Audio sync issues with Fostex FR2le?

S

sam alcoff

Member
Hey.

I got a Fostex FR2le a little while ago, but haven't used it too much because the audio people I worked with either mixed right into camera or had their own system. But Friday we went out and did some great shooting for the whole day.

Camera: HPX170, 24pn 720, Default on any audio setting other than levels.
Recorder: I think it was 48/24 Wav.
Documentary style (a lot of starting and stopping on the camera to preserve the cards, but the recorder has long clips over two hours).

It sounds great (I truly love the Sound Devices MixPre), but we have interview segments where, over twelve minutes, it starts in sync and slowly goes off sync by about a half second and is very noticeable. I think we'll be able to fix it during the editing, just resyncing every small segment, but I was curious about future shoots.
Is this a chronic issue or did I just not set it proper? If the latter, what settings should I do to keep the two in sync?
Thanks.
 
That recorder doesn't offer time code so I advise to keep a sound log and slate EVERY TIME when possible. But since it was documentary it probably was shoot and go(unless some scripted parts) I have used a Zoom H4 without TIME CODE and still have been able to later sync to picture its just a matter of taking good notes or having the script supervisor keep up with your sound file that matches the take. Hope that helps.


Nicole
 
I just got an FR2-LE and I was having the same discussion with someone here. Basically, when you don't have timecode, you should A) make sure that the khz line up or are divisible by one another (e.g. 48 on camera to 48 in recorder, 96, etc, or 44 to 44, 88, etc) and B) Keep your recording time down. I'd say anything over 10 minutes is probably going to get you into trouble (though you should be good until about 20 minutes).

I had a similar situation when I was using a professional SD cam recording at 30p for audio with my HD camcorder also shooting at 30p. Even though they should have been perfect regardless of recording, if I let it go for too long they would eventually have synch issues.

Being that you can't change what's been recorded, hopefully you can use cut-aways, etc? I doubt any kind of manipulation of the frame rate would help, and you'd then also run into an issue when converting the footage (so, even if you did change the frame rate to line up with your off-synch audio, you'd still have to settle on a single frame rate when converting to DVD, whatever).

At this point, just try to get creative in the actual edit? In the future just try not to let anyone talk for longer than 20 minutes at a time. lol
 
I have used the FR2LE extensively and the clock was always dead on. Is it possible this is a framerate or sampling rate issue? In any case, if you measure the difference in samples you can perform a simple micro pitch correction without losing SQ.
 
It seems rather surprising that these devices would slip that far apart in only 10 minutes. I would suspect that one or the other (impossible to tell which from here) may be slightly out of spec.

OTOH, 10 uninterrupted minutes of ANY shot is deadly-dull. Any place you pop in some B-roll, etc is a perfect point to "pull-up" the sync, so even with a couple of seconds after 10 minutes, I don't find it that troubling.
 
Please note, with digital file-based recordings timecode WILL NOT prevent the slippage in sync the OP has reported. Timecode establishes a single line-up point in the audio and video that is in no way any different functionally from a simple slate. All it really does is take some of the work out of the aligning the two files. With current digital video and audio recordings timecode establishes a single POSITION reference but does not provide any SPEED reference data. Timecode today is radically different from timecode in the days of analog and DAT recording in that regard. In order to prevent "slippage" over the duration of a long'ish shot, the camera and audio recorders clocks must share a common timebase. There are a variety of ways to accomplish this - with a prosumer camera that lacks genlock I/O the most common way is to feed video to a video blackburst input on the recorder such as found on the Tascam HD-P2. If the recorder doesn't have a video input, another method is to use something like a Lockit box that is jammed to the camera's video output and then sends Wordclock to the recorder. However it's accomplished, what has to happen is that the audio recorder's sample clock must be slaved to the camera's video sync clock so the two run at exactly the same speed.
 
Steve House said:
Please note, with digital file-based recordings timecode WILL NOT prevent the slippage in sync the OP has reported. Timecode establishes a single line-up point in the audio and video that is in no way any different functionally from a simple slate. All it really does is take some of the work out of the aligning the two files. With current digital video and audio recordings timecode establishes a single POSITION reference but does not provide any SPEED reference data. Timecode today is radically different from timecode in the days of analog and DAT recording in that regard. In order to prevent "slippage" over the duration of a long'ish shot, the camera and audio recorders clocks must share a common timebase. There are a variety of ways to accomplish this - with a prosumer camera that lacks genlock I/O the most common way is to feed video to a video blackburst input on the recorder such as found on the Tascam HD-P2. If the recorder doesn't have a video input, another method is to use something like a Lockit box that is jammed to the camera's video output and then sends Wordclock to the recorder. However it's accomplished, what has to happen is that the audio recorder's sample clock must be slaved to the camera's video sync clock so the two run at exactly the same speed.
Click to expand...

I recently did a shoot where the camera was recording 25 Fps. I had my Tascam HD P2 with the timecode function off. We were syncing takes with a slate. I was wondering if i should have set my tascam to record 25 Fps as well as using the slate. I have not had any comeback from post to say it can't be sync'd but i have been wondering if i had done the correct thing.
 
sam alcoff said:
Hey.

