PDA

View Full Version : What does RESAMPLE mean in the Aggregate window?


NewdestinyX
12-13-2010, 11:03 AM
What the heck does RESAMPLE mean in the Core Audio Aggregate window? When you choose one unit of 2 or more in the AudioMidi Setup of the CORE AUDIO window the other units get a checkmark by the RESAMPLE. Does that mean, more accurately, SLAVE? The term resample is misapplied in my view and confusing. If both your units are set at 48K -- there is no need for 'resampling' of any kind.

And then a follow up question.. Have any of you had any problems recording with an aggregate setup where maybe a half hour into a live recording of a bunch of tracks -- there was a loss of clock creating that 'growling' sound on all tracks. Not cliks and pops like we usually associate with clocking. Or maybe the disk slowed down or something. I've only heard this sound a few times in my career.. but it was printed on ALL tracks.. As if the whole system lost clock.. It's a growling sound. The waveforms look normal.

Thanks!

siraltus
12-14-2010, 09:06 AM
I don't know what that option exactly does - I would assume it resamples audio from slave interfaces to match the sampling rate of the master interface - the one that's set as the clock source - in case the slave interfaces are incapable of using the master's sample rate.

However, this extra math introduces a delay between the slave interfaces and the master interface causing their word clock to drift out of phase and create the digital distortion/growling sound/skips/clicks. In my case phase loss happened consistently as early as 5 minutes into playback. I turned it off and the problem was gone for good.

NewdestinyX
12-14-2010, 09:17 AM
I don't know what that option exactly does - I would assume it resamples audio from slave interfaces to match the sampling rate of the master interface - the one that's set as the clock source - in case the slave interfaces are incapable of using the master's sample rate.

However, this extra math introduces a delay between the slave interfaces and the master interface causing their word clock to drift out of phase and create the digital distortion/growling sound/skips/clicks. In my case phase loss happened consistently as early as 5 minutes into playback. I turned it off and the problem was gone for good.O SIRALTUS, Please tell me you're sure!! :-) I mean that would explain why I lost a 'whole dress rehearsal's' worth of recording but in Logic. I was using an Aggregate because I trusted my DUET's clock more than the MOTU rig. So I had an Aggregate with a 48 channel MOTU system AND my Apogee Duet. The DUET was set as clock and then 'automatically' the AUDIO MIDI.app puts a check next to MOTU. About an hour into the live dress rehearsal that 'digital growling' sound is recorded on ALL 41 tracks I was recording. From what you describe -- that was the FAULT of my having left CHECKED the RESAMPLE indicator next to the MOTU pieces???

So are you saying that I can also -- UNCHECK that box manually and I will be BETTER off and safer in the long run doing live recordings with an Aggregate? Why do you suppose the software AUTO check marks that box then when you choose one of the hardware boxes as the master?

And THANKS -- you may have solved a HUGE mystery for me.

Thanks!
Grant

siraltus
12-14-2010, 09:35 AM
O SIRALTUS, Please tell me you're sure!! :-) I mean that would explain why I lost a 'whole dress rehearsal's' worth of recording but in Logic. I was using an Aggregate because I trusted my DUET's clock more than the MOTU rig. So I had an Aggregate with a 48 channel MOTU system AND my Apogee Duet. The DUET was set as clock and then 'automatically' the AUDIO MIDI.app puts a check next to MOTU. About an hour into the live dress rehearsal that 'digital growling' sound is recorded on ALL 41 tracks I was recording. From what you describe -- that was the FAULT of my having left CHECKED the RESAMPLE indicator next to the MOTU pieces???

So are you saying that I can also -- UNCHECK that box manually and I will be BETTER off and safer in the long run doing live recordings with an Aggregate? Why do you suppose the software AUTO check marks that box then when you choose one of the hardware boxes as the master?

And THANKS -- you may have solved a HUGE mystery for me.

Thanks!
Grant


Yes - disable the "Resample" option or your word clock will drift out of phase between the aggregated interfaces and corrupt your audio!

NewdestinyX
12-14-2010, 09:42 AM
Yes - disable the "Resample" option or your word clock will drift out of phase between the aggregated interfaces and corrupt your audio!O BLESS YOU!!!! you'd mentioned problems on PLAYBACK in your first response but now you've mentioned RECORDING too!! Cool. That solves a HUGE mystery for me. That's what I LOVE about this board!.

The next thing I want CORE AUDIO to fix is the warning indicator that often says "SAMPLE RATES MISMATCH" in that same window at the bottom (in AUDIO MIDI.app) - when I've PERSONALLY set both hardware pieces to the SAME sample rate. I wonder why it gives that message ANYWAY.. Ugghh.. Makes me uneasy in an important recording situation. Makes me wanna think -- AGGREGATING is in its 'infancy' and to NOT use it when chips are down and client's payin' for flawless live performance recordings with NO DROP OUTS.

Thanks Siraltus!!
Grant

Carl Lie
12-14-2010, 09:49 AM
Wow. I've been getting this weird phased distortion and I wasn't sure what caused it so I kept resetting my hardware config to 256 buffer. Which seemed to work.

I can't wait to go home and try this. This would really make my day.

