Re: GP-10 - GUITAR TO MIDI Best Results?

Started by jassy, August 14, 2016, 11:46:50 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

jassy

I keep hearing that "much better Roland Mac drivers" and wile I think the VG99 and GR55 where not the best when I tried it and compared with RME results, Im curious about real RTL measurements with the modern Roland/Boss units like the GP10.
I used the VG99 doing audio to midi playing soft synths and the results were not the best but pretty decent anyway. I would not call that "unusable" at all. I think I have a recording playing synths with that if I find it I can upload to illustrate what I say.
About the new units, tried briefly the SY300 (returned it after a month) and in the driver side I got a good feeling with regard to use the unit as a sound card at very low latencies (64 samples).
Now, finally I got an GP10 (found a good price), and the preliminary results are very good. I can use it with my Asus laptop (i7 3610QM) at 64 samples without problems and it looks quite efficient, only slightly more than my RME Babyface. Can load several amp/effects plugins (GRig 5, Bias FX AND others), several good quality effects plugins, midi guitar and various softsynths all that Simultaneously without clicks and pops and not feeling latency while playing (for a reference on my IPad air and biasfx I feel too much latency even at the lowest latency possible). That amount of things running simultaneously is very important to know, because it's not all in the RTL numbers but also what I can run at that latency.
I decide to do the RTL test and the results at 64 samples/44,1 khz is 8.5 ms.
With the Babyface the RTL results at 64 s/44.1 khz is 5,6 ms.
Well It's obvious it uses some hidden additional buffers but does not seem so bad after all.
Im curious if it uses some hidden buffers also in the Mac side, what are the RTL numbers and how much load at it?


aliensporebomb

How are you connecting your devices to the Mac?  What Mac?  OS?  CPU?  Memory?  Host app?

Your best performance will come via MIDI over USB in my experience.

In other words:  USB out from the VG99 directly to a USB port on the Mac.

The worst performance will come from using the MIDI ports on the 99 or FC300 to the MIDI in on a midi interface connected to the Mac.

And make sure you have your GK Sense levels set to around 20 for best performance in doing guitar-to-midi operation.
My music projects online at http://www.aliensporebomb.com/

GK Devices:  Roland VG-99, Boss GP-10, Boss SY-1000.

MarkusB

Thanks Jassy, hard data is much appreciated!

alexmcginness

#3
You completely missed what I was pointing out. The Mac VG drivers will give you better tracking than the PC drivers. Theres far less glitching and tracking errors with the Mac than the PC because of the drivers.
   Ive got a fairly high powered computer setup for my music machine. It's a Windows 7, HP Z800 dual quad core XEON with 72 gigs of ram and a MOTU PCIe card with 3 -24I/O boxes. I can track without problems at 64 buffers for low latency but the issue isnt the latency its the accuracy of the Guitar 2 Midi conversion on the PC. If I plug the VG-99 or the GP 10 into the Mac the latency is more than useable but the tracking errors all but disappear.
   I use my AXON AX-100 MKII to do bass lines into the PC as its the fastest tracking box I currently own, but I use the Mac with either the 99 or the GP10 to do chord stuff as theres less tracking errors.


Quote from:  jassy on August 14, 2016, 11:46:50 AM
I keep hearing that "much better Roland Mac drivers" and wile I think the VG99 and GR55 where not the best when I tried it and compared with RME results, Im curious about real RTL measurements with the modern Roland/Boss units like the GP10.
I used the VG99 doing audio to midi playing soft synths and the results were not the best but pretty decent anyway. I would not call that "unusable" at all. I think I have a recording playing synths with that if I find it I can upload to illustrate what I say.
About the new units, tried briefly the SY300 (returned it after a month) and in the driver side I got a good feeling with regard to use the unit as a sound card at very low latencies (64 samples).
Now, finally I got an GP10 (found a good price), and the preliminary results are very good. I can use it with my Asus laptop (i7 3610QM) at 64 samples without problems and it looks quite efficient, only slightly more than my RME Babyface. Can load several amp/effects plugins (GRig 5, Bias FX AND others), several good quality effects plugins, midi guitar and various softsynths all that Simultaneously without clicks and pops and not feeling latency while playing (for a reference on my IPad air and biasfx I feel too much latency even at the lowest latency possible). That amount of things running simultaneously is very important to know, because it's not all in the RTL numbers but also what I can run at that latency.
I decide to do the RTL test and the results at 64 samples/44,1 khz is 8.5 ms.
With the Babyface the RTL results at 64 s/44.1 khz is 5,6 ms.
Well It's obvious it uses some hidden additional buffers but does not seem so bad after all.
Im curious if it uses some hidden buffers also in the Mac side, what are the RTL numbers and how much load at it?
VG-88V2, GR-50, GR-55, 4 X VG-99s,2 X FC-300,  2 X GP-10 AXON AX 100 MKII, FISHMAN TRIPLE PLAY,MIDX-10, MIDX-20, AVID 11 RACK, BEHRINGER FCB 1010, LIVID GUITAR WING, ROLAND US-20, 3 X GUYATONE TO-2. MARSHALL BLUESBREAKER, SERBIAN ELIMINATOR AMP. GR-33.

