• Global community
    • Language:
      • Deutsch
      • English
      • Español
      • Français
      • Português
  • 日本語コミュニティ
    Dedicated community for Japanese speakers
  • 한국 커뮤니티
    Dedicated community for Korean speakers
Exit
0

How to make latency compensation work? OR - suggestions for a small interface that DOES work?

Guest
Nov 29, 2011 Nov 29, 2011

Copy link to clipboard

Copied

I'm using a Roland UA-3FX interface with Audition 5.5 on windows 7.

If I load in a drum loop on one track and then record some bass on another track (monitoring direct though my interface), the recorded bass is slightly but noticably delayed from the drums.

As I understand it, the ASIO driver and Audition, between them, should be working out any latency and compensating for it.Any settings I can fiddle with? Can't find anything.

I did try using the ASIO4All driver which does have adjustable latency compensation in its settings, but this either wasn't enough or didn't do anything at all (not sure which!)

If it's an issue with the driver for the UA-3FX, can anyone else recommend a good small interface which can record in time?it's a shame 'cos the UA-3FX actually sounds great.

Ivan

Views

24.5K

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 29, 2011 Nov 29, 2011

Copy link to clipboard

Copied

I use a UA-3FX with Windows 7 64-bit, but am still using Audition 3.01 since the trial of CS5.5 expired. Unfortunately, I didn't really test for latency during the trial period, but all I can say is that I have no problems in 3.0 here. I did some tests quite recently by playing back audio on one track through phono cables between the UA-3FX's output and input and recording on another track. I couldn't detect any latency at all, which surprised me, so I, too, would be interested in learning more about latency compensation. All the effects were switched off for the test, of course.

You must be using the Advanced driver setting on the UA-3FX to be using asio, and the only other thing I can think of is maybe some usb problem. That Roland is a usb 1.1 device, and I had huge problems with the machine with an ATI chipset until I found the correct ATI "usb filter" driver. There seem to be different versions of this driver for different generations of chipset.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 29, 2011 Nov 29, 2011

Copy link to clipboard

Copied

Thanks for the reply. I am definitely using the advanced mode and the ASIO driver.

The USB Chipset is Intel. I suppose I could mess with USB drivers but unless there's a promising-looking solution, that's usually deeper than I like to dig (I was an IT guy in a previous life, and it's taught me to give up fairly fast - yes, I might have fixed it after 4 days, but you can't get the 4 days back!)

I think I have worked out that the offset is about 30ms, so I'll just try and find a quick and easy way of moving recorded tracks back 30ms. (I'll ask that in another question).

I would still be interested if anyone can suggest an interface that they have working in 5.5 which works properly in this respect.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

topomorto wrote:

As I understand it, the ASIO driver and Audition, between them, should be working out any latency and compensating for it.Any settings I can fiddle with? Can't find anything.

Perhaps you can tell me exactly what the mechanism might be for this to be worked out? I can see that it could be compensated for, but how, without a setup procedure, could you measure it? What is compensated for is VST delay, but I'm not aware that Audition makes any attempt to run tracks early to compensate for direct monitoring - which is why there are so many complaints about it.

I'm not saying it's impossible, because clearly it isn't. But without Audition having some sort of means of measuring round-trip latency (which would be quite complex, because you would need to time a known-start signal though the interface and operating system), I don't think that compensation can be any more than a guess.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

I thought the basic idea is that the ASIO driver reports the inbound and outbound latency. Assuming the manufacturer writes the driver, they are in a privileged position as they know the basic hardware latency - and the driver also knows the current buffer sizes. The DAW can then compensate for that, by 'playing early' and moving recordings forward slightly.

This article (from NI) talks about the ‘things that can be known’ :

http://www.native-instruments.com/knowledge/questions/748/%22Overall+Latency%22+Demystified%3A+why+d...

The ‘processing’ latency shouldn’t really be an issue in my case.

