Odd sync glitch

[Sean McCoy]

As mentioned in the Windows 8 thread, I experience a strange, intermittent glitch during some sessions this week, on a very fast PC with an RME Fireface running under W8. When we were punching in stereo keyboard parts across pairs of tracks, every once in awhile the right channel would slip slightly out of sync with the left. This resulted in some interesting changes of dramatic stereo phasiness. Kind of exciting, but obviously undesirable. :rolleyes:

Upon getting the files back to my studio, I found the second channel was about 250 samples behind. Easily fixable by just visually sliding into place, but weird nonetheless.

Has anybody ever experienced anything similar?

[Ian Alexander]

As mentioned in the Windows 8 thread, I experience a strange, intermittent glitch during some sessions this week, on a very fast PC with an RME Fireface running under W8. When we were punching in stereo keyboard parts across pairs of tracks, every once in awhile the right channel would slip slightly out of sync with the left. This resulted in some interesting changes of dramatic stereo phasiness. Kind of exciting, but obviously undesirable. :rolleyes:

Upon getting the files back to my studio, I found the second channel was about 250 samples behind. Easily fixable by just visually sliding into place, but weird nonetheless.

Has anybody ever experienced anything similar?
I'm quoting to aid those using threaded view.

Did the keyboard player lean to the left on punches?

[Dave Labrecque]

As mentioned in the Windows 8 thread, I experience a strange, intermittent glitch during some sessions this week, on a very fast PC with an RME Fireface running under W8. When we were punching in stereo keyboard parts across pairs of tracks, every once in awhile the right channel would slip slightly out of sync with the left. This resulted in some interesting changes of dramatic stereo phasiness. Kind of exciting, but obviously undesirable. :rolleyes:

Upon getting the files back to my studio, I found the second channel was about 250 samples behind. Easily fixable by just visually sliding into place, but weird nonetheless.

Has anybody ever experienced anything similar?

Sean -- not that I have anything helpful to say, but... do I understand right that the keyboard parts being punched were two-tracked, rather than single-track stereo?

[Sean McCoy]

Sean -- not that I have anything helpful to say, but... do I understand right that the keyboard parts being punched were two-tracked, rather than single-track stereo?
Yes. And I'm having a hard time figuring out why I did it that way. :(

[Dave Labrecque]

Yes. And I'm having a hard time figuring out why I did it that way. :(

The weird thing is that if you'd done it that way you either wouldn't have had the sync issue OR you would've, but just not noticed. :eek: Makes me wonder how often such sync issues might happen and we just don't know. :confused:

(Of course, a third possibility is that you still would have a had a sync issue between the L and R streams -- within the resulting interlaced stereo file. No. No, that couldn't happen. Could it?)

Hmph. Now I'm wondering if you had a sync issue with both L and R tracks, but you only noticed the inter-track differential. Man, am I a worry wart or what? :o

[Sean McCoy]

The sample delay on one side is very obvious as it creates a clearly unnatural, out of phase stereo signal. I suppose it could be difficult to detect any musical sync issues with a 250 sample delay, but if it were consistently on all SRP recordings that would be a different issue. While I've had to adjust the loopback latency settings before, I've never experienced mismatches between tracks.

[Dave Labrecque]

I've never experienced mismatches between tracks.

But, like you say, how would you notice something so small between tracks without common content?

[Sean McCoy]

But, like you say, how would you notice something so small between tracks without common content?
Well, I'd certainly notice them if they were on say, adjacent drum tracks. And in thinking about this further, 250 samples is more than enough to create noticeable timing issues. I have my system set with 192 samples of compensation because the discrepancy was enough to throw off even vocal overdubs. I'll need to do a latency test in that studio, which I haven't done. But even if I discover the need for compensation, it still wouldn't explain the offset between tracks when punching in.

[Dave Labrecque]

Well, I'd certainly notice them if they were on say, adjacent drum tracks. And in thinking about this further, 250 samples is more than enough to create noticeable timing issues. I have my system set with 192 samples of compensation because the discrepancy was enough to throw off even vocal overdubs. I'll need to do a latency test in that studio, which I haven't done. But even if I discover the need for compensation, it still wouldn't explain the offset between tracks when punching in.

Right. I'm just saying you noticed this because it was between two tracks with common content. What if it's happening to other punch-ins, when you wouldn't notice? It could apply to us all. Or are you saying you think you would notice? Or maybe if you wouldn't notice, it wouldn't therefore matter? Hmmm...

[Sean McCoy]

Right. I'm just saying you noticed this because it was between two tracks with common content. What if it's happening to other punch-ins, when you wouldn't notice? It could apply to us all. Or are you saying you think you would notice? Or maybe if you wouldn't notice, it wouldn't therefore matter? Hmmm...
During this same session, we did a couple of punchins of the entire drumkit. A 250 sample offset on any one of the ten tracks would have been obvious, so I was lucky in those cases. Anything is possible, of course—especially when we're dealing with bizarre occurrences that are obviously being perpetrated either by aliens or Russia.

[cgrafx]

I'm curious as to why you have needed to add compensation manually.

I've done lots of recording sessions with lots of punch ins.

I've never had to do anything with delay compensation.

Generally that is only needed if you are connected to an external piece of processing gear and it seems wouldn't be needed during basic tracking and punch-ins

[Sean McCoy]

I'm curious as to why you have needed to add compensation manually.

I've done lots of recording sessions with lots of punch ins.

I've never had to do anything with delay compensation.