I got a Fostex FR2le a little while ago, but haven't used it too much because the audio people I worked with either mixed right into camera or had their own system. But Friday we went out and did some great shooting for the whole day.

Camera: HPX170, 24pn 720, Default on any audio setting other than levels.
Recorder: I think it was 48/24 Wav.
Documentary style (a lot of starting and stopping on the camera to preserve the cards, but the recorder has long clips over two hours).

It sounds great (I truly love the Sound Devices MixPre), but we have interview segments where, over twelve minutes, it starts in sync and slowly goes off sync by about a half second and is very noticeable. I think we'll be able to fix it during the editing, just resyncing every small segment, but I was curious about future shoots.
Is this a chronic issue or did I just not set it proper? If the latter, what settings should I do to keep the two in sync?
Thanks.
Click to expand...

You haven't said what program you have your footage in. I'm betting you are editing at 24 fps and that camera (without looking at manual) I believe shoots 29.97 or 23.976 in the 24p mode.
 
stringz said:
I recently did a shoot where the camera was recording 25 Fps. I had my Tascam HD P2 with the timecode function off. We were syncing takes with a slate. I was wondering if i should have set my tascam to record 25 Fps as well as using the slate. I have not had any comeback from post to say it can't be sync'd but i have been wondering if i had done the correct thing.
Click to expand...
Wouldn't matter as there is no such thing as "frame rate" when dealing with an audio file. The Tascam's FPS setting is the frame counter in the timecode and doesn't do anything to change the recordin g or playback speed of the file. What counts is sample rate. At 48kHz there should be 48,000 samples in each second of the recording. If the clock is a tad fast, what was recorded in 1.0000 second will have too many samples. When you play it on another device (like your editing computer) whose clock is dead on, it will take a little longer than exactly 1 second to play back. That is why sync drifts - the video clock would get 48,000 samples in 1 second while the audio clock, running a little fast, might record 48,227 samples in 1 second. Play them back together and even though both files were recorded in exactly 1 second they play back at slightly different rates and sync drifts off over time. Slaving the clocks together, which is what happens when you feed composite video or blackburst into the Tascam's video input, makes them run in lock-step together at exactly the same speed.

If you are using a timecode workflow, the Tasmcam's timecode register (which is essentially a time-of-day clock completely separate from the sample clock rate) needs to count frames at the same rate the camera does, so when the camera says this frame was recorded at 00:23:17:18, the audio recorder's head of file timestamp plus the number of samples to the audio bit recorded with that frame says exactly the same thing. So if your camera is PAL running at 25FPS, the recorder's frame rate should also be set to 25FPS. It won't effect the recording speed but it will effect the first-sample timestamp.
 
Last edited:
Steve hit pretty much everything I was going to point out about TC and modern work flow. Two hour takes on the Fostex are probably silly and you probably shouldn't be doing that BUT since the camera isn't running constantly that shouldn't affect sync, no cumulative drift. So my first guess is a clocking issue. FCP can do this "intentionally" if you use a true 24fps time line and have not set things up just right. Another common culprit is having some other digital device set to the wrong SR and having it digitally connected to a digital input on an audio device. Depending on settings etc this can have the effect of clocking your 48K audio at 44.1 even when your set to 48.
The physical clocking issues are not that hard to track down if you look, the FCP issues can be really hard because FCP does black magic under the hood and SR converts and (when it thinks it needs to) do pullu-ps/ downs with out letting the user know what or why. I know FCP has some of these issues but probably other NLE's also especially the more "user friendly" ones. What you describe (if the drift is really 1/2 sec over 10 min or so) sounds like a pull-down issue, i.e. the NLE is pulling down the audio when it shouldn't. So I would look to a setting or you have entered the FCP X.0 framerate Twilight Zone nightmare.

I don't particularly like the Fostex box but if it isn't broken and wasn't being clocked off a CD deck or something then it is very unlikely to be the problem.
 
Thanks Everyone

Thanks Everyone

OP here. Thanks for the great feedback.
One little note is, if you read my OP closely, I'm not going to have any long uninterrupted shots in the final edit, so I can just resync everything; I was just curious if they could be set in a way to prevent that in the future. Doesn't seem to be the case, but serious thanks for the insightful answers.
 
I shot my niece's wedding a few weeks ago and had 8-10 audio tracks running into my Digimax 48k's, through my 2408, and into Vegas. Normally, I clock the entire chain to one of my Digimax's because I've never had a problem with their clocks in over 10 years of use. Unfortunately, during set up, I broke the cheap wordclock in connector on the 2408 and was forced to sync to the either the 2408 or the PCI-424 card (not sure which one I chose) since I have never been able to get a lock using the ADAT litepipe connectors without experiencing digital "noise".

A few days later while dumping all of the footage from three cams into Vegas, I noticed the multitracked audio was waaay off by the end of the ~30min ceremony. Fortunately, the rate of change was consistent throughout the entire period. So it was as simple as time stretching all the audio tracks the extra 3+ seconds required to re-establish sync throughout the ceremony.

