Does Sonar compensate for latency recording?

I posted more or less the same problem (mine was for using external effects) here
http://www.cakewalk.com/forum/tm.asp?m=163701&mpage=1&key=&anchor#164252
and got no reply

I contacted Cakewalk tech support and was told that this was a driver problem, as well as the time it takes to make a D/A conversion and then a A/D conversion and nothing to do with Sonar.

I was told that Sonar records exactly when it hears it, but that's just not happening.

I get a 67 sample delay when using optical out to an FX unit and back in optically- no A/D or D/A happening there is there? I get a 135 sample delay when going out of analog out to analog in (I understand that because of the conversions, but not the digital out to ins.)
I've tried running straight out and in without going through any effects and the times are the same-67 samples for digital to digital, 135 for analog to analog.
I'm using an Echo Layla24 and a Gina24 on another PC.

These times seem absolute-no changes anywhere seem to alter them. What I wondered was if they were indeed absolute. Then I could just slip those tracks along by that amount (a bit of work, but it won't break my back!!)

I also was wondering if it were possible to write some sort of CAL script that could do this for me . I was thinking would it be possible to modify one of the CAL programs (one let you nudge a track one tick left (or right). I presume the "ticks" would depend on how you have your midi resolution set up. How would you change that to samples?


I'm a bit distressed now to here that Cubase/Nuendo doesn't have this problem- Cake Tech said all DAWS do.

Yes I did Sani- Eco Layla24 and Gina 24.

thats interesting,

i wondered about (and posted) the exact same problem last year and nobody responded.

But now there is a thread called "Article on Sample Accurate Recording" that really goes into this topic.

what makes me a bit sad is, that there was the conclusion that every sequenzer has this problem.
but apparently the cubase users dont have to slide their recordings everytime they have recorded material, because it autoslides their recordings!!!!

i recommend reading the article and please you to post comparisons if you have sonar and cubase. Id also like to hear Ron on that topic

Jeff

Here is the link:

http://www.cakewalk.com/forum/tm.asp?m=130201&mpage=2&key=article&anchor#174528

Ron already posted some ideas (AD/DA conversion) but if Cubase´s Recording are on time there is no excuse for Sonar s Recordings being late and out of time

since i encountered this problem (and many others) a long time ago i found this issue VERY important and annoying...

Like you say there is no driver problem but some kind of calibrating problem

Jeff

Sliding is not an option! I can slide a drumtrack or a percussive instrument, but I don't want to spend my time to slied the vocals or a guitar to its "right" place.

Ok, we are not three users with different but known soundcards that can be considered as pro or better semipro. I'd like to hear somebody who made the test and can say that the second wave is sample accurate in sonar.

you can read the thread, there are some tests written down.

ORIGINAL: Xavier

I am really glad this was brought up because I have fought the same issues...

Nice work djkepi!!!
Now I can figure out how to line my tracks up more accurately than by ear like I have been doing...


As far as a fix from Cakewalk (and other vendors for their software), it seems to me that one should be able to calibrate each sound card in the system by routing a signal out of the soundcard and back in, recording at the same time with a specific test signal that comes with Sonar. I realize that the delay out may be different than the delay in, but there must be ways to detect that or do a calibration differently. Calibration is the key word here. You do it once and no more monkying around...

All this wave form sliding seems rediculous to me. The software should automatically compensate for this. Sonar already knows what sound card one is recording through, so (if there are more than one) each sound card's calibration factors could be automatically used (or each individual channel if they vary that much).

The software would probably have TWO record buttons then: "Record External" or "Record via Sound Card Loop". One would use "Record External" when tracking more external instruments like we normally do. "Record via Sound Card Loop" would be used when routing a signal out of Sonar, through some external FX or live reverb chamber, then back into Sonar. Then everything would be automagically lined up.

Any thoughts on the viability of this before I submit a request?


<X>

P.S. At the same time there is probaby a way to use pink or white noise in the same loop to check for frequency related phase issues, and the like, of our sound card, and then even test frequency related phase issues of EQ's and other plugins. Of course add a calibrated mic and we could check our speakers and rooms. Probaby not something we would get directy from Cakewalk, especially anytime soon, but I can wish...

I found this an interesting thought from Xavier. That's just what I'm looking for.

If I run the same test with Sonar 3.1.1 on my RME HDSP 9632 card
with RME's v 2.70 ASIO drivers, I can detect no skew whatesoever
on an analog in/out loopback, even if I zoom to the point of being able
to see individual sample points.

There *is* intrinsic delay in the output/input process, and it's dependant
on the design of the card. It's up to the driver to either compensate
or provide Sonar with compensated timing. It sounds like RME's
drivers do this (no suprise, their driver support is one of the
best things about using their gear), but that the other vendors
do not, at least not as perfectly.

we should all test this..

at least with my M-Audio cards (Delta 44, Mobile Pre) there is no sample accuracy..

jeff

thanks for the info...

i will try these settings - maybe its that simple

jeff

Out of curiousity, sani, are you using *exaclty* the same driver settings
for clocks, timing, etc, between Cubase and Sonar? The two programs
may have different defaults, and Sonar's may be suboptimal for your
driver. I think that's at least as likely as my previous theory about
different drivers having different compensation techniques - though
one can't rule that out, either.