Carl

siraltus
12-14-2010, 12:48 PM
Wow. I've been getting this weird phased distortion and I wasn't sure what caused it so I kept resetting my hardware config to 256 buffer. Which seemed to work.

I can't wait to go home and try this. This would really make my day.

Carl

I had the same thing happening in Cubase - I kept going to preferences and resetting the CoreAudio device, which worked for about 5 minutes, then it would get all screwy again. It was always the slaved interface that got screwy, too. I bet this will fix your problem.

siraltus
12-14-2010, 12:50 PM
O BLESS YOU!!!! you'd mentioned problems on PLAYBACK in your first response but now you've mentioned RECORDING too!! Cool. That solves a HUGE mystery for me. That's what I LOVE about this board!.

The next thing I want CORE AUDIO to fix is the warning indicator that often says "SAMPLE RATES MISMATCH" in that same window at the bottom (in AUDIO MIDI.app) - when I've PERSONALLY set both hardware pieces to the SAME sample rate. I wonder why it gives that message ANYWAY.. Ugghh.. Makes me uneasy in an important recording situation. Makes me wanna think -- AGGREGATING is in its 'infancy' and to NOT use it when chips are down and client's payin' for flawless live performance recordings with NO DROP OUTS.

Thanks Siraltus!!
Grant

My pleasure! Yes, bad clock will screw both inputs and outputs, so both playback and recording are affected. BTW, I've been using my aggregate device with 108 I/O for 2.5 years now with no problems this way (since OS X 10.5.4) so I think it's a pretty stable technology. If you're having further issues, I'd look into updating the drivers for your interfaces or getting different interfaces altogether ;P

Carl Lie
12-14-2010, 05:29 PM
I went in and unchecked the resample and everything was fine UNTIL I decided to mess with my buffer size and all hell broke loose.

Total distortion. I tried going back to 256 and it still distorted. Rebooted and so on...no luck. I had to switch to my digital I/O to get things working again.

So I guess my luck with the Duet has run out for the time being.

Any suggestions?

Carl

NewdestinyX
12-14-2010, 07:31 PM
I went in and unchecked the resample and everything was fine UNTIL I decided to mess with my buffer size and all hell broke loose.

Total distortion. I tried going back to 256 and it still distorted. Rebooted and so on...no luck. I had to switch to my digital I/O to get things working again.

So I guess my luck with the Duet has run out for the time being.

Any suggestions?

CarlWhat are you trying to 'aggregate' the DUET with, Carl? Naturally I'm very interested as I have a DUET too. Are you using the DUET as the master clock? And what I know too is that you can't change anything in the AUDIO MIDI.app while ProTools open without all hell breaking loose. You have to quit ProTools before you make any changes in the AUDIO MIDI.app window. Then Relaunch ProTools.

Siraltus seems to be saying that this 'aggregate' window has been a part of Core Audio MIDI.app for several years now. I wonder why I never saw it before ProTools 9?... I know it's not a part of PT9. Maybe I just never needed it before since I've been a PTHD guy for a couple of years now. Tell us more about your setup Carl so there's a chance of diagnosing the situation with you.

Amin7b5
12-15-2010, 09:18 AM
I tried running the Duet with a Presonus Firestudio and had the same issues. Tried every combo of Resample checked/unchecked... They would work ok for awhile, then gradually start popping/clicking, like the clock was drifting the longer it was on. :(

getz76
12-15-2010, 09:22 AM
From the FAQ:

Q: Does Pro Tools 9 Support Aggregate Audio Devices in Mac OS X? (FAQ added 12/1/2010)

A: Mac OS X device aggregation of audio I/O hardware is not officially supported in Pro Tools 9. While certain configurations may work, they were not tested and qualified by the Pro Tools development team. The only exception to this rule is the "Pro Tools Aggregate Device" that is created upon launching Pro Tools to provide users of built-in Mac hardware with simultaneous record and playback capabilities.

That said, it should work. Set your interfaces to the desired bit and sample rate, including the Mac's internal audio. Use the Mac as the master clock.

getz76
12-15-2010, 09:26 AM
Also, the "resample" checkbox will do what it says; if you have a device set at 48kHz and you send a 96kHz output to it, it will resample the audio to 48kHz. If you have all the devices set to the same sample rate, you should not have "resample" checked.

NewdestinyX
12-15-2010, 10:01 AM
I tried running the Duet with a Presonus Firestudio and had the same issues. Tried every combo of Resample checked/unchecked... They would work ok for awhile, then gradually start popping/clicking, like the clock was drifting the longer it was on. :(UGGH... So it's not as simple as unchecking that box.. Well -- it was a nice hope. But you have a similar set up to mine and I can't ever again trust an aggregate setting for an important live recording unless I have MDM redundant backup or something.

From the FAQ:

Q: Does Pro Tools 9 Support Aggregate Audio Devices in Mac OS X? (FAQ added 12/1/2010)

A: Mac OS X device aggregation of audio I/O hardware is not officially supported in Pro Tools 9. While certain configurations may work, they were not tested and qualified by the Pro Tools development team. The only exception to this rule is the "Pro Tools Aggregate Device" that is created upon launching Pro Tools to provide users of built-in Mac hardware with simultaneous record and playback capabilities.