It may be true that there is still some guesswork involved (e.g. time it takes for the data to move through the OS, as described in the NI article) - but in that case, there needs to be a user definable adjustment somewhere to compensate for it. I guess this should really be on the driver, but seeing as it apparently doesn’t always work, it would be nice if Audition had it as a user-definable setting too (“Ignore reported latency and use these figures instead”) which you could calibrate by doing a manual round - trip test with an impulse or drum sound or something, recording from one track to another (which I might try later, actually, to see if I can refine my 30ms figure)

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

aaLatency1.jpg

What we have here is a screenshot showing that the audio (a square wave) passing through Audition 3.01 via my UA-3FX using asio has negative latency.

The top track is the original The UA-3FX is plugged analog output to input. Track 2 is the recording of this signal using asio.

Track 3 doesn't show the signal recorded through the same hardware loop using the WDM drivers. The latency in that case is approximately 70 ms so the square wave is way off to the right.

There obviously is some sort of compenstion occurring somewhere when asio drivers are in use.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

Interesting, thanks a lot for that!

If I'm reading that right, that means that Audition 3, with your system, is compensating correctly to within one sample (in fact, over-compensating by one sample), which to my mind is very good.

I will do the same test when I get home from work.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

OK, So I did the same in cs 5.5.

nolatcomp.png

1) Green - original

2) Pink - this is through the UA-3FX and its native driver. As you can see, there is a just over 30ms delay.

3) and 4) Blue - this is the UA-3FX using the Asio4All driver. The first one is with 2048 samples of latency reported in Asio4all (1024 in, 1024 out), the second with none - as you can see, they are no different.

5) and 6) Yellow - this is my internal soundcard using the Asio4All driver. Again, the first one is with 2048 samples of latency reported in Asio4all the second with none - again, they are no different.

So, if I am reading it right:

- Audition CS5.5 is not compensating properly, as tested with two different soundcards

- It seems that Audition CS5.5 is not paying attention to the reported latency from the driver at all, insofar as changing the reported latency on Asio4all makes no difference.

This is just one guy and one system...

I'd be really interested if someone else with 5.5 could do the same test - that is, load up a suitable spiky audio file in one track, join your soundcard output to the input, and record it to another track.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

I can find no evidence that ASIO inherently reports anything to do with latency at all. What I did find though was a utility that will let you measure it: Latency Test Utility. This appears to work in the manner I suggested, replacing in this case Audition with its own software.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

SteveG,

From native instruments:

"Note that the values shown in this dialog window are based on what the driver reports to the software. There are some drivers which do not report the correct information, hence you might find unexpected values shown with some drivers."

http://www.native-instruments.com/knowledge/questions/748/%22Overall+Latency%22+Demystified%3A+why+d...

From asio4all:

"The values displayed there are the values ASIO4ALL reports for the purpose of latency compensation. This means that recorded audio can be aligned perfectly with the rest and if that does not work out of the box, you can use the "Latency Compensation" sliders in the advanced section of the control panel to make it perfect!"

http://www.asio4all.com/ufaq.html

Of course I'm sure there is more I can learn about this, but there does seem to be such a mechanism.

Wild_Duck's tests suggest that there is some mechanism for compensation - it could be just a guess, but in that case he's really lucky that he gets his round-tripped signal back again within 1 sample of where it came from.

Steve, do you have 5.5? Would you be kind enough to take a minute or two to do the same tests and post the results here?

If this is just my issue, it's just my issue - but if CS5.5 actually can't record in time, it's something we should all know about!

Cheers,

Ivan

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

I have run the Centrance latency tester on the 64-bit Windows 7 machine, and it reports the latency as 24.17 ms.  I remain confused, but the fact is that I can play guitar happily along to other tracks in Audition 3, and I don't have a latency problem.

