-
Manually Correcting for Audio Drift: what’s your favorite way?
Much of drift correction today seems to be “under the hood”. Plural Eyes has Automatic Drift Correction, or it can correct for drift when it gets above a certain threshold. FCPX and Premier Pro can sync based on audio when importing. It seems like these programs are just estimating corrections??
To my knowledge (at the moment) the only way to accurately correct for audio drift is manually. So far I’ve found two methods that work.
FCPX: click on externally recorded audio , hit “Command R” and time stretch transients to camera audio. Easy!
Adobe PP: export sequence to Audition, Toggle on “Global Clip Stretching”, click on externally recorded audio, time stretch transients to camera audio. With this method you have to export to Audition, render (High Quality) in Audition, save time stretched clip, and import back in to Premier Pro. Wow – much more time!
FCPX works great for audio stretch, but sometimes I cannot use FCPX because it does not see my footage.
How do you AVID folks do it? I just searched the 2018 Media Composer Manual for “drift” and “time stretch” with no results.
Can anyone suggest better or alternative workflows?
-
4 Replies
-
I NEVER stretch the audio track. The audience can hear artifacts from diddling the audio MUCH more than they can detect the odd duplicated (or dropped) frame of video. And, of course you can pull-up the sync at each edit point, so you rarely have to diddle the picture track, either.
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder. -
Hello Conrad and welcome to the Cow Audio Forum.
You raise interesting questions. Can you show us some before and after samples to let us hear if we can hear the artifacts that Richard is concerned about?
I once had a client for radio spots. He was also the voice talent and copywriter. He usually over-wrote the spots and could never get the copy done in 60 seconds. I had an Ultra Harmonizer with extra RAM and time compression. We found that for his voice and the copy, we could speed him up by 11% before his voice sounded weird. Slowing down is a bigger problem because you’re stretching the audio instead of pulling out bits and pieces. In his case, of course, we were speeding him up (without pitch change). We actually got to liking the energy from the speed boost.
Thanks!
Ty Ford
Cow Audio Forum LeaderWant better production audio?: Ty Ford\’s Audio Bootcamp Field Guide
Ty Ford Blog: Ty Ford\’s Blog -
Yes 11% sounds like a lot! I’m filming corporate events where lectures can go on for hours. The scenario I’m bringing up is when two devices are recording audio at the same sample rate at the same event. One recording device is the camera audio; the second is the audio mixing board. I’m adjusting the mixing board audio to match the camera audio. Once matched, I delete the camera audio and use the audio from the mixing board.
I know that this sounds like a ton of trouble, but the mixing board audio sounds so much better!
- BLUE AUDIO is CAMERA AUDIO
- GREEN AUDIO is EXTERNAL AUDIO
- 75 minutes at 30 FPS (using 30 instead of 29.97 just to make this easier)
- 75 minutes = 4500 seconds = 135,000 frames
- External audio shortend by just over 6 frames in order to match transients in camera audio.
- ~6 frames is .0044444 percent of 135,000 frames.
[how’s my math? =\ I’m still a little fried right now]
First picture is UNADJUSTED audio.
Second picture is ADJUSTED AUDIO
When the adjustment is rendered at High Quality, I cannot hear any artifacts, which makes sense give the percentage is so small. I find it super interesting that Adobe Audition does not even show this percentage in the ADJUSTED audio file. Audition shows 100% in both the unadjusted and adjusted files. I should point out I am not worried about artifacts since I do not see drift going over 1%. I’m just looking for alternative workflows. I’m also just curious how other folks compensate for tiny audio drift between two devices that simultaneously recorded a single event.
Just as Creative COW will not approve of uploading copyrighted images, I believe it’s the same for copyrighted audio files. Indeed I do not own the rights to the audio, so I’m not posting it. I hope the screen shots help illustrate the workflow I’m describing.
-
In my many years of futzing with audio for video – I’ve found that 99 times out of 100 audio drift comes from sample rate mismatches between the camera and the timeline.
If you simply take the source audio file and drop it on a utility like MediaInfo – note the parameters of the recording (taking special note of the recorded sample rate) and set up your video project to match that — drift disappears.
If the field audio was done with setting far from the digital video sampling norms of 48 or 96Khz, (e.g. 32khz) then you MIGHT need a transcode step on the incoming files if your software of choice can’t do the repair automatically. But I find that rare. The software folks were aware of the typical sample rates and have algorithms in place to handle most all of them.
There are times, of course, that something is TRULY strange – like an old audio recorder that doesn’t have even crystal sync or where bad design allows truly drifting timekeeping – and in those cases, the ONLY real solution is manual re-timing. But thank god I haven’t seen that type of situation in twenty years at least.
Just my thoughts.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery.
Reply to this Discussion! Login or Sign Up