I have never had a problem like this while clocking to one of the Presonus Digimax's. But, at least the drift was stable. So I think you, like I, experienced the ramifications of variations between clocks in non-locked digital devices.

I think the solution that Steve mentioned sounds interesting for small scale outboard audio kits. But I would never trust the clock on a camcorder to serve as the master to a large multitracking rig. And it's interesting reading some of the comments about long takes. Anytime you shoot a live concert, for example, and are tracking the show's audio to an outboard multitracker, you're going to have to maintain sync throughout the whole show since there will not be any audio breaks in the finished product (at least not the way I produce most live shoots).
 
Last edited:
ajcourtney said:
...
I think the solution that Steve mentioned sounds interesting for small scale outboard audio kits. But I would never trust the clock on a camcorder to serve as the master to a large multitracking rig. And it's interesting reading some of the comments about long takes. Anytime you shoot a live concert, for example, and are tracking the show's audio to an outboard multitracker, you're going to have to maintain sync throughout the whole show since there will not be any audio breaks in the finished product (at least not the way I produce most live shoots).
Click to expand...

If you were shooting something like multicamera coverage of a concert with the audio on a separate recorder, whether multitrack or not, the pro way to do it is to use cameras that have genlock inputs, not a typical prosumer camcorder (with the exception of a few topshelf cameras like the Canon XHG1 and XLH1 that have them). You equip each camera and the audio recorder with individual Lockit boxes tuned to a common reference to provide sync, either blackburst for SD video or trilevel for HD to the cameras, and wordclock to the audio. Tuned Lockits are spec'ed to hold accuracy to within 1 frame drift in 24 hours. Just like the hard-wire distributed "house sync" in a broadcast studio, Lockits give you a common timebase for all audio and video devices on the set.
 
Last edited:
^^yeah, I looked at those little boxes several years ago. I'd say they're fine for syncing your cam's with portable field recorders, but what happens to your audio rig's WC when it fails? And I realize I'm speaking strictly on the audio side right now. That's why you buy something like an Apogee Big Ben for your audio rig. But even with that box, it appears that the SureLock feature is a FAIL based upon some experiences I've read recently, which is why I will probably pick up a Brainstorm DCD-8.
 
ajcourtney said:
^^yeah, I looked at those little boxes several years ago. I'd say they're fine for syncing your cam's with portable field recorders, but what happens to your audio rig's WC when it fails? And I realize I'm speaking strictly on the audio side right now. That's why you buy something like an Apogee Big Ben for your audio rig. But even with that box, it appears that the SureLock feature is a FAIL based upon some experiences I've read recently, which is why I will probably pick up a Brainstorm DCD-8.
Click to expand...

Lockit's from Ambient and comparable boxes from Deneke are widely used in network TV and feature film location work. They're built to professional standards (and priced to match) and while nothing is perfect, their failure rate is quite low. That technique is a staple of network broadcast television and theatrical feature production where the original is shot on video rather than film.
 
Last edited:
they are nifty little boxes for sure and that's why I took a look at them initially. My only point in this discussion was that, from the audio engineer's POV, I would never clock to one of those boxes without having a solid backup clock source because when your digital audio kit loses WC, it is a disaster.

I cannot imagine tracking a live show's 20+ stage mic's through my mic pre's and converters, looping one set of analog outs to the FOH mix, locking to someone's Black, losing WC due to a video failure, and bringing down the entire show's FOH, LOL!
 
ajcourtney, is there a reason you're not using a split snake? As someone that does a lot of FOH work, I can't imagine setting things up the way you're describing. FOH should be completely unaffected by any failure on the recording crew's end.
 
ajcourtney said:
they are nifty little boxes for sure and that's why I took a look at them initially. My only point in this discussion was that, from the audio engineer's POV, I would never clock to one of those boxes without having a solid backup clock source because when your digital audio kit loses WC, it is a disaster.

I cannot imagine tracking a live show's 20+ stage mic's through my mic pre's and converters, looping one set of analog outs to the FOH mix, locking to someone's Black, losing WC due to a video failure, and bringing down the entire show's FOH, LOL!
Click to expand...

You still have the problem of getting all the cameras and the audio kit to share the same timebase. However you go about it, the point is that you have to have the SAME clock reference for video and audio to avoid long shots drifting out of sync as they progress. In a large stage show such as you cite, you wouldn't use one of the cameras as the master, you would have a stable house clock providing sync for both audio and video. Lockits synced to the house clock would go on the cameras if they weren't cabled. On the other hand, if you were shooting a doco or feature with a small crew, audio on a cart or in a bag, Lockits on both is a viable approach to insure sync is maintained.
 
JordanBlock said:
ajcourtney, is there a reason you're not using a split snake? As someone that does a lot of FOH work, I can't imagine setting things up the way you're describing. FOH should be completely unaffected by any failure on the recording crew's end.
Click to expand...
All of the stuff I do is fairly small scale so I've never needed to split. I make it a point to work very closely with FOH, typically setting up close by, and of course prefer to use my mic's and my pre's and simply send analog outs to the house board. Those analog outs should be unaffected by any clock failure. But I don't know for sure since I've never had a clock failure while tracking.
 
You must log in or register to reply here.
Back
Top