jassy

Quote from:  alexmcginness on August 16, 2016, 09:31:27 AM
You completely missed what I was pointing out. The Mac VG drivers will give you better tracking than the PC drivers. Theres far less glitching and tracking errors with the Mac than the PC because of the drivers.
   Ive got a fairly high powered computer setup for my music machine. It's a Windows 7, HP Z800 dual quad core XEON with 72 gigs of ram and a MOTU PCIe card with 3 -24I/O boxes. I can track without problems at 64 buffers for low latency but the issue isnt the latency its the accuracy of the Guitar 2 Midi conversion on the PC. If I plug the VG-99 or the GP 10 into the Mac the latency is more than useable but the tracking errors all but disappear.
   I use my AXON AX-100 MKII to do bass lines into the PC as its the fastest tracking box I currently own, but I use the Mac with either the 99 or the GP10 to do chord stuff as theres less tracking errors.
Sorry but I cant understand that at all. The audio to midi conversión is done with the VG/GR/GP device itself, not the driver or the computer. After the conversión is done in the unit, the driver can only be responsible to do the data transfer at the lowest buffer possible (latency) and the most efficient way (cpu hit, less important if we have an actual computer), so apart of that how can be the driver or the computer be responsible of anything?
Could you do the RTL test with your mac in order to know some objetive numbers? maybe we have here some answers.
Its very easy:
Just download the RTL utility here http://www.oblique-audio.com/free/rtlutility
Connect your L/mono out to the guitar input of the GP10 and run the utility, thats all.
Say your Os and pc specs and let us know the results.
To other GP10 users: please if you have a momment do the test and maybe we can have some light to this question.

Elantric

#5
QuoteThe audio to midi conversión is done with the VG/GR/GP device itself, not the driver or the computer.

They are all part of a system and i confirm same results as Alex McGinness here:
https://www.vguitarforums.com/smf/index.php?topic=18729.msg133591#msg133591
Quotethe strength of the mac lies in the roland drivers for it vs the drivers for the PC. The Mac drivers work far better for the GP 10 the GR 55 and especially the VG-99. Guitar to midi using the 99 on a PC just sucks. On the Mac its totally useable and again. this is coming from a dedicated PC guy. The macbook I have is a 2.4 ghz intel core i7 with 8 gigs of ram  and OSX 10.95. late 2011 15" screen. It use is primarilty a guitar to midi/soft synth interface. I have an Axon AX 100 mkII that I use for bass lines as its faster on the lower strings.
   Again, find someone with a fast enough mac and see for yourself. I paid $950 CDN for the used Mac and some can find them cheaper.


QuoteCould you do the RTL test with your mac in order to know some objetive numbers? maybe we have here some answers.
Its very easy:
Just download the RTL utility here http://www.oblique-audio.com/free/rtlutility
Windows only

http://www.oblique-audio.com/free/rtlutility

FWIW - for real world Audio latency measurement -  I only trust this


And explain how here:
https://www.vguitarforums.com/smf/index.php?topic=8929.msg133954#msg133954


jassy

Quote from:  Elantric on August 16, 2016, 10:08:26 AM
They are all part of a system and i confirm same results as Alex McGinness here:
https://www.vguitarforums.com/smf/index.php?topic=18729.msg134499#msg134499
So you are saying that the guitar2midi conversion is done in some way in the computer? So if we disconnect the computer then there is not conversion possible?
Maybe the driver is sending something more than simple midi data (in this midi2guitar job)? or maybe the midi data is been corrupted in some way?
I'm sorry but any of these possibilities are totally out of logic for me and would imply that the driver is not very efficient, but it's a disaster.