Just for the record, I did correspond some time ago with someone who also played guitar and used Audition 3 with an Edirol UA-4FX. He was also happy.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

topomorto wrote:

From native instruments:

"Note that the values shown in this dialog window are based on what the driver reports to the software. There are some drivers which do not report the correct information, hence you might find unexpected values shown with some drivers."

That's reported, not measured... and only applies to the device itself.

From asio4all:

"The values displayed there are the values ASIO4ALL reports for the purpose of latency compensation. This means that recorded audio can be aligned perfectly with the rest and if that does not work out of the box, you can use the "Latency Compensation" sliders in the advanced section of the control panel to make it perfect!"

ASIO4ALL isn't a driver. It's a conversion utility that interfaces ASIO software with native Windows drivers on sound devices - not the same thing at all. If it reports values, it is only because it's been given them by a hardware device, which it passes on.

Yes I have 5.5 (and every other version of Cool Edit/Audition there's ever been...) and if I get a chance over the next day or so I'll check it with the MOTU 8pre that's currently connected to the DAW.

Yes I know that the Centrance latency checker said XP on it, but doesn't most XP software run in emulation mode on that pile of cr..  W7?

I spent some time yesterday 'persuading' W7 to let me do some rather basic things with file sharing - and managed easily to provoke a BSD out of it - twice. It really isn't very good, and won't be going near any DAW I own any time soon.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Nov 30, 2011 Nov 30, 2011

Copy link to clipboard

Copied

I know that the latencies aren't measured, live - that's fine.

Actually I think I found the place in the ASIO code that gets them... ASIOGetLatencies().

http://www.lasi.arizona.edu/Dist%20Disk%201/ASIO%20Distribution%20Disk/Audio/asio.h

"inputLatency will hold the 'age' of the first sample frame

  in the input buffer when the hosts reads it in bufferSwitch()

  (this is theoretical, meaning it does not include the overhead

  and delay between the actual physical switch, and the time

  when bufferSitch() enters).

  This will usually be the size of one block in sample frames, plus

  device specific latencies.

  outputLatency will specify the time between the buffer switch,

  and the time when the next play buffer will start to sound."

If I read that right, It says there that it's not perfect, but it does include the buffer size ("size of one block in sample frames") and what the manufacturer knows about the hardware ("device specific latencies")

ASIO4All gets the values it reports from... us!

Latency Compensation, in the top right corner.

http://img.brothersoft.com/screenshots/softimage/a/asio4all-274791-1250587869.jpeg

But as you can see from my measuring results, changing those sliders doesn't make any difference in Audition, which I don't understand.

As I mentioned before, I'll try some other software when I get a chance.

It would be great if you could repeat the experiment I and Wild_Duck have done... would really appreciate it! 

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Dec 02, 2011 Dec 02, 2011

Copy link to clipboard

Copied

Okay, CS5.5 latency tests with MOTU ASIO driver, and also ASIO4ALL drive converter: Most importantly, this is using XP though, not that pox-ridden W7!

One 5-second file, with a 0.02s square tone burst in the middle of it - easy to measure the edges accurately from. Play it out of track 1, and re-record it on track two with a loop-through on the back of the 8Pre.

With the ASIO driver, the ASIO buffer size works best at 512 samples - this goes for using ASIO4ALL as well. Using the MOTU driver, latency is consistent, at 57 samples. Checked this on AA3.0.1, and it's exactly the same.

Installed ASIO4ALL driver (don't need it on the DAW, but don't mind it being there). With no latency compensation at all, the round-trip latency is 1194 samples - about 27ms. If you whack the input slider over to 1024 samples, then the latency improves to 200 samples - about 4.5ms. This doesn't necessarily mean that there's an error - it just means that not all of the latency is on the input. If you then set the output slider to 1024 as well, you get overcompensation - the resultant file is recorded early - about -18ms. What you'd have to do in this situation is to juggle both of the faders so that the input and output compensation values were similar - after all, it's a round trip you are considering. It would be relatively easy to get it down to a value that would be completely insignificant.

