Converting non-drop frame audio to drop frame

[Patryk_Rebisz]

Camera recorded 23.97 video while audio was recorded with European non-drop frame standard thus the audio drifts a bit over time. Is there a way to convert the audio to DF?

 

[Steve House]

Camera recorded 23.97 video while audio was recorded with European non-drop frame standard thus the audio drifts a bit over time. Is there a way to convert the audio to DF?

Audio doesn't have a "frame rate" so there is no conversion possible or necessary. With modern file-based workflows, timecode does not establish a speed reference, only a single position reference similar in function to that of a conventional slate, thus the playback speed of the audio is completely independent of the timecode frame rate recorded in it. Your observed drift has nothing to do with the frame rate setting on the recorder. What caused it is the lack of a common timebase for the sync clocks in both camera and recorder leading to a slight difference in playback speed of the audio with respect to the video when they are both aligned in the editor's timeline and the sample clock rate is not influenced by the timecode settings and clock in any way. The editor is playing back the files at exactly 48kHz but the audio sample rate is close to but not exactly 48kHz. As a result the audio plays slightly faster or slightly slower than when originally recorded. Timecode in the audio file is simply an entry in the file header that records the timecode value of the first audio sample in the file and how the TC display should be updated during playback - as said, the value there has no effect on playback speed whatsoever.

FYI, there is no difference in speed between drop frame and non-drop frame. Frames aren't 'missing' in drop frame but rather frame numbers are skipped. The frames are there but the counter jumps periodically. If the frame rate is 30 FPS, the count normal goes ... 28, 29, 00, 01, 02, 03 ... but every so often it skips and goes ... 28, 29, 00, 03, 04 ... The same number of frames goes by each second but the numbers assigned to the individual frames skip.

 

[Patryk_Rebisz]

what would you suggest would be the best way of dealing with the drift?

 

[TyFord]

Hello Patryk,

Before you give up, try Wave Agent from Sound Devices. It's a free software utility. Here's a link.

Bring up one of your files and change the time code in the header. See if it works for you. Others i have suggested Wave Agent to have found it helpful.

Steve's right, though. Without a real sync, you're gambling. I'm looking at a TASCAM HS-P82 right now and it takes all kinds of sync from camera to lock the two devices together.

Please let us know what happens with Wave Agent.

Regards,

Ty Ford

 

[Steve House]

As Ty says, it's relatively easy to change the timecode frame rate noted in the audio file header and the free Wave Agent software he suggests is a good tool to use for that. The gist of my post is that changing it won't make any difference. The embedded timecode does NOT control the playback speed, hence the duration, of the audio file. Back in the analog tape and DAT days, when timecode would control playback speed at the time the original master tapes were "resolved" by being played in real time in order to transfer the audio to magnetic perf or videotape, it could have, but not with today's file-based workflows. If the audio is playing a tad too slow now, for example, so that after lining up the head of the two files in perfect sync you find the audio is drifting later and later with respect to picture as the files play out, after 'correcting' the timecode frame rate in the audio file it will STILL playback in exactly the same manner and it will still drift. You only fix is going to be a laborious process of aligning the head of the two files and either stretching or compressing the audio file manually until it's the proper length. Hope you've got slates at the start of each take! And if the drift is variable, audio sometimes leading and other times lagging video over the course of the shot, and by different amounts, lot's 'o luck.

The speed is exactly the same in drop frame or non-drop frame. The only difference between 29.97 DF and 29.97 NDF is in the frame numbers displayed in the editing software timeline, not the actual number of frames in X seconds of program. With DF the timeline reads 1 hour at program end when the prgram lasts 1 hour of real time - the time displayed on the timeline corresponds to real clock time. With NDF it doesn't. Most broadcast and theatrical is shot NDF while ENG and commercials are often shot DF to simply cutting to fit a specific time window during the edit. But regardless of how the video was shot, you can freely put it on a DF or an NDF timeline when dropping it into the editor. A TV episodic is frequently shot and edited in NDF but the timeline is switched to DF for the final rendering so the timing indicated in the embedded timecode corresponds to real clock time during broadcast.

BTW, to the best of my knowledge there is no such thing as 23.98 DF. DF and NDF refers to standard NTSC 29.97 FPS video.

 

[Richard Crowley]

My way of dealing with drift:
1) DO NOT attempt to "stretch" or "compress" (in time) audio. Leave it as it came from the recorder. Unless you are very careful and using high-quality tools, you may end up with artifacts MUCH WORSE than a slight time problem.
2) The audio track is the MASTER REFERENCE. The video is SECONDARY to the audio track. At any point where there is a video transition (cut to another camera, insert B-roll footage, insert presentation slides, etc. etc.) you can "pull-up" the video to sync with the audio. It isn't difficult to do this manually, but there are now software tools (Plural Eyes, etc.) that will do this automatically for you.

A single, unbroken shot that runs for many minutes is deadly dull and something we try to avoid, so there should always be opportunities to break up the video to keep it interesting. And even if you DO have a long, unbroken shot adding or dropping frames is typically much less disturbing (or even perceptible) to the average viewer than attempting to diddle with the audio.

Remember that DF or NDF is only a numbering scheme. It is NOT a sampling rate.

 

[Noiz2]

My way of dealing with drift:
1) DO NOT attempt to "stretch" or "compress" (in time) audio. Leave it as it came from the recorder. Unless you are very careful and using high-quality tools, you may end up with artifacts MUCH WORSE than a slight time problem.
2) The audio track is the MASTER REFERENCE. The video is SECONDARY to the audio track. At any point where there is a video transition (cut to another camera, insert B-roll footage, insert presentation slides, etc. etc.) you can "pull-up" the video to sync with the audio. It isn't difficult to do this manually, but there are now software tools (Plural Eyes, etc.) that will do this automatically for you....

First I would disagree with some of the above. If your working in film or video the IMAGE is your master reference. Two reasons, first that is the post workflow pretty much everyone except maybe music videos do (good luck telling the director and picture editor that sound is going to adjust the picture edits). Second picture only moves in full frames (1/24 of a second in your case). People can sense a quarter frame off sync, you can't move picture a 1/4 frame. You can move sound by a sample (generally 1/48,000 of a second).

Music is an exception because you can't go cutting and moving music with out serious problems.

Also you may well be able to do full clip stretches etc to get the timing back. With decent software and such a small change you are not likely to pick up any ugly artifacts. Again music is generally an exception.

---------******************----------------


To your original post. While it is true that sound records at a sample rate and video at a frame rate and the two run along happily and will usually stay in sync for quite some time, some software WILL act like frame rate relates to sound speed.

Final Cut Pro is notorious for changing the speed of double system sound because of a session and media frame rate conflict. I don't know if this is fixed in the latest and (greatest?) FCPX but was certainly a problem with FCP6 and I believe also with FCP7. Apple is VERY slow to fix things they don't think should be a problem even if everyone using the app thinks it's a problem so I doubt it's fixed in FCPX but?

If your on FCP then there is something you can and probably should do to "fix" your footage before you start editing. You need to do it before because once you have a FCP project that is set up "wrong" there is no way to "fix" it. The problem comes up in systems that have been used or are defaulted to one frame rate and then you create a project with out changing ALL of the preferences to the new frame rate BEFORE you create the project.

Once you have a satanic FCP project it will sample rate convert all audio brought into the project on the fly in the background with no warnings or alerts.

If this sounds like you drop me a line and I can send you a link to help.

 

[Richard Crowley]

First I would disagree with some of the above. If your working in film or video the IMAGE is your master reference. Two reasons, first that is the post workflow pretty much everyone except maybe music videos...

Yes, you are correct. I should have said that I do mostly music productions where the audio track is more important than when doing mainly speech/dialog driven work.