Quote
Windows only

http://www.oblique-audio.com/free/rtlutility
How sad, because we need objetive numbers. Subjectivity is by definition very inexact.

Quote
FWIW - for real world Audio latency measurement -  I only trust this


And explain how here:
https://www.vguitarforums.com/smf/index.php?topic=8929.msg133954#msg133954

I hope we can do something to obtain some light than purchase an very expensive device.
In fact in the link you provided there is a software, also for MAC, which seems capable of meassuring the RTL. Maybe someone can try.


Elantric

#7
QuoteSo you are saying that the guitar2midi conversion is done in some way in the computer? So if we disconnect the computer then there is not conversion possible?

No - i'm saying there are several components which impact Guitar to MIDI  conversion - specific to the  Boss GP-10

QuoteSo if we disconnect the computer then there is not conversion possible?

Correct - since the Boss GP-10 lacks any 5pin MIDi I/O - there MUST be a computer with USB Host function that supports the Boss GP-10 MIDI over USB ($35 Raspberry Pi is the cheapest)  The MIDI over USB relies on the quality and latency of the USB Driver, and Operating System's real time implementation  - all are important elements that contribute to the Boss GP-10's Guitar to MIDI performance.

QuoteIn fact in the link you provided there is a software, also for MAC, which seems capable of measuring the RTL. Maybe someone can try.

Not likely since neither GP-10 or GR-55 respond to MIDI note on/ off messages ( required to use the midi-latency-analyser
https://expressiveelectronicsformusicians.com/our-products/midi-latency-analyser


alexmcginness

#8
Im a hardcore PC guy. Back in the day I use to relish the Mac vs PC thing as on more than one occasion I clearly destroyed the Mac megahertz myth myth in head to head speed tests. Things have changed with the newer macs and the Intel processors. I kept noticing however, on videos that the Mac guys were using the VG-99 as their only guitar to Midi interface and weren't having the same glitching issues as I was having with a windows based system, so I bought a used Mac book Pro just to see. After all I could always dump it if there was little or no difference. Guess what? There 'is' a difference and a very noticeable difference at that.
   I understand that for many PC guys that it doesnt make sense, but the proof is in the puddin as they say.
   Don't believe it or you cant see why? Go find a Mac, load up the drivers and take it for a test drive. A/B it with your PC doing the GT 2 Midi thing and see for yourself. Until you do you will find yourself pontificating and rationalizing why it cant be so for whatever seemingly logical reasons you can come up with.
  Again, this isnt a latency issue but a conversion thing. Theres less glitching with the Mac Roland drivers than there is using a PC with the Roland drivers over USB.
VG-88V2, GR-50, GR-55, 4 X VG-99s,2 X FC-300,  2 X GP-10 AXON AX 100 MKII, FISHMAN TRIPLE PLAY,MIDX-10, MIDX-20, AVID 11 RACK, BEHRINGER FCB 1010, LIVID GUITAR WING, ROLAND US-20, 3 X GUYATONE TO-2. MARSHALL BLUESBREAKER, SERBIAN ELIMINATOR AMP. GR-33.

rolandvg99

Peripherals "buy" audio bandwidth "factory direct" in OSX /MacOS.  Windows relies on "distributors" to handle the audio business. Which again makes Windows audio more "expensive" and error prone.
To V or not to V: That is the question.

My little Soundcloud corner

shawnb

I must say I'm with jassy on this one.

I fully understand the crap architecture on the PC.  I understand Mac, just by virtue of a superior architecture, will likely have a few ms less latency than a PC.  I also expect better consistency - my measurements of PC latency vary a few ms, same note, multiple readings. 

But 'glitches' in the Roland world are clearly due to the MIDI traffic.   I'm talking the squeals, etc.  These are mostly due to pitchwheel MIDI events with large values, sometimes NOTE events.  This is output from the GP-10. 

I won't believe it until I see it!  Video would help. 

Or, better yet, I have a US-20, a Mac & a PC, so if someone sends me TWO GP-10s I'll do one!    ;D

I strongly doubt there is different MIDI traffic.   
Address the process rather than the outcome.  Then, the outcome becomes more likely.   - Fripp

Elantric

#11
QuoteOr, better yet, I have a US-20, a Mac & a PC, so if someone sends me TWO GP-10s I'll do one! 

