GM-800 MIDI Tracking Results: Midi Guitar 2 vs. Fishman Tripleplay vs. GM-800

[Bluesbird]

Received my GM-800 about an hour ago. I adjusted the pickup sensitivities, turned off all scene parts and measured pitch to midi using the usb output on the GM-800. I used a GK Kit guitar connected to GKC-AD, and connected to the GM-800 with a 1-foot trs connector. The result of measuring all open strings, notes at the 5th fret and notes at the 12th fret, dividing by 18 and adding the 1.84 input latency yielded a result of 20.8 ms average. A second test had a similar result with an average of 19.78 ms. The pitch to midi latency average was close to the Wayne Joness GR-55 external midi result of 19.01. The Tripleplay has a latency half that of the GM-800. Please, don't shoot the messenger--you are welcome to do the tests yourself and compare to my results. Sorry to report the bad news.

You cannot view this attachment.

Second GM-800 Latency Test

You cannot view this attachment.

Test Methodology

I set up a midi track in Ableton using the GM-800 Daw Ctrl as midi input. I set up a track with just the audio going directly from guitar into Focusrite Saffire 14 interface (1.84 input latency).  I had the tracks side by side and analyzed the difference in position along the timeline in arrangement view between the onset of the guitar audio and the onset of a midi note. I made sure latency compensation was disabled in Ableton.

Comparison to Wayne Joness Midi Speed Results

https://www.joness.com/gr300/MIDI_SPEED.htm

Jam Origin Midi Guitar 2 latency was tested in mono mode. 

[Elantric]

We have access to great tools

I hate using computers at a gig, so internal sounds are what are required for live gigs
I find the GM-800 internal sounds trigger very fast
Faster than its Guitar to external 5 pin MIDI performance.

Im positive the Fishman Tripleplay  guitar to MIDI  is faster, lower latency vs GM-800 Guitar to MIDI.

IF GUITAR TO MIDI is your only need - look at other gear
( send your GM-800 back and enjoy alternatives )
But no Guitar to MIDI SYSTEM or Roland GR , or Boss  GM-800 -  even the wired fret Systems can replace the play experience I accomplish with VG-99  VGuitar System

https://www.roland.com/us/support/knowledge_base/201922959/

[germanicus]

Thanks for testing and posting this.

One further test would be to record the gm800's audio output and compare it to the other signals as well. Doing this we could then see definitively if their triggering of internal sounds is faster than the pitch to midi output.

I would guess it might be slightly faster, as in the sy1000 this was the case with OSC synth (as well as past units like the gr33). I measured the OSC synth in the sy1000 to be slightly faster than the midi output (but nowhere near as fast as the COSM modeled dynamic or gr300 HRM synths).

[arkieboy]

That isn't the greatest of news, but I'm with Elantric here - it's the performance of the internal sounds not the midi out that I'm interested in.  Is there any chance you could work out how to reliably add the audio-out latency to that comparison?

Also a stab at the sample variance of tracking for ten or twenty triggers on some of the lower strings would be good too if you're up for that - the thing that kills pitch-detection synths for me is inconsistent tracking, rather than slow tracking.

[Bluesbird]

Here is a test comparing direct audio of high e string to midi note and to the audio coming directly from gm-800 using patch 0013 Suitcase EP. The stereo phones out of the GM-800 was input into Focusrite Saffire Pro 14 line inputs 3 and 4. The latency between guitar audio and midi note for this single high e string test was 15 ms. And the difference between the midi note and the GM-800 audio was 8 ms. The GM-800 audio occured 8 ms. after the midi note. The difference between the guitar audio and the GM-800 audio was 23 ms. The on-board sounds do not trigger faster than the guitar to midi from the USB.

You cannot view this attachment.

[Elantric]

That isn't the greatest of news, but I'm with Elantric here - it's the performance of the internal sounds not the midi out that I'm interested in.  Is there any chance you could work out how to reliably add the audio-out latency to that comparison?

Also a stab at the sample variance of tracking for ten or twenty triggers on some of the lower strings would be good too if you're up for that - the thing that kills pitch-detection synths for me is inconsistent tracking, rather than slow tracking.

Realize the specific GM-800 Preset employed for the test WILL impact results - many GM-800 tones are not fast attack , instead they have built in slower ramp up

[germanicus]

With the FTP, it hits a performance "floor" somewhere between g3 and b3, of 6ms. No subsequent raising of pitch leads to faster results.
The roland never, even several octaves higher, gets that low. Maybe if you played above 12th fret on E4 string.

[BROCKSTAR]

At the end of the day it's all just numbers, however if you enjoy it and it works for you and your style then it's great! To me the gm-800 is one of the best things they have done so far!

[germanicus]

Realize the specific GM-800 Preset employed for the test WILL impact results - many GM-800 tones are not fast attack , instead they have built in slower ramp up

Yes best to use a sound with a percussive attack, like a piano.

[Bluesbird]

Here is a result with a fast attack drum using the rhythm part only. The GM-800 drum sound using the high e as a trigger occured 5 ms. after the midi note.

You cannot view this attachment.

[germanicus]

The internal sounds are slower than the PTM? Thats kind of surprising.

[Elantric]

Next measure the audio output latency response of the Third party Synth you are triggering via MIDI

[Bluesbird]

Next measure the audio output latency response of the Third party Synth you are triggering via MIDI

Here is an open e test comparing guitar audio to midi note, Omnisphere quick attack audio, and GM-800 quick attack audio. Omnisphere registered within 1 ms after midi note and GM-800 audio occurred 4 ms later.

You cannot view this attachment.