That said, it should work. Set your interfaces to the desired bit and sample rate, including the Mac's internal audio. Use the Mac as the master clock.Yes.. I know I should remember that it's not officially supported. The problem is the 'growling' recording happened in Logic. So there's something wrong with CORE UAUDIO aggregating with the DUET I fear. One thing I hadn't considered is that maybe I should checkmark the MAC input and output and set IT as the master clock. Which of course would only work if I want to record at 44 or 48. MAc's can't do 96. I don't think. I guess I feel the Apogee has the more stable clock but it never occurred to me that the softwares are running IN a MAC OS so maybe the MAC itself needs to be 'clocked' and part of any AGGREGATE. But that's only a guess.

Also, the "resample" checkbox will do what it says; if you have a device set at 48kHz and you send a 96kHz output to it, it will resample the audio to 48kHz. If you have all the devices set to the same sample rate, you should not have "resample" checked.Yes, that much I know. Siraltus would have us just 'buy new hardware' until it works.. But I'm sorry... I need a higher standard from the software people and CORE AUDIO itself. If you're going to 'make it available' -- it needs to work... with any ASIO based hardware. Is that too much to ask??? :rolleyes: :(

Carl Lie
12-15-2010, 10:29 AM
I'm aggregating it with the digital I/O on my mac. When I just run through the digital I/O i can go to a buffer 64 with no problems. When I run out through the Duet I get distortion at anything below 256.

I'm cool not to use the Duet as it's headphone amp doesn't crank the AKG's very well anyway. My only problem is that I can't select the digital out for my Master fader for some reason.

I'l try to set it up outside of PT and then see what happens.

Carl

Top Jimmy
12-15-2010, 10:50 AM
If you have all the devices set to the same sample rate, you should not have "resample" checked.

Bull crap! If you have no other way to clock sync the interfaces, you need to resample all of the ones that aren't the clock master. No two clocks are exactly alike when set at the same rate. One might be 48.004 and the other 47.997. This is enough to be a problem. Always use resample unless you can externally clock sync the interfaces. "There can be only one."

siraltus
12-15-2010, 11:12 AM
I'm aggregating it with the digital I/O on my mac. When I just run through the digital I/O i can go to a buffer 64 with no problems. When I run out through the Duet I get distortion at anything below 256.

I'm cool not to use the Duet as it's headphone amp doesn't crank the AKG's very well anyway. My only problem is that I can't select the digital out for my Master fader for some reason.

I'l try to set it up outside of PT and then see what happens.

Carl

Looks like the problem here is not with device aggregation and word clock, but the fact that the Duet is seemingly unable to deliver stutter-free audio at anything below a 256 sample buffer?

getz76
12-15-2010, 11:16 AM
Not bull crap, actually. You should be providing a master clock and determing the clock source on your interface (either on hardware or software). The resample checkbox has nothing to do with determining the clock source.

NewdestinyX
12-15-2010, 01:00 PM
Bull crap! If you have no other way to clock sync the interfaces, you need to resample all of the ones that aren't the clock master. No two clocks are exactly alike when set at the same rate. One might be 48.004 and the other 47.997. This is enough to be a problem. Always use resample unless you can externally clock sync the interfaces. "There can be only one."

Not bull crap, actually. You should be providing a master clock and determing the clock source on your interface (either on hardware or software). The resample checkbox has nothing to with determing the clock source.Sorry, TopJimmy, Getz is right on this one. When it comes to the interface itself - Core audio allows you to choose one of the interfaces in the Aggregate as the Master Clock -- which it then use to 'clock' the 'INTERNAL' clock of the other device. So in the 'hardware config' of each 'individual' device the clocks would be set to 'internal' for both units -- but CORE audio aggregation sends the clock from one unit to another depending on which you choose in the Aggregate window of the AUDIO MIDI.app. Now if you have 'other' external gear -- like digital PRE's (like my Presonus Digimax LT's and FS) you do have to send them 'actual' word synch from some hardware interface and make sure the PRE's are in External Clock. But 'interface to interface' clocking is SUPPOSED TO BE ABLE TO be done by the Aggregate window itself. Though I'm beginning to wonder if it really works -- or if it's buggy with some pieces. My first concern upon buying the DUET was noticing that in the tail out -- there was no WORD CLOCK in or out jack. So it either has to be the master -- or IF Core Audio aggregation works properly it can be a slave. I LOVE the way this thing sounds -- I sure hope I can make the Aggregation window work with it.

Top Jimmy
12-15-2010, 01:49 PM
Not bull crap, actually. You should be providing a master clock and determing the clock source on your interface (either on hardware or software). The resample checkbox has nothing to do with determining the clock source.

Never said it did. What I did say was that if you don't sync all the hardware to the master clock then you need to resample. Some of Avid's HD interfaces support real-time sample rate conversion (resampling) just for this situation.