But, as I mentioned earlier, this is with XP - an OS far better suited to audio than W7, whatever its apologists say. And also this is using a known-good MOTU external audio device. I don't have the means to run the same checks using W7 here on the premises, and even if I used the machine that's potentially available, it wouldn't be the same because it's a different set of PC hardware.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 02, 2011 Dec 02, 2011

Copy link to clipboard

Copied

Great, thanks a lot. Sounds like, in your system, Audition does seem to be compensating (otherwise you'd have a round-trip latency bigger than your buffer), further evidenced by the fact that it's paying attention to the ASIO4ALL sliders - which for me don't make a blind bit of difference.

Nearly the weekend... I will see if I can do the same test in Ableton Live or some other DAW on my win7 system, and take things from there.

All good fun!

EDIT: as your results (and instincts!) suggested there may be a win7 problem, I did a google search...

dragged up a couple of threads:

http://www.sevenforums.com/sound-audio/41314-delay-sound-coming-invalid-asio-driver-3.html

http://forum.cockos.com/showthread.php?t=90912

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Engaged ,
Dec 02, 2011 Dec 02, 2011

Copy link to clipboard

Copied

SteveG(AudioMasters) wrote:

Yes I have 5.5 (and every other version of Cool Edit/Audition there's ever been...) and if I get a chance over the next day or so I'll check it with the MOTU 8pre that's currently connected to the DAW.

I spent some time yesterday 'persuading' W7 to let me do some rather basic things with file sharing - and managed easily to provoke a BSD out of it - twice. It really isn't very good, and won't be going near any DAW I own any time soon.

Well, the fact that you're not going near W7 would explain, then, why you're running every version of SCEPAA (Syntrillium Cool Edit Pro Adobe Audition) - XP can run all of them, right?

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

therealdobro wrote:

Well, the fact that you're not going near W7 would explain, then, why you're running every version of SCEPAA (Syntrillium Cool Edit Pro Adobe Audition) - XP can run all of them, right?

Nearly right. I don't have all of them on one machine,  but XP will run more of them properly than any other OS, yes.

And it's not that I haven't gone near W7 - far from it. Only the other day I got a BSD from it, just from altering some of their ridiculous file permissions... any so-called 'OS' that does that doesn't deserve houseroom.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

SteveG(AudioMasters) wrote:

And it's not that I haven't gone near W7 - far from it. Only the other day I got a BSD from it, just from altering some of their ridiculous file permissions... any so-called 'OS' that does that doesn't deserve houseroom.

          You seem to feel about win7 the same way I feel about our drummer's Rottweiler!

          Audio matters aside - and obviously I am not feeling well-disposed to it on that front ATM - I've always thought Windows7's overall stability is very good, on the scale of things...

          I work in a team of about 25, all on Windows 7, all on different hardware... only BSOD's I can think of were where there was some underlying hardware problem (graphic card fan had died, disk was on the way out, etc.).

          Which doesn't mean it doesn't hate some people, I'm sure....(Like me and the Rottweiler)

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Community Expert ,
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

Well, the problem was on a laptop that came pre-installed with it. And if the Rottweiler bit you, then you're quite entitled to be anti-Rotweiller, for just the same reason!

If W7 is a Rotweiller, then in comparison, XP is an Old English Sheepdog, I think. Altogether more pleasant and useful.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

Yes, W7 is in some ways a dog, but back to business.......

Before getting too much into workarounds, the thing I would try is going into the Windows Control Panel audio section and disabling the UA-3FX in both record and playback tabs. This, to a great extent, removes W7's audio from the equation. You will then have to use only the Edirol asio drivers.

Some of this suggests to me that it might be a usb problem rather than a driver problem. I'm a great believer in xperf and logman as ways of seeing usb problems, if not actually resolving anything.

I've done a quick test here with Reaper and have no problems with latency with that either.