Generally that is only needed if you are connected to an external piece of processing gear and it seems wouldn't be needed during basic tracking and punch-ins
I'm not doing it manually, except for this particular problem in this session done in a remote studio where the adjacent tracks were sometimes out of sync. But in my studio, using an RME Digiface and a very fast core i7 system, I have to set the Record LoopBack Latency Adjust to 192 samples to get perfect sync during punchins. I've never been clear on why that is, but making the adjustment is pretty simple. Interestingly, using the same hardware in XP, the adjustment was substantially higher than it is in Win 8.

[Carl G.]

During this same session, we did a couple of punchins of the entire drumkit. A 250 sample offset on any one of the ten tracks would have been obvious, so I was lucky in those cases. Anything is possible, of course***8212;especially when we're dealing with bizarre occurrences that are obviously being perpetrated either by aliens or Russia.

Delays in punch in are an obvious effect of the buffer settings.
They're small differences... but so are phase differences.

It seems like you have something patched (or somehow ghosting) in one of the channels that has a tad latency on it, causing the minor phase shift when you have stereo using two channels instead of one (stereo).

Have you tried a quick test on a fresh new EDL (with no plugins on any tracks).
Then add one at a time of the plugins you have.
Or... how about just isolating those two tracks to see the phase difference.... then try coping those to two other tracks... is the difference still there?
Perhaps the difference was actually a mic phase placement problem on just those two mics???? (recorded that way - maybe Ian was right)

At worst case... spread the track width apart in the mix... and let the end user worry about it (yep...problems in their stereo!) :)

[Carl G.]

I'm quoting to aid those using threaded view.

Did the keyboard player lean to the left on punches?

:)
No.... the punches didn't phase him a bit.

[UpTilDawn]

...Has anybody ever experienced anything similar?

Seems to me that the only time this kind of thing ever happened to me was when the two channels in question happened to be split between two soundcards.... the soundcards either lost sync through some quirk (maybe as a result of the punch-in process and your latency settings in your case), or were not sync'd properly to begin with.

Could this have been the case?

I've never needed to use latency adjustment settings in the rare punch-in sessions I've done in SAW. I'm not sure I understand why you need to do this in your case. Would you mind explaining?

It's been a couple of years since I last had to do a punch-in session (even for looped/layer dubbing sessions), so sorry for my ignorance on the matter.

__

[Dave Labrecque]

During this same session, we did a couple of punchins of the entire drumkit. A 250 sample offset on any one of the ten tracks would have been obvious, so I was lucky in those cases. Anything is possible, of course—especially when we're dealing with bizarre occurrences that are obviously being perpetrated either by aliens or Russia.

Aliens or Russia?

[Dave Labrecque]

Delays in punch in are an obvious effect of the buffer settings.
They're small differences... but so are phase differences.



Actually, my understanding is that SAW handles all the timing and buffering so that you get sample-accurate punches and SRP in general regardless of in or out buffer settings. Otherwise, we'd be moving stuff back into sync all the time after recording, which I never do.

[Sean McCoy]

Delays in punch in are an obvious effect of the buffer settings.
They're small differences... but so are phase differences.

It seems like you have something patched (or somehow ghosting) in one of the channels that has a tad latency on it, causing the minor phase shift when you have stereo using two channels instead of one (stereo).

Have you tried a quick test on a fresh new EDL (with no plugins on any tracks).
Then add one at a time of the plugins you have.
Or... how about just isolating those two tracks to see the phase difference.... then try coping those to two other tracks... is the difference still there?
Perhaps the difference was actually a mic phase placement problem on just those two mics???? (recorded that way - maybe Ian was right)
No, this was a fresh, raw recording session with no plugins or effects, and the problem was intermittent.

At worst case... spread the track width apart in the mix... and let the end user worry about it (yep...problems in their stereo!) :)
No, this was a fresh, raw recording session with no plugins or effects added, and the problem was intermittent.

[Sean McCoy]

Seems to me that the only time this kind of thing ever happened to me was when the two channels in question happened to be split between two soundcards.... the soundcards either lost sync through some quirk (maybe as a result of the punch-in process and your latency settings in your case), or were not sync'd properly to begin with.

Could this have been the case?

I've never needed to use latency adjustment settings in the rare punch-in sessions I've done in SAW. I'm not sure I understand why you need to do this in your case. Would you mind explaining?

It's been a couple of years since I last had to do a punch-in session (even for looped/layer dubbing sessions), so sorry for my ignorance on the matter.

__
A valid point as different interfaces and converters might have slightly different latency times, but these were all going through the same multi-channel interface.

[Sean McCoy]

Aliens or Russia?
Where's your conspiratorial spirit? :D

[Dave Labrecque]

No, this was a fresh, raw recording session with no plugins or effects added, and the problem was intermittent.

Very strange. Here's me thinking in areas in which I have no business do so: multi-core processor? Hyper-threading? Could different recording streams going through different processor threads slip out of sync from each other (and/or the reference MT audio data)?

[Dave Labrecque]

Where's your conspiratorial spirit? :D

You make it sound like aliens and Russia aren't perfectly legitimate explanations for things we can't otherwise explain. ;)

[Sean McCoy]

You make it sound like aliens and Russia aren't perfectly legitimate explanations for things we can't otherwise explain. ;)
My bad.

[Bob L]

Perhaps you actually dropped some buffers on the one chan... check all your tweak settings and also try single or dual cpu... no hyperthreading.

Bob L