Sorry, TopJimmy, Getz is right on this one. When it comes to the interface itself - Core audio allows you to choose one of the interfaces in the Aggregate as the Master Clock -- which it then use to 'clock' the 'INTERNAL' clock of the other device. So in the 'hardware config' of each 'individual' device the clocks would be set to 'internal' for both units -- but CORE audio aggregation sends the clock from one unit to another depending on which you choose in the Aggregate window of the AUDIO MIDI.app. Now if you have 'other' external gear -- like digital PRE's (like my Presonus Digimax LT's and FS) you do have to send them 'actual' word synch from some hardware interface and make sure the PRE's are in External Clock. But 'interface to interface' clocking is SUPPOSED TO BE ABLE TO be done by the Aggregate window itself. Though I'm beginning to wonder if it really works -- or if it's buggy with some pieces. My first concern upon buying the DUET was noticing that in the tail out -- there was no WORD CLOCK in or out jack. So it either has to be the master -- or IF Core Audio aggregation works properly it can be a slave. I LOVE the way this thing sounds -- I sure hope I can make the Aggregation window work with it.

I must be completely ignorant of this magical technology. You're aggregating various pieces of hardware that each and of themselves do not support distributing clock via the interface layer. They rely on an internal crystal oscillator that can be switched to sync to other hardware sources, that's all. The rules of digital have not changed yet. You must sync all digital clocks or you must resample, even when the clocks are set the same. There's no getting around that. If I am wrong then show me the error and I will publicly admit it.

getz76
12-15-2010, 01:55 PM
Never said it did. What I did say was that if you don't sync all the hardware to the master clock then you need to resample. Some of Avid's HD interfaces support real-time sample rate conversion (resampling) just for this situation.

Resampling via dedicated hardware is one thing, but using the resample of CoreAudio in place of a clock is a non-starter for DAW application. Latency and quality are going to go out the window.

How about this? You need to distribute a clock source. <full stop> :)

Top Jimmy
12-15-2010, 02:05 PM
You need to distribute a clock source.:)

Agreed, but one man's necessity is another man's excess. Nowhere did the OP state that he was distributing clock. If he's not going to do so, then there's no alternative and he must resample.

NewdestinyX
12-15-2010, 03:09 PM
I must be completely ignorant of this magical technology. You're aggregating various pieces of hardware that each and of themselves do not support distributing clock via the interface layer. They rely on an internal crystal oscillator that can be switched to sync to other hardware sources, that's all. The rules of digital have not changed yet. You must sync all digital clocks or you must resample, even when the clocks are set the same. There's no getting around that. If I am wrong then show me the error and I will publicly admit it.I believe where you be erring is in your belief that the 'internal' mode of 2 pieces of hardware 'can't' be given clocking info FROM Core Audio. And that core audio can't 'detect' the clock from an attached piece of hardware.

If Core Audio 'can't' do this - then aggregating would be an impossibility. The reason I'm pretty darn sure I'm right on this one -- is that IF the MAC via core audio WEREN'T altering the 'internal' clock of one of the attached hardwares we'd be immediately hearing 'clicks and pops like you do in any clock interference situation.

Now you do have my interest peaked... in that the only guy that's chimed in so far that he's been aggregating with perfection for over 2 years (Siraltus) is a guy who has 95% of his I/O from ONE PCI424 card from MOTU which all 'clocks to itself thru the PCI424 card.. OR there may be word clock cables connected to each hardware piece.

Since the DUET has no word clock cables are you saying - the only way to make it work would be to put my MOTU rig in RESAMPLE mode? I already tried that with disastrous results... digital growling thru the last hour of a live show. So I KNOW 'that' doesn't work. But let's not have a 'pissing' contest to see who can 'grunt' the loudest and be the 'rightest'.. I'm interested in a good, detailed 'digital clock/aggregating' discussions. But I need to know the 'whys' and 'wherefores'.


How about this? You need to distribute a clock source. <full stop> :)
Agreed, but one man's necessity is another man's excess. Nowhere did the OP state that he was distributing clock. If he's not going to do so, then there's no alternative and he must resample.CORE AUDIO "aggregating" claims to 'distribute clock'. That's the point of Aggregating.

Top Jimmy
12-15-2010, 05:34 PM
I believe where you be erring is in your belief that the 'internal' mode of 2 pieces of hardware 'can't' be given clocking info FROM Core Audio. And that core audio can't 'detect' the clock from an attached piece of hardware.

If Core Audio 'can't' do this - then aggregating would be an impossibility. The reason I'm pretty darn sure I'm right on this one -- is that IF the MAC via core audio WEREN'T altering the 'internal' clock of one of the attached hardwares we'd be immediately hearing 'clicks and pops like you do in any clock interference situation.

Now you do have my interest peaked... in that the only guy that's chimed in so far that he's been aggregating with perfection for over 2 years (Siraltus) is a guy who has 95% of his I/O from ONE PCI424 card from MOTU which all 'clocks to itself thru the PCI424 card.. OR there may be word clock cables connected to each hardware piece.

Since the DUET has no word clock cables are you saying - the only way to make it work would be to put my MOTU rig in RESAMPLE mode? I already tried that with disastrous results... digital growling thru the last hour of a live show. So I KNOW 'that' doesn't work. But let's not have a 'pissing' contest to see who can 'grunt' the loudest and be the 'rightest'.. I'm interested in a good, detailed 'digital clock/aggregating' discussions. But I need to know the 'whys' and 'wherefores'.
CORE AUDIO "aggregating" claims to 'distribute clock'. That's the point of Aggregating.

http://support.apple.com/kb/HT3956

Expand the "Setup An Aggregate Device" section and review #6 & #7.