There was a similar thread on the Audiomasters forum from someone in Helsinki. He doesn't seem to have posted since going off to try XP.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

the thing I would try is going into the Windows Control Panel audio section and disabling the UA-3FX in both record and playback tabs. This, to a great extent, removes W7's audio from the equation. You will then have to use only the Edirol asio drivers.

               Tried that already... made no difference, sadly!

                I also tried the opposite approach - using the MME drivers to see if they'd be compensated for correctly, despite the longer latency, but... no.

Some of this suggests to me that it might be a usb problem rather than a driver problem. I'm a great believer in xperf and logman as ways of seeing usb problems, if not actually resolving anything.

          Could be,,, to be honest I'm not inclined to spend too much more time going into that level of troubleshooting - unless you have any to-the-point links you'd be kind enough to share?

          After all, it only takes a few seconds per track to apply the delay, solo and bounce to fix things... not ideal, but at least the audio is clean!

         

There was a similar thread on the Audiomasters forum from someone in Helsinki. He doesn't seem to have posted since going off to try XP.

          I checked out what my chances are of getting this laptop to run XP, but they don't look good.... something to do with the BIOS.

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

Topomorto, just to check - you did disable the UA-3FX, not just make it not the default device? If you did and it makes no difference, it probably eliminates any weird effects from the Windows audio side of things.

Once set up to run from batch files, xperf is easy to use and does provide a useful insight into what is going on with usb.The zoomable graphs with a timeline do give a picture of usb events and let you see whether they coincide with many other events in the machine. One definitive Microsoft document can be found by googling USB Event Tracing for Windows, which brings up a rather heavy Microsoft paper on the subject. Alternatively, if you PM me, I can send some instructions.

I understnd the concept of getting on and using workarounds, but I like to be sure what the bug is that I'm working around and whether it's consistent or likely to bite me in future. It took me ages and much posting of questions to get my Win 7 laptop working well, but now I use it without thinking and it "just works"..

The asio code tht you posted about a few days ago looked really useful, so thnks for that. I need more time (I have other non-audio problems at the moment caused by recent high winds).


Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

Wild_Duck wrote:

Topomorto, just to check - you did disable the UA-3FX, not just make it not the default device? If you did and it makes no difference, it probably eliminates any weird effects from the Windows audio side of things. 

          Definitely disabled it in both tabs in the audio control panel. For a giggle I also tried disabling it in device manager to see if asio4all could magically find it (it can't )

         

Once set up to run from batch files, xperf is easy to use and does provide a useful insight into what is going on with usb.The zoomable graphs with a timeline do give a picture of usb events and let you see whether they coincide with many other events in the machine. One definitive Microsoft document can be found by googling USB Event Tracing for Windows, which brings up a rather heavy Microsoft paper on the subject. Alternatively, if you PM me, I can send some instructions.

I understnd the concept of getting on and using workarounds, but I like to be sure what the bug is that I'm working around and whether it's consistent or likely to bite me in future. It took me ages and much posting of questions to get my Win 7 laptop working well, but now I use it without thinking and it "just works"..


          Sounds really useful. I have a band session coming up this week and need to sort out some Xmas stuff so probably can't mess with it too much for a while - maybe over the christmas hols?

The asio code tht you posted about a few days ago looked really useful, so thnks for that. I need more time (I have other non-audio problems at the moment caused by recent high winds).

          I won't pretend that I truly understand it... I do some coding but not driver-level stuff!   If time was really unlimited I'd write my own demo recording program and see if that held any clues!

          Are you in Western USA? 

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

topomorto wrote:

          Are you in Western USA? 

No. in the UK, but don't let that put you off!

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Guest
Dec 03, 2011 Dec 03, 2011

Copy link to clipboard

Copied

Me too... no winds in Cambridge though, so wondered what i'd missed.... just typed 'winds' into google news!

Votes

Translate

Translate

Report

Report
Community guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines