Another sync issue: Sony PCM-D50 to 25fps

youngindiefilms

youngindiefilms

Well-known member
This is the first time I shot a interview where I recorded audio separately. Has been shot with Canon 7D and audio on a Sony PCM-D50. Audio was recorded at 48khz, 24bit. Video shot at 25fps as the final output is intended for PAL countries.
At the time of recording I tried to find a setting in the Sony to sync for 25fps, but there is nothing to choose from.

Once matched up in FCP the audio goes out of sync after a while, what is the best way / program to get it in sync?
 
Video frame rates have nothing to do with audio sync. Your video camera and audio recorder are not using the same clock, and even if they both are good units, the speed match is never perfect (or you are extremely lucky). The only way to have perfect sync on long takes is to lock both units to the same wordclock. With cheapesh prosumer gear that is not possible.

There are tools for stretching/shrinking the audio in post. Do either that or just make slight cuts in the video using inserts making it longer or shorter as needed (hope you shot those, or editing a one-camera one-take interview is next to impossible), or splice the audio so that you use the same bit twice in some quiet parts or take out small sections where unnoticeable.
 
Are you sure about that?

I found this thread here:
http://www.gearslutz.com/board/post...o-shot-23-976-fps-audio-rolled-24fps-48k.html

And someone suggests to export the FCP timeline to look for "NTSC" true/false... I found that there is a 25 which relates to the framerate of my shot video and there is a 24 which relates to the external audio track. That would suggest that the Sony PCM has a 24(or23.976?) framerate as timebase!?

NTSC video plays back slower than PAL, hence the audio is also slower.

I have enough inserts to fix the audio by splicing up the clips.. but I think stretching will be in order since the occasional part of the interview is long and audio will run out of sync...
was just wondering if there is any dedicated program/plugin or a good way to fix this in either FCP or Soundtrack Pro maybe.
 
youngindiefilms said:
Are you sure about that?

I found this thread here:
http://www.gearslutz.com/board/post...o-shot-23-976-fps-audio-rolled-24fps-48k.html
Click to expand...

That thread is full of a bunch of bad information (except for a post from user Georgia). As Petrus pointed out, audio does not play by frame rate. Yes, there are machines that can record TC with the audio, but that is nothing more than a reference... and a sync tool when working double-system with camera and audio that can be jammed together.

Sample rate is what matters with audio (well, bit rate matters but for different reasons). And what you face is audio and video that may both be recorded at "48kHz" in theory, but in practice one may be 48.098kHz while the other may be 47.998kHz. Those numbers are arbitrary, by the way, the point being that either or both devices may be recording at a slightly imperfect sample rate.

The other term for this is a difference in clock speed. Each device, camera and audio recorder, is using its internal clock. Which variance on 48kHz it uses depends on the manufacturer. The only way to avoid this during production is to use equipment that can, as Petrus mentioned, use a common master clock.

Put audio recorded at 48.098kHz against a 48kHz sequence and the audio is going to lag after a while. Audio recorded at 47.998kHz in the same sequence will play faster and thus lead the video after a while. You can adjust playback speed of the audio to time comrpress/expand it as needed.
 
OK. So in short, there is no standard formula or easy fix other than fuzzing around with stretching and cutting/realigning to fix it. :(
 
youngindiefilms said:
OK. So in short, there is no standard formula or easy fix other than fuzzing around with stretching and cutting/realigning to fix it. :(
Click to expand...

Pretty much. It's just part of working with consumer/prosumer gear. You can get amazing results for low cost, but it does take some elbow grease in post.
 
I never had that issue before, only once I got rid of my HVX and started doing interviews and other stuff with 5D/7D I became aware of this problem. Figure I will get myself a "proper" video camera again in the near future and run audio from my external mics directly to the cam, problem solved :)
 
youngindiefilms said:
OK. So in short, there is no standard formula or easy fix other than fuzzing around with stretching and cutting/realigning to fix it. :(
Click to expand...


If you do not have a 'common sync' mechanism, then yes there is no 'standard formula'.

Basically here's the situation. The 'crystal' in the audio device, generates the time base for the audio sample rate. The engineers have chosen a device that has some 'tolerance', that is between each unit there could be say 100 parts per million difference in the fundamental frequency generated by the crystals in the units.
So, even with two audio recording devices, recording the same material, you can see 'errors' in synchronization after 'a while'...

The same story is true for the camera manufacturer.

So, with out a common 'sync' pulse keeping both (or more) pieces of equipment in Lock-Step, they were eventually end up unsynchronized. The error between the two will be less noticeable for 'short' takes, and generally a frame time's worth of difference between the audio track and the visual track can be spotted by viewers.