NewdestinyX
12-15-2010, 07:45 PM
http://support.apple.com/kb/HT3956

Expand the "Setup An Aggregate Device" section and review #6 & #7.Alrightey then! Well I stand corrected, Top Jimmy -- you were right all along. Bottom line.. The only way to assure clocking accuracy in Aggregates between two hardware devices is to physically connect their work clocks to each other IF they have them. A DUET does not have one. Therefore it will always need to be in RESAMPLE mode when in an Aggregate.

The guys on Gearslutz were saying that whenever you're trying to aggregate two devices WITHOUT their work clocks physically connected you need to put BOTH hardware devices in RESAMPLE. Does that seem right to all here?

What I'm still left dumbfounded by in all this is WHY, then, does CORE AUDIO give you a MASTER CLOCK choice in an Aggregate device? HOW is it being a MASTER? Master - for the resampling comparison? If that's so then the thread conclusion at Gearslutz is wrong. And the MASTER device would NOT be in resample and the other devices would.

Oh -- I may have moved too fast in getting this DUET. Boy do I love the way it sounds - and the Ensemble's just out of my range financially. But I need rock stable clock between my ProFire LightBridge (which does have a word clock input) and the Duet (no word clock). Uggh.

Thanks all!

getz76
12-15-2010, 08:04 PM
The guys on Gearslutz were saying that whenever you're trying to aggregate two devices WITHOUT their work (sic) clocks physically connected you need to put BOTH hardware devices in RESAMPLE. Does that seem right to all here?

Well, yes, but I would not do it for anything critical. No way that is going to be stable.

What I'm still left dumbfounded by in all this is WHY, then, does CORE AUDIO give you a MASTER CLOCK choice in an Aggregate device? HOW is it being a MASTER? Master - for the resampling comparison? If that's so then the thread conclusion at Gearslutz is wrong. And the MASTER device would NOT be in resample and the other devices would.

Resampling is not the same as a clock.

You need to tell CoreAudio what the master clock is because that is what the DAW software will sync to. Simples.

If you are using multiple sets of AD/DA, you need to have a clock physically connected (usually through a BNC, Toslink or coaxial cable). You could use the resampling instead, but it is not going to be stable enough for low-latency, real-time audio work in my opinion.

NewdestinyX
12-16-2010, 08:21 AM
Well, yes, but I would not do it for anything critical. No way that is going to be stable.That explains my loss of lock and growling audio after an hour in Logic aggregating the DUET and a MOTU 48 channel system -- where MOTU was slave. Boy -- I made so many 'recent purchase' decision based on the fact that this was all gonna work.. uggh... But I do that every time. I get out of one platform and JUMP in.. It's my own fault.. You guys are SO generous with your information and time. I just need to read more and ask more questions BEFORE buying. Now I'm stuck with a GREAT sounding interface, the DUET, that has no physical word clock out or in that I'll never be able to aggregate for the hardware inserting I want to create like my HD system had. Uggh... I just can't afford the Ensemble. But I need a GREAT converter to listen through for my mixing decisions AND to be able to create hardware inserts to my analog outboard gear... uggh... more thinking to do.
Resampling is not the same as a clock.

You need to tell CoreAudio what the master clock is because that is what the DAW software will sync to. Simples [sic].So do you AGREE with the gearslutz guys that IF you're going to go 'without' a physical word clock connection -- BOTH units need to be set to 'resample' in AUDIO MIDI? Even the one you set to MASTER? That's not what others said here before. So I'm a little confused.

If you are using multiple sets of AD/DA, you need to have a clock physically connected (usually through a BNC, Toslink or coaxial cable). You could use the resampling instead, but it is not going to be stable enough for low-latency, real-time audio work in my opinion.I get that aggregating work most stably with a physical word clock connection.

1)So all these 'my DUET loses lock in an aggregate' threads I read everywhere are simply due to the DUET not having word clock external ability?
2) So IF I'm not 'recording' and needing 256k buffers for listen thru do you think things would be 'stable enough' (using double 'resample') for '1024' buffer in mixing mode? I of course need the DUET as stable as possible to 'hear' clearly my mix -- and all I need the PROFIRE LIGHTBRIDGE + Digimax FS for is to create hardware inserts to my Distressors, etc.

Whatcha think?

getz76
12-16-2010, 08:23 AM
So do you AGREE with the gearslutz guys that IF you're going to go 'without' a physical word clock connection -- BOTH units need to be set to 'resample' in AUDIO MIDI? Even the one you set to MASTER? That's not what others said here before. So I'm a little confused.

No. I do not agree. It is like using a spoon to cut a steak.

Buy an interface that works for the application.

NewdestinyX
12-16-2010, 08:34 AM
No. I do not agree. It is like using a spoon to cut a steak.