Here's the big worry for me-
If the loop test shows up this latency- That means that after your original click track is down, any recordings/overdubs made using that track as a guide will be the 135 samples (or whatever your soundcard is showing) behind. Now, that's less than 2 ms but it's still enough to ruin a groove. I'd really love everybody to do this loop test with a hi-hat and post their findings.
What I'm saying is that not only is this a problem when looping. Logic will tell you it's also happening when you track-affecting everyone, not just external effects users.


Like I said- if I knew this was just a fact of life (it used to be a fact of life with ProTools) I'd just slide it over by 135 samples. It wasn't a big problem then and it wouldn't be a big problem to do now. All I really want to know-from Cakewalk or EchoAudio- is if this is the way it is-a fact. I have a tech support e-mail in with EchoAudio, so I'll also post their reply here.

My Layla 24 plays back late as well. I know digi 001 also have this problem when used with Pro Tools, but I hear Cubase doesn't have this problem. Sonar 4 needs to address this.

My only workaround is to record/bounce everything at the same latency setting. This usually mean choosing a high enough latency to record as many tracks as you want (not really a painful workaround to me). Another thing is to stay minimal on the softsynths or cope with the high latency or play the synths in a non real-time manner (AGONY!).

ORIGINAL: sani
While trying both programs to do the same thing, I didn't change the asio settings for the audiocards. I'm glad to hear that your system behaves as it should. My both programs are set to use the clock of the audiocard. In Cubase it doesn't matter how high or low I set the latency. I could set the latency for hundreds of miliseconds and the record would still be sample accurate, but not in Sonar.

My synchronization does not seem to be dependent on the configured record
latency, either. Indeed, Sonar and/or the drivers of just about all audio cards
manage to compensate for that sort of buffering latency just fine. You have
not actually posted what delays you are seeing with sonar, but I doubt that
it's tracking the record latency (and if it is, I suspect you've got a configuration
problem elsewhere).

It's unfortunate that the old Cakewalk NNTP server newsgroup Product.Sonar
is gone and was never archived, because a bunch of us went through this
exercise last year. It does seem to vary with the hardware and the driver.
My recollection is that the driver is supposed to communicate to the host
program (Sonar in this case) how much delay is in the output and input
paths, so that the program can act accordingly. Some drivers are calibrated
very carefully (as RME's seem to be), while others, most notoriously the
Creative cards, provide nothing useful. If I recall correctly, Sonar trusts
the driver to provide the right information. Cubase may do something
different, like have a table of values for different leading cards that were
measured by Steinberg, or perform some sort of automated measurement.
I don't know that that's true, but it would be a smart thing to do if you know
you're going to be selling to a lot of people with consumer-grade sound cards.
This could create a vicious circle, where because Cubase doesn't require
the drivers to do the right thing, driver writers stop bothering to do it.

But at the end, if cubase works well with the asiodrivers of the card and sonar not, it still means to me, that sonar is not quite asio compatible as it should be.

Cakewalk doesn't like to admit it, but Steinberg defines ASIO, and it's
tough for anyone else to be quite as compatible. The only real Cakewalk
ASIO bug that I know of in the current product is the off-by-one-buffer-latency
MIDI timing thing that can be worked around easily in the Audio settings,
but this could be another one. Cakewalk's engineers are really great guys
when it comes to fixing a real bug. The problem is that there are so many
cases of operator error and/or third party components causing false alarms
that it can be difficult to get them to accept a bug as real.

I made further investigation yesterday and here are my results:

I downloaded the demos audition (former cool edit), sony vegas, sony acid 4 (I guess). I have forgotten to download samplitude.
Before I started I set my terratec soundcard to 1024 samples latency. I just wanted to try how the programs would or would not compensate.
I set the drivers in every program, loaded a audiofile and made the test: from audio output thru the mixer and back into the soundcard.
The result: audition, vegas, acid doesn't have any latency!!! In Sonar I don't even need to zoom in, I see the late recording.

One more thing: terratec says on their site that the drivers are absolutely sample accurate in the input/output recording.

My conclusion is that it is not the fault of badly written drivers, that Cubase is not the only program able to compensate (it seems that most other software do this as well) and that Sonar has a problem.

People with a very fast computer can set their latency down to 2-4 ms and maybe they don't see this problem, but at this time I will not use Sonar anymore for anything else but making some loops. BTW, waiting for Sonar4 to handle this is IMHO silly. Sonar is dedicated to record and track music. Accurate recording is the most important feature above all other things we want to see in the next version. Music happens in a linear time and Sonar should support it and not destroy it. If a non-dedicated multitracking program like Acid can do this better than it is a shame. I have done everything I could to see where the problem is and it seems that it is not the fault of my soundcard.

It's not just an ASIO problem. I use WDM exclusively.