noise where you can shrink or stretch. Moving a frame or two with a crossfade is pretty well seamless. I've never had slippage that wasn't easily fixable. Worst examples come from the cheaper consumer equipment where I guess quality control doesn't extend to absolute timing - because it's not really an essential feature.

As long as you have audio recorded on the camera, just looking at the waveforms lets you check sync really easy, and fixing it too, is easier.

What I can't understand is why this is turning into a big deal. Very often distance makes sync inaccurate anyway - so just treat fixing sync as part of the editing process and just get good at doing it. Most of my stuff is multi cam so sync fixing is what you have to do. If a video source is a less accurate camera, then sync fixing for picture and sound is just how it is!
 
You must log in or register to reply here.
Back
Top