For 25 fps that would be about 40 ms per frame, for an audio sample rate of 48K, that would be aobut 2000 samples... after about 5 minutes at 48K samples per second, there's about 14 Million samples, so '2000' samples 'off' only represents an error of about 0.013 percent... and would be probably with in the 'tolerance' set by the consumer device engineers...

The 'fix' is, to stretch the audio slightly, or, through cutting, make the 'resynch' points less noticeable.

Even for documentaries, having someone talk for more than a 15-20 seconds, before a 'cut' is rather 'long'. Often even when one person is talking there will be cut-aways to other shots, so you can 'tighten' up the audio as needed when the speaker, and their lips, are again seen on screen.
 
youngindiefilms said:
I never had that issue before, only once I got rid of my HVX and started doing interviews and other stuff with 5D/7D I became aware of this problem. Figure I will get myself a "proper" video camera again in the near future and run audio from my external mics directly to the cam, problem solved :)
Click to expand...

If you're recording straight to the camera, a la your HVX, this issue is non-existent because the audio is recorded to the same device using the camera's clock. But don't let the presence of this sync drift be your deciding factor about replacing your DSLR.

The choice between a DSLR and a true video camera comes down to the tool you need to tell your visual story. How you capture the audio part of that story is just as important, but a camera is a visual tool first and foremost. If the DSLR gives you the visual tools you want/need, then there are ways to get the audio you need as well (though, as we've discussed here, they may take some post-production work to sync).
 
All the above is good info but I think your problem is the 25fps and FCP. You need to have FCP set up correctly for a X.0 FPS or you will get really bad drift. The drift that has been talked about above should be VERY small unless something is broken, like a frame or two in an hour.

FCP has this issue that I am fuzzy on the exact solution for. If I remember correctly what causes it is having your project set to a X.0 frame rate and your capture settings set to a NTSC frame rate.
What ever the cause what happens is that FCP "assumes" things about your work flow that aint so and changes the speed of your sound to "fix" the problem. You can find a lot of talk about the issue if you Google 5d Mrk II audio sync problems. It showed up with the Mrk II's and their 30FPS video.

Sound has no frame rate but picture (NTSC land) has two major "speeds". Video speed and film speed. When you shoot on one and then move to the other your speed changes and FCP is trying to "fix" that problem. YOUR problem however is that you havent changed speeds but FCP thinks you have. If you change the FPS # you found in the sound files to 25 you might get FCP to leave it alone. If you haven't really started editing you might try to set up a new project and make sure ALL the settings are set to PAL 25 FPS, save it and then try bringing in the video and audio and see if they stay in sync.
You can also try Sebsky's tools (not sure I spelled that right?) you can find it on my blog in the post about the 5D Mrk II. You can use it to "wrap" your sound files in an appropriate frame rate Quick Time wrapper and that is supposed to get FCP to leave it alone.

www.scottkouesound.com
 
C2V said:
If you're recording straight to the camera, a la your HVX, this issue is non-existent because the audio is recorded to the same device using the camera's clock. But don't let the presence of this sync drift be your deciding factor about replacing your DSLR.

The choice between a DSLR and a true video camera comes down to the tool you need to tell your visual story. How you capture the audio part of that story is just as important, but a camera is a visual tool first and foremost. If the DSLR gives you the visual tools you want/need, then there are ways to get the audio you need as well (though, as we've discussed here, they may take some post-production work to sync).
Click to expand...

I know the problem doesn't exist when recording directly to camera :)
It's not the only reason I want to shoot with a proper camera again, the DSLRs simply lack so many features, dedicated controls and a better formfactor.
I shoot a number of different things from corporate to films, have been shooting all that stuff with DSLRs in the past year and was much more painful and more work in the end than shooting with my HVX or Z1 for example.

Noiz2 said:
All the above is good info but I think your problem is the 25fps and FCP. You need to have FCP set up correctly for a X.0 FPS or you will get really bad drift. The drift that has been talked about above should be VERY small unless something is broken, like a frame or two in an hour.

FCP has this issue that I am fuzzy on the exact solution for. If I remember correctly what causes it is having your project set to a X.0 frame rate and your capture settings set to a NTSC frame rate.
What ever the cause what happens is that FCP "assumes" things about your work flow that aint so and changes the speed of your sound to "fix" the problem. You can find a lot of talk about the issue if you Google 5d Mrk II audio sync problems. It showed up with the Mrk II's and their 30FPS video.

