What's new

Does "near to zero" latency possible playing Virtual Instrument ?

Well, I think, that an "ultra fast" interface is no warranty to get better results with a MOTU m4. I'd think different. Try it on a usb 2 interface. Why? Because I had many issues due to not observed standards with usb 3 interfaces. And MOTU m4 is made for usb 2. USB 2 deserves less speed but in some circumstances better latency. Had some problems with my WACOM tablet on a usb 3 interface. Plug into a usb 2 interface and all problems were gone. Backward compatibility is often only a dream ...
That's a good point and could be a factor but i rather change my interface than change computer...
 
Hey Guys
I was able to start to do some tests with two interface,my Motu M4 (usb) and my new "used" Presonus Quantum 2 ( Thunderbolt 2 ).I'll spare you all the details but i'll give you a quick explanation.
I've use two DAW.One was only use for the recording and the other was use only for the Virtual Instrument Garritan CFX.So the one use for the Virtual Instrument was not affected by the recording. I've always play my piano using the onboard sound of a stage digital piano thru headphone so that was my reference.The setup was,one set of headphone from the RD-2000 and one set of headphone from the interface.i've use one mic per headphone set,i've put each mic as close as possible of the "speaker" from each headphone set.(inside of course).So every time i was hitting a note on the keyboard,two note was recorded and i've calculated and compiled the difference between the two note.The unit that i've use are "sample".I've notice that there was a slight variation in the result for each note,play several times,most probably because of the round-robins and/or the layers...
I've play the same note for about 8 times for each buffer,i've keep the 3 most constant result and calculated the average for each buffer
I'm sure there was a better way to do that kind of test but i got the answer i was looking for...
I've include a humble chart with my result.
 

Attachments

  • Screen Shot 2021-06-12 at 2.12.21 AM.png
    Screen Shot 2021-06-12 at 2.12.21 AM.png
    81.9 KB · Views: 18
So every time i was hitting a note on the keyboard,two note was recorded and i've calculated and compiled the difference between the two note.The unit that i've use are "sample".I've notice that there was a slight variation in the result for each note,play several times,most probably because of the round-robins and/or the layers...
Interesting study! Does the table indicate that the delay of headphone output for Garritan CFX vs. Roland on-board sounds would be:

--Buffer 16
MOTU 453/44,100 ~ 10.2ms
Quantum 160/44,100 ~3.6ms

--Buffer 64
MOTU 567/44,100 ~12.9 ms
Quantum 274/44,100 ~6.2ms

And that the Quantum output is about 6.6ms faster than that of the MOTU USB for these 2 cases?

In real life playing, how low can you get the Garritan CFX buffers without dropouts with these interfaces? I've seen quite a few people typically running buffers of 64 or 128 with consumer USB interfaces with minimal dropouts.
 
Interesting study! Does the table indicate that the delay of headphone output for Garritan CFX vs. Roland on-board sounds would be:

--Buffer 16
MOTU 453/44,100 ~ 10.2ms
Quantum 160/44,100 ~3.6ms

--Buffer 64
MOTU 567/44,100 ~12.9 ms
Quantum 274/44,100 ~6.2ms

And that the Quantum output is about 6.6ms faster than that of the MOTU USB for these 2 cases?

In real life playing, how low can you get the Garritan CFX buffers without dropouts with these interfaces? I've seen quite a few people typically running buffers of 64 or 128 with consumer USB interfaces with minimal dropouts.
I'm gonna do a second test and i'm gonna change a few things:
I'm gonna use ms instead of samples
I'm gonna use a basic virtual instrument sound that don't have any layers
in order to try to minimize the variation that happen during the first test.
Using Digital Performer,i was able to use Garritan+Quantum at the lowest setting possible ( 96 at buffer 32 ) with very little pops and dropouts,not perfect but definitively usable for practicing purpose and no dropouts at all at (96 , buffer 64) which gave me almost the same feel that i've when i'm using my RD-2000.
Using Garritan,the Quantum beat the M4 easily,before doing the test,i've compare both interface for about a hour,switching from interface every 5 mins and i could "feel" the difference.The test have just confirm what i've "felt"
 
Using Garritan,the Quantum beat the M4 easily,before doing the test,i've compare both interface for about a hour,switching from interface every 5 mins and i could "feel" the difference.The test have just confirm what i've "felt"
no dropouts at all at (96 , buffer 64) which gave me almost the same feel that i've when i'm using my RD-2000.
That is excellent performance. Glad this worked out for you.
 
I did another "batch" of test and got very interesting result.
For those tests i've use two "static sound" in order to try to avoid the variation that i had between the same note played few times in a row,so no layering or round-robins,the sound from the RD-2000 was always identical and i did the same with a sound from Kontakt.
Even doing this,there was still a good variation in the timing of the note so that variation is not cause by the layer or the round-robins...
I've notice that my result were "worse" compare to my previous tests and i could not understand why...
After much thinking,i've realize that,unlike my previous test where i was comparing the RD-2000 to Garritan,this time i was comparing the RD-2000 to a sound using Kontakt.
The Kontakt "engine" seem to be a "slow one".
It seem to had some internal latency compare to the Garritan's engine and other engine that i've also tested.
And it make sense to me cause the first piano library that i've ever bought was NOIRE.
I was not crazy about the sound but the "deal breaker" for me was the playability.I was never able to adapt to the "feel" of NOIRE,i was noticing too much "delay" or latency...
Few months later,i took another chance and bought Garritan.I like the sound better but more importantly the playability was definitively better than the NOIRE one but i had never make the link that they use different "engine" and the latest test confirm that theory.
And since the Quantum was performing better,i was wondering if it would do even better i was using it in a different DAW than Digital Performer...
So i've download and install Studio One 5
For that test i've return to a RD-2000 to Garritan comparaison and the result were better than with Digital Performer
To a point were at the best setting,my Garritan "act or feel" like my Roland...same responsiveness.
Exactly what i was looking for
All the result in the chart are the difference in ms between the sound of the RD-2000 vs the sound of Garritan.
 

Attachments

  • test 3.png
    test 3.png
    67.7 KB · Views: 15
Top Bottom