Its equally possible to verify Guitar to MIDI (over USB) using appropriate Roland USB drivers with GR-55 and VG-99 - both work better on a Mac than a Windows PC

But the GP-10 will have lowest latency Guitar to MIDI from a Roland / Boss device in 2016

Vernon Reid's GP-10 Review >
https://www.vguitarforums.com/smf/index.php?topic=11506.msg86127#msg86127

I just did a GP-10 pitch to midi test via Ableton Live and THIS IS THE BEST OUT OF THE BOX EXTERNAL TRACKING ROLAND HAS EVER ACHIEVED. NO TWEAKING.


shawnb

#12
Better performance, in terms of consistency & a few ms, I'll buy.

You're saying no squeals & doubled notes, too?

That's the part of this thread I'm having a hard time with - no "glitches". 

I had done a 5-pin vs USB comparison of GR-55 traffic (& VG-99 also) a while back on a PC.  The traffic was identical across 5-pin & USB, as expected, only a ms or 2 difference here & there.
https://www.vguitarforums.com/smf/index.php?topic=870.msg44868#msg44868
Address the process rather than the outcome.  Then, the outcome becomes more likely.   - Fripp

alexmcginness

VG-88V2, GR-50, GR-55, 4 X VG-99s,2 X FC-300,  2 X GP-10 AXON AX 100 MKII, FISHMAN TRIPLE PLAY,MIDX-10, MIDX-20, AVID 11 RACK, BEHRINGER FCB 1010, LIVID GUITAR WING, ROLAND US-20, 3 X GUYATONE TO-2. MARSHALL BLUESBREAKER, SERBIAN ELIMINATOR AMP. GR-33.

jassy

Quote from: Elantric on August 16, 2016, 10:42:15 AM
No - i'm saying there are several components which impact Guitar to MIDI  conversion - specific to the  Boss GP-10
Exactly, "specific to the Boss GP-10", its the unit which does all the G2M work and then sends "midi" data to the computer or USB host with all the mistakes and successes that the technology included in the GP-10 is capable of performing.

Quote
Correct - since the Boss GP-10 lacks any 5pin MIDi I/O - there MUST be a computer with USB Host function that supports the Boss GP-10 MIDI over USB ($35 Raspberry Pi is the cheapest)  The MIDI over USB relies on the quality and latency of the USB Driver, and Operating System's real time implementation  - all are important elements that contribute to the Boss GP-10's Guitar to MIDI performance.

Although the GP10 does not have a midi port, does not mean at all that the full G2M conversion is not done entirely inside the unit.
The fact is that the quality of the G2M conversion in my GP10 with my 2 pcs is very good (not so perfect in the lower notes, but thats a fact of the Roland conversion metod and maybe some tweaks to do in the gp10 menu). So if you are obtaining that bad G2M behaviour with your windows pc, well, maybe is the configuration of your particular pc, because here is a very different history.


shawnb

Quote from:  sec6
I just did a test with a GC-1 and the GP-10 as audio interface hooked up to PC set to 48 samples (lowest sample size possible) in Ableton and received latency between 9 and 10 ms on high e at 7th fret.  So we just need some one to measure the distance between audio and midi note on a Mac.  Just arm the audio and midi tracks and hit record playing staccato notes at high e 7th fret. Zoom in to one ms resolution and measure the distance.

I'm fairly certain they'll beat that by a few ms. 

The question isn't latency, though. 

There are a few here claiming there are fewer "glitches" as well - and that term usually refers to ghost notes, double notes, squeals, etc., on the MIDI output of the GP-10.  That's the crux of this discussion. 