Buy an interface that works for the application.Thanks - but with all due respect and appreciation - I don't really need advice on 'hardware purchases'. For the moment I just need 'technical questions answered about AUDIO MIDI'. So - the one marked 'master' should NOT be in Resample mode - only the slave - agreed?

getz76
12-16-2010, 08:48 AM
Thanks - but with all due respect and appreciation - I don't really need advice on 'hardware purchases'. For the moment I just need 'technical questions answered about AUDIO MIDI'. So - the one marked 'master' should NOT be in Resample mode - only the slave - agreed?

Good luck.

NewdestinyX
12-16-2010, 08:56 AM
For the moment I just need 'technical questions answered about AUDIO MIDI'. So - the one marked 'master' should NOT be in Resample mode - only the slave - agreed?Good luck.
:rolleyes:

Top Jimmy
12-16-2010, 09:33 AM
The guys on Gearslutz were saying that whenever you're trying to aggregate two devices WITHOUT their work clocks physically connected you need to put BOTH hardware devices in RESAMPLE. Does that seem right to all here?

That is absolutely incorrect. One of the interfaces has to provide an absolute clock. The software cannot "make-up" a clock and then sample rate convert all incoming streams to it. The clock master in the aggregated interfaces setup should not be set to resample.

Since you cannot clock the Duet externally and its only function is for monitoring, I would make the Profire the clock master. This way, there's no resampling going on on the input and insert sides of the mix, only in the monitoring path.

As for getz76's assertion that the Coreaudio AU sample-rate conversion should be avoided at all costs, that's up to you to put to the test. Few people realize that we listen to this function all the time when listening via iTunes, Quicktime, et al. Whenever media we're listening to changes sample rate, the converter clock does not automatically switch. Instead, this SRC function is silently occurring.

NewdestinyX
12-16-2010, 09:49 AM
That is absolutely incorrect. One of the interfaces has to provide an absolute clock. The software cannot "make-up" a clock and then sample rate convert all incoming streams to it. The clock master in the aggregated interfaces setup should not be set to resample.

Since you cannot clock the Duet externally and its only function is for monitoring, I would make the Profire the clock master. This way, there's no resampling going on on the input and insert sides of the mix, only in the monitoring path.Thanks, TopJimmy. THat's what I was beginning to conclude too. Though I hate that idea -- since the Apogee Duet's clock is a baby brother to the BigBen - -even in a $499 piece. I think I agree. On the 'profire side' of things there are lots of WordSync connections and I want that all to be rock stable. Might as well have that be the master.. And I still own a LUCID Genx6 master clock machine which I can use as the ultimate master to the Profire as well. Then I'll just hope that thru Core Audio resampling the DUET can remain 'stable' and still sounding like its amazing self. In the end there are still many mixes I don't touch my Distressors and other outboard gear for - so in those situations I'll have the DUET as master and be all the happier.

I think I have a working situation. I'll be setting it all up this afternoon since all pieces of the puzzle are here. I was waiting still on my Digimax FS which came yesterday. I can now plug it all in and open up my 'biggest' mix from my HD1 rig and 'see what happens'. I LOVE 'tech days'.. :D I know I'll have to find a 'Hardware Insert Delay amount' in this new rig. What's the best way to 'discover the offset'? Use your ears?
As for getz76's assertion that the Coreaudio AU sample-rate conversion should be avoided at all costs, that's up to you to put to the test. Few people realize that we listen to this function all the time when listening via iTunes, Quicktime, et al. Whenever media we're listening to changes sample rate, the converter clock does not automatically switch. Instead, this SRC function is silently occurring.But I HATE that sound and can hear it. My grandmother could hear the difference. That's what I'm most concerned about... but -- we'll see. I'm one of those 'nuts' that can hear the difference between a piece of gear inserted on a channel in BYPASS mode and 'not inserted'. If a Distressor is 'in' the insert chain -- in bypass -- my track sounds different than if I 'unplug' it from the chain. So 'how something sounds' is one of those things I'm super sensitized to.

Thanks all!

Top Jimmy
12-16-2010, 10:03 AM
I know I'll have to find a 'Hardware Insert Delay amount' in this new rig. What's the best way to 'discover the offset'? Use your ears?

I've always just drawn a spike with the pencil tool and then printed the return to a different track. Measure the difference with the cursor tool and it will show the samples in the counter if you switch it to display samples. ADC should handle the 2x buffer itself, the only additional you'd want is the small amount of actual conversion delay.

NewdestinyX
12-16-2010, 10:07 AM
I've always just drawn a spike with the pencil tool and then printed the return to a different track. Measure the difference with the cursor tool and it will show the samples in the counter if you switch it to display samples. ADC should handle the 2x buffer itself, the only additional you'd want is the small amount of actual conversion delay.Forgive my ignorance, TopJimmy, but what do you mean by the 2x buffer? And how does it affect my 'math equation' after I measure the lag using your approach? I always use the 'longest' (4096 at 48k/8192 at 96k) ADC turned on in ProTools9 and run the buffer at 512k when mixing. Should I do my pencil draw test with ADC ON or OFF and then how does it factor into my math equation?

