Craig & Pals -- hardware (converter) latency comp -- what DAW ware has it?

[blue2blue]

[ Look at the can of worms that got opened in the "Do you use guitar plugs" thread... I apologize in advance for the length of this and I excuse Craig or anyone who doesn't have the time or inclination to deal with this right now -- or ever. ]

 

 

When I was using my desktop box with its Echo Mia PCI audio interface, its conversion/transport latency was low enough that I sort of shined on the issue of 'track misalignment' due to hardware/converter latency. With that card, with Sonar 'optimized' for it with the Profiler, track misalignment due to converter latency was around 4.5 ms.

 

Since I typically put down drums and use that as the primary time reference when I overdub, this amount of 'misalignment' between the drums and subsequent tracks did not become too problematic.

 

(Where you might run into problems is if, say, you cue off drums to cut the bass, then cue off the the bass to play the piano, then cue off the piano to cut the guitar, and finally cue off the guitar to sing... in an unlikely worst case scenario all that potential chronological misalignment could cascade like a bozo bucket brigade of lagging chronological landmarks. If you catch my drift. No pun intended. But, of course, in the real world you'll probably stay fairly anchored to the drums.)

 

Even so, a lot of times, after I was done tracking, I'd slide the drum track around a bit to taste.

 

 

But, my MOTU 828 mkII can only get down to 128 sample buffers w/o running into problems (it can go to 96 but not on my rig), which works out to a one way latency of 2.9 ms.

 

But, of course, overdubbing is a 'two way street' -- you are dealing with whatever playback/conversion latency you have and adding to that any input/conversion latency you have, tossing other processing latency on top of it... and still hopefully coming up with something negligible.

 

On my Motu that works out to a total of 366 samples of 'delay' (about 8.3 ms at 44.1 kHz), meaning an 8.3 ms misalignment between existing tracks and newly recorded tracks.

 

(The way to establish this figure, as you probably know, is to route a previously recorded sound, something with a clean, easy-to-spot-the-beginning-of transient, out one analog output and into an an analog in, record it, and measure the 'gap' between where the original begins and where the newly recorded copy begins. [Of course, one must be careful to avoid a potentially devastating feedback loop.])

 

 

Now, yes, that is the time it takes for sound to travel 8 or 9 feet, as I'm well aware. But it is also a musically significant amount of time, one which can be annotated (at least at higher tempos -- I think it's a quasihemidemisemiquaver at around 180 bpm or so ).

 

Let's put it this way so that people are clear I'm not trying to make myself out as a goldenears -- I'm just not a good enough musician to not be thrown off by that.

 

 

Anyhow, before I got it, I had been thinking that the current version of Sonar had 'converter latency compensation' for this -- but as far as I can tell, it doesn't. Pro Tools LE also does not have compensation for this 'conversion latency' either. (Sonar does have automatic plugin delay compensation, however. A whole 'nother, if related, issue. PT LE doesn't have APDC yet, apparently, at least if the UAS site is still right.)

 

 

So, anyhow, I'm getting into the habit of using the Sonar Nudge control (a rather clumsy implementation, I must say) to 'manually' nudge newly recorded tracks 366 samples left (earlier). [Obviously, I don't need to do that on the very first track, but it doesn't matter if I do, since wherever that track ends up is, de facto, the 'reference' point.]

 

But, every time I do this rote manual adjustment, being a programmer (and, ergo 'lazy' when it comes to repeated tasks) I find myself thinking -- wouldn't it be great if Sonar would do that automatically.

 

(And then I wouldn't have to worry about accidentally moving it 366 samples right or something, either. But that's more a complaint about the bad nudge implementation in my beloved Sonar, anyhow.)

 

 

I've been told that Cubase/Nuendo has such a feature. But when I looked over the Cubase feature list just now, I didn't see it.

 

But there seems to be a lot of confusion about all the different latency issues that come up in audio work: monitoring latency, this above-discussed hardware/conversion latency, plug in latency, etc.

 

(So much confusion that I was told, for instance, that PT LE had added the kind of conversion latency compensation I'm talking about here, but it appears that was incorrect.)

 

Someone also suggested that Logic might have such a conversion latency compensation. Is this true of "Cubendo" or Logic? How about boutique DAWs like Samplitude/Sequoia? (I just got back from checking the Samplitude site, and I'm still not sure. Doesn't look like it, though. I did a site google looking for latency compensation and it all looked like plugins. Happy to be otherwise corrected. )

 

 

Can o' worms, anyone?

 

 

[PS... Here's a pretty good article on latency in Sound On Sound -- but it doesn't, I don't think, address this particular issue: http://www.soundonsound.com/sos/Sep02/articles/pcmusician0902.asp ]

 

[Alndln2]

 

Originally posted by blue2blue I've been told that Cubase/Nuendo has such a feature.

From what I read(Craigs review),is that you can loop hardware effects in your extra i/o in either insert or aux and compensation is not automatic,but rather you have to calculate and adjust the settings youself.Since I use neither app anymore,I'm not sure if this is global or adjustable to specific i/o.

 

But there seems to be a lot of confusion about all the different latency issues that come up in audio work: monitoring latency, this above-discussed hardware/conversion latency, plug in latency, etc.

Sonar will compensate for everything software including live input of plugs,but not lopped external hardware.

 

(So much confusion that I was told, for instance, that PT LE had added the kind of conversion latency compensation I'm talking about here, but it appears that was incorrect.)

Correct,PTHD recently added PDC similar to Sonar/Cubase etc,but not PTLE(yet).I'm not clear however,wether PTHD compensates for external hardware.

 

Someone also suggested that Logic might have such a conversion latency compensation. Is this true of "Cubendo" or Logic? How about boutique DAWs like Samplitude/Sequoia?

I read about Logics i/o plugin for inserts a while back,and in it's earlier incarnation there was no hardware latency compensation,but I haven't kept up regarding update features,but I do know that in Cubase and Nuendo,hardware latency is adjustable.To what degree as far as individual i/o I have no idea.

[blue2blue]

I think you're on my wavelength, here, but just to make sure no one else might be confused:

 

The particular "latency" I'm talking about here is a completely different issue than latency induced by plug ins -- or overall monitoring latency aggravated by plug in delay compensation.

 

What I'm calling hardware/conversion latency is a result of delays introduced by an audio interface's i/o buffering added to the time it takes to actually do the D/A and A/D conversions necessitated by overdubbing. It happens, of course, whether or not there are software plugs in the processing chain.

 

It manifests itself (for my purposes here) as a 'misalignment' of newly recorded tracks vis a vis the absolute position of existing tracks unless compensated for in some fashion. (Which I do by nudging -- but wish I could do automatically.)

 

 

I belabor this (perhaps ridiculously) because I brought this up on the CW Sonar forum and was met by a huge amount of confusion about what I was fundamentally talking about.

 

Some folks fixated on the measurement test I outlined above, thinking -- despite the fact I said a number of times it was a test to determine the extent of this particular latency issue in my rig -- that I was trying to achieve some sort of effect.

 

It was pretty funny, actually. People can write, but sometimes I wonder if they can read. [Then again, I realize I have a way of overexplaining things that probably is counterproductive. Not to mention I might have -- you proably haven't noticed it yourselves -- a slight tendancy to ramble... oh, wait, I'll save that for your next thread. ]