I'm having a hard time seeing how that is possible.  The driver doesn't - and shouldn't - alter the MIDI traffic, his job is to pass it - untouched - to the app. 
Address the process rather than the outcome.  Then, the outcome becomes more likely.   - Fripp

jassy

#16
Quote from: alexmcginness on August 16, 2016, 10:45:27 AM
Im a hardcore PC guy. Back in the day I use to relish the Mac vs PC thing as on more than one occasion I clearly destroyed the Mac megahertz myth myth in head to head speed tests. Things have changed with the newer macs and the Intel processors. I kept noticing however, on videos that the Mac guys were using the VG-99 as their only guitar to Midi interface and weren't having the same glitching issues as I was having with a windows based system, so I bought a used Mac book Pro just to see. After all I could always dump it if there was little or no difference. Guess what? There 'is' a difference and a very noticeable difference at that.
   I understand that for many PC guys that it doesnt make sense, but the proof is in the puddin as they say.
   Don't believe it or you cant see why? Go find a Mac, load up the drivers and take it for a test drive. A/B it with your PC doing the GT 2 Midi thing and see for yourself. Until you do you will find yourself pontificating and rationalizing why it cant be so for whatever seemingly logical reasons you can come up with.
  Again, this isnt a latency issue but a conversion thing. Theres less glitching with the Mac Roland drivers than there is using a PC with the Roland drivers over USB.
Again I think the conversion is done in the Roland unit, but anyway here we are talking about the GP-10 and the actual drivers Roland is delivering (2016), not previous units and old drivers. That is I thing a big difference.
The GP-10 is performing very well here so I not going to spend 1000 € in a second hand Macbook to try that.
If this was not working as well, I simply would use my Babyface.

jassy

Quote from: shawnb on August 16, 2016, 04:46:26 PM
I'm fairly certain they'll beat that by a few ms. 
I will love to see some real numbers about it, im not sure even of it.

QuoteThe question isn't latency, though. 

There are a few here claiming there are fewer "glitches" as well - and that term usually refers to ghost notes, double notes, squeals, etc., on the MIDI output of the GP-10.  That's the crux of this discussion. 

I'm having a hard time seeing how that is possible.  The driver doesn't - and shouldn't - alter the MIDI traffic, his job is to pass it - untouched - to the app.
Im totally agreed with you, how in the hell the driver can add double notes, squeals, glitches? its an "intelligent deformation of the data stream"? because if its not an intelligent process, that degradation or corruption of data would result simply in an unusable data and not "ghost, double notes, etc". that,s simply impossible by logic.
So well, maybe we are talking about latency anyway... maybe? in that case some numbers would be very welcome.

alexmcginness

VG-88V2, GR-50, GR-55, 4 X VG-99s,2 X FC-300,  2 X GP-10 AXON AX 100 MKII, FISHMAN TRIPLE PLAY,MIDX-10, MIDX-20, AVID 11 RACK, BEHRINGER FCB 1010, LIVID GUITAR WING, ROLAND US-20, 3 X GUYATONE TO-2. MARSHALL BLUESBREAKER, SERBIAN ELIMINATOR AMP. GR-33.

jassy

Quote from:  sec6
I just did a test with a GC-1 and the GP-10 as audio interface hooked up to PC set to 48 samples (lowest sample size possible) in Ableton and received latency between 9 and 10 ms on high e at 7th fret.  So we just need some one to measure the distance between audio and midi note on a Mac.  Just arm the audio and midi tracks and hit record playing staccato notes at high e 7th fret. Zoom in to one ms resolution and measure the distance.
Sorry not very clear to me the measurement you have done.
You mean that you recorded the guitar signal and the midi signal generated by the GP10 and the midi note was 9-10 ms after the audio?

gumtown

What you have done there is only compare the difference between audio and midi latency,
not measure the actual latency difference between physical guitar string note and the end point.
You need to add the audio latency to the midi result.
Free "GR-55 FloorBoard" editor software from https://sourceforge.net/projects/grfloorboard/

jassy

Im still testing my GP10 as sound card (while as guitar processor) in my laptop Asus i7 3810q and getting quite acceptable and consistent results in terms of latency realiability and quality of sound. I m the two last days using the lowest setting of the driver (something completely impossible with the VG99 or GR55 in the same laptop) and so far I have not had any crash or problem yet. The RTL is 6.3 ms and overload is similar to that produced my Babyface 48 samples (4.2ms RTL), so the result is quite impresive and promising. I have been using this latency with the different sounds that I use (midi2guitar, zebra, Omnisphere, Kontakt, etc and many effects) and so far everything works perfect, without clicks or pops.
So I can say that the quality of this driver in Windows is at a high level and is very usable, maybe its time to regain confidence in the quality of Roland audio drivers in windows, at least with the GP10 (with the SY300 my feeling was also very positive).
I still think that in this thread would be very enlightening if someone can bring real measurements of latency in OSX, so we can have a base of comparison.