Sound has no frame rate but picture (NTSC land) has two major "speeds". Video speed and film speed. When you shoot on one and then move to the other your speed changes and FCP is trying to "fix" that problem. YOUR problem however is that you havent changed speeds but FCP thinks you have. If you change the FPS # you found in the sound files to 25 you might get FCP to leave it alone. If you haven't really started editing you might try to set up a new project and make sure ALL the settings are set to PAL 25 FPS, save it and then try bringing in the video and audio and see if they stay in sync.
You can also try Sebsky's tools (not sure I spelled that right?) you can find it on my blog in the post about the 5D Mrk II. You can use it to "wrap" your sound files in an appropriate frame rate Quick Time wrapper and that is supposed to get FCP to leave it alone.

www.scottkouesound.com
Click to expand...

That's a good point. there is something fishy going on with FCP when you import the files. I had made sure that all settings are to PAL 25fps when I setup the project but looks like the audio got imported on a 24fps base.

Anyway, thanks guys for all of your comments and ideas, it's not like I'm helpless on the issue, just wanted to make sure if there wasn't a simple fix to this problem before I start doing it the hard way.

My docu is pretty straight forward, interviews with lots of inserts so I easily can re-align the audio and fix it that way.

Going to shoot a short on film in the near future, maybe I'm going to invest in a crystal sync device for the camera after all. Seems to be a good idea now.
 
Last edited:
Noiz2 said:
FCP has this issue that I am fuzzy on the exact solution for. If I remember correctly what causes it is having your project set to a X.0 frame rate and your capture settings set to a NTSC frame rate.
What ever the cause what happens is that FCP "assumes" things about your work flow that aint so and changes the speed of your sound to "fix" the problem. You can find a lot of talk about the issue if you Google 5d Mrk II audio sync problems. It showed up with the Mrk II's and their 30FPS video.

Sound has no frame rate but picture (NTSC land) has two major "speeds". Video speed and film speed. When you shoot on one and then move to the other your speed changes and FCP is trying to "fix" that problem. YOUR problem however is that you havent changed speeds but FCP thinks you have. If you change the FPS # you found in the sound files to 25 you might get FCP to leave it alone. If you haven't really started editing you might try to set up a new project and make sure ALL the settings are set to PAL 25 FPS, save it and then try bringing in the video and audio and see if they stay in sync.
You can also try Sebsky's tools (not sure I spelled that right?) you can find it on my blog in the post about the 5D Mrk II. You can use it to "wrap" your sound files in an appropriate frame rate Quick Time wrapper and that is supposed to get FCP to leave it alone.
Click to expand...

I wonder if the issue you're talking about might be this one: How Final Cut and Avid Import BWF Time Code Metadata

However, I can't imagine the audio recorder in question was stamping framerate or timecode. Having said that, one can always check and prep files for import by using Sound Device's free Wave Agent software to "conform" audio metadata to any project spec. They're still letting anyone download it for free whether you're an SD user or not. And taking the few minutes to add framerate data to audio files so that it matches the project/sequence settings can never hurt. Provided you pick the right frame rate for FCP!

I noticed that particular "tech note" when they issued an update for my 552 which stopped it from stamping 30NDF as the default frame rate in the audio metadata when no external timecode was detected. Why on earth they chose 30NDF still puzzles everyone who uses the thing but that's neither here nor there.

The problem reared its head at the time I first started using the 552. When I had dig up a backup for some audio fix in post and it would quickly go out of sync with the 23.976 project being worked on. The issue appeared again on the next 29.97 project - though that was more insidious since it was upredictable and much less obvious.

It didn't make sense since we were doing single system sound with the 552 recording a backup and no timecode (since the HVX200 doesn't have a TC input to begin with!). When that firmware fix was released it all became clear. I now prep all files in advance of bringing them into FCP or sending them off to another editor.
 
Last edited:
I am pretty sure that's what's going on. I've exported one of the sequences to XML and checked it with an editor and turned out that Final Cut assigns the audio a 24fps as base.
Have to finish up some more urgent work now and then I'm all over it!
 
Finally had time to use Wave Agent, it's a pretty useful tool and free on top.
I tagged my audio files with 25fps and re-imported them to the FCP project, unfortunately it still goes out of sync but not as bad as it was before.
 
This synch problem seems to be common. We had a problem with an LS-10 and GH1 not agreeing about time ;-). But I wanted to post this to say that anybody dealing with this might want to check out the free (in every sense of the word) and quite wonderful SoX (Sound eXchange). It's a command line utility that's billed as the "swiss army knife for audio". If you're not comfortable with the command line then at least on Mac I used to use (before I switched to gnu/linux) soxwrap, which was a nice gui "wrapper" for sox. But the command line way to use sox is straightforward in fact. Hopefully this helps some people. By the way there's also lbsndfile, with a command line "front end" called I think sndfile-convert, but I haven't used it. It also looks good from the documentation anyway, like sox.
 
You must log in or register to reply here.
Back
Top