Thanks!
Grant

siraltus
12-16-2010, 11:30 AM
Now you do have my interest peaked... in that the only guy that's chimed in so far that he's been aggregating with perfection for over 2 years (Siraltus) is a guy who has 95% of his I/O from ONE PCI424 card from MOTU which all 'clocks to itself thru the PCI424 card.. OR there may be word clock cables connected to each hardware piece.


Just to clarify:

I have a MOTU PCI-424 card with two 2408mk3 and one 24I/O boxes sitting on it, as well as an M-Audio ProFire Lightbridge.

Both are being clocked externally via BNC from the Apogee clock card in my Mackie d8b mixer (and so are all my external digital FX processors and synthesizers/samplers).

NewdestinyX
12-16-2010, 05:35 PM
Just to clarify:

I have a MOTU PCI-424 card with two 2408mk3 and one 24I/O boxes sitting on it, as well as an M-Audio ProFire Lightbridge.

Both are being clocked externally via BNC from the Apogee clock card in my Mackie d8b mixer (and so are all my external digital FX processors and synthesizers/samplers).Thanks for the details! Those MOTU mega systems are rock stable. It's a shame after DP 4 the software became so glutted an unstable. Beta tested for them for 6 years and then at one release they embarrassed me in front of a client with a screwy release for the 'last time' and I've been ProTools ever since. My son's stayed in DP the whole run and V7.2 at about 16 tracks still has the counter all jiggy on playback and red overs in the CPU performance meter. MOTU=Great hardware-- clever comprehensive but glutty unstable software.

Top Jimmy
12-16-2010, 07:43 PM
Forgive my ignorance, TopJimmy, but what do you mean by the 2x buffer? And how does it affect my 'math equation' after I measure the lag using your approach? I always use the 'longest' (4096 at 48k/8192 at 96k) ADC turned on in ProTools9 and run the buffer at 512k when mixing. Should I do my pencil draw test with ADC ON or OFF and then how does it factor into my math equation?

When using hardware inserts, the primary delay is caused by the hardware input/output buffer. If it's set to 512, then your round trip delay will be 512+512+AD/DA delay. ADC will cover the delay from the buffer, but if you want to be sample accurate, you'll have to measure the AD/DA delay and enter that into the H/W Insert Delay tab in the I/O setup. This value is in milliseconds so you'll have measure it with the time selected on the counter.

With ADC on, draw a spike or have some easily measurable region on a track. Create a hardware insert on this track and loop a cable from the output back to the input. Route this track to another track and print it. Measure the difference in milliseconds between the region in the first track and the second. Enter this number in the appropriate field in the I/O setup. Now your hardware inserts will be sample accurate.

NewdestinyX
12-16-2010, 07:51 PM
When using hardware inserts, the primary delay is caused by the hardware input/output buffer. If it's set to 512, then your round trip delay will be 512+512+AD/DA delay. ADC will cover the delay from the buffer, but if you want to be sample accurate, you'll have to measure the AD/DA delay and enter that into the H/W Insert Delay tab in the I/O setup. This value is in milliseconds so you'll have measure it with the time selected on the counter.

With ADC on, draw a spike or have some easily measurable region on a track. Create a hardware insert on this track and loop a cable from the output back to the input. Route this track to another track and print it. Measure the difference in milliseconds between the region in the first track and the second. Enter this number in the appropriate field in the I/O setup. Now your hardware inserts will be sample accurate.Thanks a million, TopJimmy! Very clear instructions. That's what I'll do. In terms of what's actually gonna then end up happening. I guess ADC will add 'extra' 'moving ahead of the other audio tracks (not having inserts) based on that number you enter in the I/O delay window -- since there's no way to move 'real time' earlier -- except in the StarTrek universe.. :rolleyes:.. So it's accomplishing sample lock by adding even 'more' movement 'earlier in time' of the other existing audio printed tracks.. Right? So if you're already close to 'max' ADC on a couple of tracks in a mix -- if the ADC window in ProTools has 'all green indicators' and one 'yellow' ('yellow' I think showing the 'greatest movement needed).. it's likely adding HW insert delays could cause a couple of tracks needing the most ADC to go into the 'red' in Delay comp window.. Agreed?

Top Jimmy
12-17-2010, 07:58 AM
Thanks a million, TopJimmy! Very clear instructions. That's what I'll do. In terms of what's actually gonna then end up happening. I guess ADC will add 'extra' 'moving ahead of the other audio tracks (not having inserts) based on that number you enter in the I/O delay window -- since there's no way to move 'real time' earlier -- except in the StarTrek universe.. :rolleyes:.. So it's accomplishing sample lock by adding even 'more' movement 'earlier in time' of the other existing audio printed tracks.. Right? So if you're already close to 'max' ADC on a couple of tracks in a mix -- if the ADC window in ProTools has 'all green indicators' and one 'yellow' ('yellow' I think showing the 'greatest movement needed).. it's likely adding HW insert delays could cause a couple of tracks needing the most ADC to go into the 'red' in Delay comp window.. Agreed?

While it would seem intuitive to advance the delayed tracks, something which could be done with zero active processing for audio tracks, Pro Tools actually actively delays all tracks to line up with the longest one.