[Elantric]

In actual use, I cant hear or feel any latency until its above 16 milliseconds

I did play many gigs in the 1980s with the 40 millisecond GM-70 Guitar to MIDI


Regardless , Guitar to MIDI remains a high priorty pursuit for those who hate the sound of guitar - which explains its general malaise among shredders
https://www.joness.com/gr300/MIDI_SPEED.htm

[Nobulusprime]

Here is an open e test comparing guitar audio to midi note, Omnisphere quick attack audio, and GM-800 quick attack audio. Omnisphere registered within 1 ms after midi note and GM-800 audio occurred 4 ms later.

You cannot view this attachment.

Interesting results, I'd like to see the low E internal audio (fast attack) of the GM800 comparison to the regular audio as that's where the greatest variation of latency occurs. I really notice the difference on low pitched notes on my Roland gear  compared to the FTP

The main question I have is tracking stability and spurious notes compared to the FTP and other Roland/Boss products. I'd image it's light years ahead of the SY1000 OSC and midi out but hopefully inline with the FTP?

[Bluesbird]

I'd imagine it's light years ahead of the SY1000 OSC and midi out but hopefully inline with the FTP?

As an FTP user, the GM-800 feels sluggish. The FTP is still by far the best in terms of stable and fast tracking. It's no contest. I can't recommend the GM-800, but don't take my word for it—-you should check it out to see if you like it.

[arkieboy]

I did play many gigs in the 1980s with the 40 millisecond GM-70 Guitar to MIDI

Me too.  The only guitar synth I would play a piano part live with confidence.

The thing I notice when I watch vids of people playing the GM-800 is that there doesn't seem to be any hesitancy or uncertainty in any of the timing in the notes played.  The performance hangs together - it has feel.  Inconsistent tracking makes you sound like your playing is amateurish and uncontrolled.

Maybe that's just because I'm aware of it when I'm playing ...

----

Also I agree that using Wayne's method is good for comparison with other units because of the published data, but I don't think averaging all those numbers out gives you a meaningful picture of how well a guitar synth performs.  For one, the arithmetic mean expects a normal distribution of errors, but the delay variance would be expected to be more like a Poisson distribution https://en.wikipedia.org/wiki/Poisson_distribution because no matter what your variance in detection is, its never going to be faster than the capability of the unit, faster than sampling theorem dictates, or ultimately faster than the audio from the string itself.  But a really sloppy guitar synth could have a nominal variance that introduces the potential for time travel ;-)

You would expect high frets on high strings to give you practical minima for the pitch detection software.  Then you would want to know how that minimum changes as pitch falls, particularly if there is any noticeable breakpoint where physics takes over from the processing algorithm.  Then ultimately how the lowest regular note performs.  And of course I think variance is important here - 16ms is (according to google, and the speed of sound at 330m/s STP) the same as being 5.2 metres from your amplifier, so you ought to be able to cope with that.  But there is almost no variance in the speed of sound from second-to-second stood in front of your amp, which you rely on for your feel.

</mini-rant>

[arkieboy]

The FTP is still by far the best in terms of stable and fast tracking. It's no contest.

Which of course beggars the question 'why the flying f@*k doesn't Fishman make a TriplePlay unit with a 13-pin connector'?

[Nobulusprime]

I think confident play feel is the most important thing. I do find the low pitch string off putting and distracting.

Eventually I'll try the GM800 out if I can return it without charge. The fact I need to have the GKDA box makes it even less attractive.

If the play feel is as good or better than the FTP I'll probably want to keep it as I'm not a massive fan of using IPads or laptops for live either.

[Elantric]

Which of course beggars the question 'why the flying f@*k doesn't Fishman make a TriplePlay unit with a 13-pin connector'?

Because there is Zero MIDI on the GK13 connector

[GuitarBuilder]

Let's face it, FTP is the benchmark, just like Axon was before that.  Having said that, it's all about what latency you can tolerate...

[arkieboy]

Because there is Zero MIDI on the GK13 connector

Of course there isn't!  You put the guitar wart gubbins in the triple play controller, add a small amount of circuitry if necessary to cope with any differences in the pickups and charge through the nose (Ed.  Sorry, Brit phrase for 'price it by as much as the market would bear, and then a little more').

They'd fly off the shelves, I'm sure.

[IMH1234]

Of course there isn't!  You put the guitar wart gubbins in the triple play controller, add a small amount of circuitry if necessary to cope with any differences in the pickups and charge through the nose (Ed.  Sorry, Brit phrase for 'price it by as much as the market would bear, and then a little more').

They'd fly off the shelves, I'm sure.

Absolutely - FTP tracking could be so fast it plays notes before I pick them and it is still a none starter without being able to connect in the same system as my VG99. Fishman have a great guitar-midi conversion system to be sure but without proper hardware synth integration and Roland compatibility it is always going to be the Betamax of the guitar synth world.

[pasha811]

Thanks to Bluesbird for all kind testing even on demand!
Personally I have never used an FTP but as Elantric said my VG99 experience has been so far the best.
As I said into another post, it all about what you need to do and what is your approach. If you are a live musician I think that GM800 (not owning one) given testing and comments is the best thing Roland did to perform
Live with G2M. If as I am you need a good solo synth sound live VG99 SY1000 SY300 and GP10 can help (no OSC) no need for a fully fledged GM800.I am a bedroom guitarist with my Daw and that's why GM800 can be an investment in the wrong direction as I have Jam Origin and VG99 Midi out to JV1010 to DAW that works great for me... but now that I was inspired by Bluesbird I want to test my setup in the same way...

[BROCKSTAR]

I'm a "bedroom" guitarist too and gm-800 is good for me I think having units like this is just better and less complicated then having to fire up your DAW, plug-ins and other things. I have jam origin as well but I don't find it as thrilling as having an actual unit but we're all different