• Wondering which camera, gear, computer, or software to buy? Ask in our Gear Guide.

Audio drifting out of sync

So my DSLR footage is 23.976fps, with the onboard audio track at 48khz. External audio was recorded at 44.1khz

The video project is Adobe Prem Pro CS4, with project settings matching the DSLR footage. (23.976 @ 48khz audio) I've resampled the 44.1khz external audio to match the project settings, but it's still drifing out of sync the longer it plays.

Any ideas? :)
 
More than usual? And what kind of drift are you experiencing (i.e. how many frames off over how much time)?

And which recorder are you using?

Non-sync audio for DSLR is going to drift over time, simply because there is no shared master clock between the recorder and the camera. Some recorders drift worse than others. This is dependent on the timing crystals in each device, and how close to 48kHz they are actually running. Hint: neither is running at the exact 48kHz that you may think they are, and neither is running at the same offset as the other.

And even though your external recorder was set to 44.1k, it wasn't a perfect 44.1. The conversion to 48k didn't fix the issue, and may have made it worse. Always record audio for video at 48k.
 
Side note: digital audio isn't at all dependent on frame rate. Recorders that can utilize TC are merely putting a meta stamp at the head of each recorded file for reference. Sample rate is what matters here.
 
The external recorder was a Zoom H4 (not the n).

Over 9 minutes of video, the audio becomes a good 1.25 seconds out of sync. Meh. :(

Ah well. Time to start chopping & resyncing every 20 seconds. :lol:

.
 
Side note: digital audio isn't at all dependent on frame rate. Recorders that can utilize TC are merely putting a meta stamp at the head of each recorded file for reference. Sample rate is what matters here.

Yup, knew that. Was just mentioning that as part of the whole thing. Thinking out loud. :)
 
Unless you want to have a really long shot of an actor delivering a monologue over 20 seconds, then it probably won't be a problem since almost every sentence of dialogue in a movie does not go on for 20 seconds. Unless I missed something :)
 
Unfortunately the H4 seems to be particularly bad about drifiting - the H4n is much better, and doesn't have any appreciable drift in my tests over the typical 12-minute clip limit.

Hmm, well that's no fun. Would I get better results with an H2? I have one of those hangin' about. I just don't know much about the performance of these devices.

Regardless, I have 9 full episodes recorded at 44.1 now, with the H4. There's much less drift after I resample to 48khz in Soundbooth (as opposed to letting Prem Pro just render the audio), but being able to get a no-drift match would be awesome. This is pretty much a one-camera talking-head show reading from a teleprompter, so being able to sync with no issues... well, yeah.
 
Hmm, well that's no fun. Would I get better results with an H2? I have one of those hangin' about. I just don't know much about the performance of these devices.

I'm not a fan of Zoom, though I have yet to get my hands on the H6. I had an original H4 and liked it for what it was, but it bit the dust about 2 years in. Used as a line-level recorder from a mixer, it did just fine, but drift was an issue.

Handheld recorders in general are going to have some of the worst issues with drift. I now keep a DR-40 in my kit, but would never use it as a primary recorder and especially not for long takes unless i had no alternative. It also drifts quite a bit. Part of keeping costs down for these little devices is using cheaper parts, and the timing crystals are part of that.

Move up to something better. I recently ran a Fostex FR2-LE for over an hour straight on a live event, non-sync with multiple cameras, and experienced no discernible drift. I was honestly surprised at the results, but also have never had issues over the 12-minute limit of my DSLR when I've recorded interviews and the like.

Regardless, I have 9 full episodes recorded at 44.1 now, with the H4.

Live and learn. Always, always, always use 48k. (Unless, of course, you're recording something that will be subject to heavy manipulation in audio post, in which case 96k or even 192k will come in handy.)
 
Last edited:
Thanks for extra infos. Man, I wish this project had an audio guy on it. :lol:

The simplest solution for the future (still have another 9 episodes to shoot) would be to feed the wireless lav straight into the camera then, it seems. I don't see any better gear materializing on this project. :no:
 
A lot of the fix articles for the H4n may not apply to the H4 - there were a lot of sync issues with DSLRs and FCP which weren't due to the recorder but rather the default project settings; get the settings right and there's no appreciable drift with the the H4n. The H4, on the other hand, seems to drift badly internally, so I think the only options are to either cut/slip or retime your audio.

Handheld recorders in general are going to have some of the worst issues with drift. I now keep a DR-40 in my kit, but would never use it as a primary recorder and especially not for long takes unless i had no alternative. It also drifts quite a bit.

In my experience the DR-100 seems to do well without much drift, while the zoom H1 drifts pretty badly (similar to the H4). Clearly it can vary quite a bit within a given manufacturers product line up, but it appears to improve as you step up from their entry level to the mid-range. I suspect it probably varies from unit to unit as well though, so it may be worth making a sync test the first thing you do if you get any of these recorders.
 
^ Good points.

I use a Tascam DR-40 and have never had any drift issues. Have synced 2 features and 2 shorts to date without issues.

Going forward, may be best for you to swap out the H4.
 
Side note: digital audio isn't at all dependent on frame rate.

In theory you're right, in practise it doesn't always work out that way! It depends what is being done, by whom and with what software.

Over 9 minutes of video, the audio becomes a good 1.25 seconds out of sync.

Mmm, that really is excessive! With 9 minutes of video I would expect no more than about 10 frames of drift, even with a poor quality recorder, and roughly half that (or less) with a decent recorder. It sounds to me that there could be an issue beyond an inaccurate clock in the H4, although I've never used a H4 and I suppose it is possible that it's internal clock really is that atrocious. But an alternate explanation could be, for example, an audio pull up/down between video speed and film speed, which over a 9 min clip would account for just over half a second of drift. Is it possible that your software is using the frame rate of your recorder's audio and has applied a pull up/down (or an equivalent time stretch/compression) on import and/or when you sample rate converted to 48kHz? Two of these audio pull ups/downs would account for about 1.1 secs of drift. It seems fairly unlikely you have inadvertently applied 2 pull up/down operations but it would be worth checking. Try to edit the metadata of the audio file and change to the same frame rate as your footage, convert the audio file to 48kHz (using different software to what you used to convert previously) and then import this into your NLE and see if there's any difference.

If that doesn't solve the problem try; Edit and line up the start of your recorder's audio file (to the clapperboard) and do the same with your camera's audio file. Next, edit the end of the recorder's audio file as close as you can (by ear) to the same audio point as the end of your camera audio. Now apply a time compression/expansion on this edited recorder's audio file to match the duration of the camera audio file. You might be lucky and your recorder's audio file may now be in sync with your camera's audio. If it works, make note of your time compression/expansion ratio setting, which might work for your other clips as well. A bit of a pain but much quicker than manually slicing and dicing all your dialogue, if it works of course! Watch out for time compression/expansion artefacts though.

If this doesn't work, your last option is as you have suggested and to start editing away. Vocalign or some similar program like Plural Eyes might help speed things up a little and make the results a bit more accurate, as you've got camera audio as a reference.

I'm certain you already know, but for the benefit of others, just to emphasise what AcousticAl said:
1. Always record audio at 48kHz, and
2. There will always be drift between an audio recorder and a camera! The only question is how much drift and at what point your editing equipment and your powers of observation make it noticeable. The ONLY way of avoiding this and guaranteeing sync between camera and audio recorder is to lock them both to the same masterclock.

G
 
Last edited:
Back
Top