Ordinarily, using at least one hardware insert per track should not be a problem unless the mixer routing that you've implemented is convoluted, complex, and uses a lot of high latency plugs. Keeping the mixer layout simple and straightforward helps a great deal. Think along the lines of Source Tracks -> Submix Tracks -> Master Mix. In Pro Tools, it's easy to implement a convoluted routing that couldn't exist on an analog console.

NewdestinyX
12-17-2010, 08:29 AM
While it would seem intuitive to advance the delayed tracks, something which could be done with zero active processing for audio tracks, Pro Tools actually actively delays all tracks to line up with the longest one.

Ordinarily, using at least one hardware insert per track should not be a problem unless the mixer routing that you've implemented is convoluted, complex, and uses a lot of high latency plugs. Keeping the mixer layout simple and straightforward helps a great deal. Think along the lines of Source Tracks -> Submix Tracks -> Master Mix. In Pro Tools, it's easy to implemented a convoluted routing that couldn't exist on an analog console.Yes, of course.. Silly me.. It would 'delay' the tracks to meet the 'longest delay' needed. Makes total sense. And yes.. I get the stay away from 'convoluted' concept. I've been a recording/PT engineer for 'way' too many years and beta tested for 'MOTU' for 6 years where they asked us to 'push the system to its limits' - as if we engineers don't do that EVERY DAY with our software DAW's anyway!!! :D.

The outboard gear, that I still think 'makes a difference' in my mixes, are mostly compressors which, for most of us, are a 'first in line' insert step after source anyway. I have a TUBE MP by ART that I 'love what it does' as well -- again, a 'first in order' insert - right back to PT and then a couple of RTAS plugs after 'if' that.. mostly just an EQ after that. So I will see if I push any of the tracks over into 'the red' on ADC.. And even then sometimes it's only 2-100 samples over the limit which our ears can't detect as delay anyway.

Thanks again for the tutorial back into the NATIVE world. I 'loved' not having to think about this stuff having been in HD for the last 2.5 years. But times being what they are -- the clients have slowed down. And the 'cash' was needed more than the 'ease' of HD. And with the advent of PT9's 'happy bedfellowing' with ASIO hardware - and the speed of the computers these days (and, sorry Avid, but no sonic difference between RTAS and TDM plugs anymore) -- the time was right to move back to native where I spent MANY years anyway.

In the end -- for 99% of us on this user board -- it's about our 'experience' and our 'ears' that make for great mixing - not an Apogee 8000 + PT HD system. I can still make great sounding records on 3 ADAT 20bit MDM's running into a Behringer board if I HAD to.. LOL!!! - - - because I know 'how' to mix. :cool:

Thanks again!
Grant

NewdestinyX
12-18-2010, 02:29 PM
As a postscript to this thread for others' benefit in considering master/slave relationships through Aggregates -- I can verify, after everything getting hooked up today that the best scenario for those of us using Apogee Duets in an Aggregate with another interface -- that it's probably best to have the Apogee be the slave and set to 'resample'.

I have a Profire Lightbridge that's connected via word clock to a Presonus Digimax FS that I'm using to add hardware inserts to PT9 for my analog gear. When I tried to run the Duet as MASTER and Profire as slave (resampling) -- the track I ran through the insert to the FS was all digitally growling. Even after restarts of all gear -- every time -- the growling. So resampling the Profire -- the insert side of my equation does NOT work at all. Unusable.

But when the DUET is slave and set to 'resample' -- ProFire the master -- all is well for hours on end. Though I'm not thrilled with the idea of my 'main 2 mix' monitoring environment being set to 'resample' mode - it's working and I'm 'pretty' sure I can't hear any difference with it as slave or master. Who knows. I'll only ever be sending a 2 mix to it. Until I can save up for an Ensemble this will have to do.

The main take-away from this thread for me is that Aggregating via the Audio MIDI.app [Core Audio], though you do need to choose a master and slave -- does IN NO WAY have the ability to access the 'internal clocks' of your third party (or Avid) interfaces. All your interfaces in an Aggregate MUST be 'Word clocked' together, physically, via BNC Word 1x, AES or SP/DIF cables with one interface being the Synch master. If you have a piece, like the Apogee Duet, that canNOT be externally synched or be a Master via Word 1x (AES or SP/DIF) then your ONLY choice is to have that piece be in 'resample' mode in Core Audio - which does introduce a 'lugubrious' calculation process and 'could' introduce noise/jitter into whatever's coming out of the 'resampled' piece.

The DUET is nearly perfect except for no word 1x out. I understand. Just like Avid needed to artificially cap PT9 Standard at 32 I/O - Apogee had to make DUET a standalone, non word synchable piece -- to preserve dual to tri layers of price points in their line. The 'Ensemble', at $1995, is their 'mid price point' and of course has Word 1x out. Just like HD|Native is Avid's mid price point and has more than 32I/O (I think.....).

Thanks for all the help, gang! :-)

zoff
12-26-2010, 02:57 PM
I probably should start a new thread but seeing as you guys have experience with the Duet and Aggregate I/O I'll ask here. When I set PT 9 to use Aggregate as its I/O the CPU hovers around 100% just for PT even without a session open!
My hardware is MacBook Pro Core 2 Duo, Duet (for monitoring) and the ProFrire 2626.

Thanks.