GR-55 - Tracking Solved - No bad notes

Started by Moja, April 25, 2017, 09:40:19 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

bryanz

I remind my fellow GR-55ers of a trick I tried with a piano sound, that I think helped a little (just a little): Putting the same or similar voices for PCM1 and PCM2, and giving one a very short sustain so that it dies after the initial attack, and the other one a slow attack--so that the attack and sustain are split between the 2 voices--the theory (and I think actuality) is that just the attack portion of a wrong note is not as noticeable, and by the time the sustain voice kicks in the Gr-55 has figured out what note you are actually trying to play.

Shingles

Quote from: bryanz on April 07, 2019, 06:47:15 PM
I remind my fellow GR-55ers of a trick I tried with a piano sound, that I think helped a little (just a little): Putting the same or similar voices for PCM1 and PCM2, and giving one a very short sustain so that it dies after the initial attack, and the other one a slow attack--so that the attack and sustain are split between the 2 voices--the theory (and I think actuality) is that just the attack portion of a wrong note is not as noticeable, and by the time the sustain voice kicks in the Gr-55 has figured out what note you are actually trying to play.

That's similar to Roland's trick with internal sounds in their guitar synths that goes right back to the GR50. It depends on their voice architecture that was first seen in the D50 synth, with one wave for the attack of the note and another for the sustain. The attack portion is triggered before the string pitch is stable but that's ok, as the attack wave does not include critical pitch information. There is then time for stable pitch recognition to trigger the sustain portion. This is how the GR30 and GR3, for example, manage tracking with their internal sounds that is faster than MIDI to external sound modules.
Nik
--------------------------------
Tonelab, VG99, Axon AX100, EDP, Repeater
Godin, PRS, Crafter and Roland guitars
Center Point Stereo Spacestation V3

Aengelantir

#127
Experiment: Input parameters for max OSC tracking. Ideas?

Has anyone done an experiment like this?

Recently got my hands on a GP-10 and GK-3, which I installed on a 335-model guitar. Even after much reading (thanks for the suggestions on this forum) and trial and error on GK-3 settings, midi vol&feel settings and trying to clean up my playing, I can't get good results from the OSC synth. So I have decided to set up an experiment with the use of my DAW Ableton. Because shouldn't I be able to get results like this? (watch ).

The search is for the parameters that I can influence and have the most impact on a good result in translating to midi/OSC. I like to understand before I start luthering in one of my other guitar (:->

The Ableton set up takes in everything seperately: all unprocessed hex channels and PU's, all midi per string (mono) and all midi signal combined (poly). All audio is max - 3dB. It works very well to re-amp the PU's and re-guitar/synth the hex on the GP-10 so I can tweak away. In the DAW I've als redirected every hex channel and the PU's each to a seperate jamorigin VST channel. All these midi channels (14 all together) have a seperate midi synth vst and deliver sound.
But GP-10 OSC and the midi synths in the DAW (!) glitch and play wrong notes.
I suppose that the main reasons are transients, unwanted harmonics and crossfeeding from other channels.

So I recorded
- a muted stroke over all strings at once to get string transients
and per string
- a muted attack for a transient
- a open string with other strings unmuted
- a note on the 9 fret (eg a C on the 6 string) with other strings muted.
Since input volume and frequencies are the essential information for transalting to midi I put a bandpassfilter and noise gate on each hex channel. So far I only experimented with the low E (6) string and here are some observation. (which I will reassess)
Volume:
-Transients deliver unpredictable notes (off course)
-But it is possible to set a min volume floor (around -20db relative to the input) for which the transient disappears while the notes do not, even if the transient is as loud as the notes.
Frequencies:
-The right note is found even if the BPF cuts of above the fundamental or octave harmonic (2x Hz) of the note played. Same at the second octave (4xHz). This is without any high cut.
-The right note is found if the BPF is set arround the fundamental, the octave and the double octave.
-Transients cannot be disgarded by using the BPF.

So there is some intelligence in the translation to midi. E.g also higher frequencies are used (at last dor low strings). And there something to do away with the transients.

Before I dive into this for hours, does anyone have ideas about
- what assumption should be tested? What to test to decide upon where to improve: guitar, guitar setup, playing style, GK-3 parameters, GP-10 (midi) parameters, etc?.
- what to test and measure?
- improve the experiment set up e.g. is there something to learn from the GP-10 midi output or from the Jamorigin midi output.

If you came this far reading, thanks for that. Ideas or additional information are welcome.



Gibson L5 CES, Ibanez 2630 Artist + GK-3, Fender Am Strat, Ibanez ST-300, Aria Nylon, Fina steelstring, GT-10, GP-10, Ableton.

admin

#128
The GP-10 OSC SYNTH  is extremely poor tracking ( SY-1000 OSC Synth is better , but also not perfect
This is unique to OSC synth only.

- while the Guitar to MIDI is much better - AFTER  applying the Guitar to MIDI Settings recommended for VG-99, GR-55 documented earlier in this thread and in the DOWNLOADS area

https://www.vguitarforums.com/smf/index.php?action=downloads;sa=view;down=26

https://www.vguitarforums.com/smf/index.php?topic=4439.0#msg30065

Few docs exist specifically for GP-10.

billbax

QuoteHas anyone done an experiment like this?

Separate strings and Midi Guitar 2 can be accessed with a simple £80 GK breakout cable. Plug each string cable into your daw interface, and route selected analog strings to MG2.

As shown here. MG2 action starts at 2:15.

Standard 1/4 inch jack breakout cable. £80GBP.

Aengelantir

Hey billbax,

Saw that video (and your others too)! Very interesting. I see that you use a noise gate as well. In my setup the GP-10 bassically is that fanout and audio interface. I use also Jamorigin but just for comparison because I wouldn't like to take the laptop to the stage .

There's a lot automation and sequencing going on. My experiment is more aimed at playing chords on jazzy changes ( eg with Ainsoph's e-piano patch) and melodic soloing (like Michael Breceker on the EWI). Still tweaking and cleaning up my playing to get rid of the glitches.
Gibson L5 CES, Ibanez 2630 Artist + GK-3, Fender Am Strat, Ibanez ST-300, Aria Nylon, Fina steelstring, GT-10, GP-10, Ableton.

abbaby

It's also best to dampen strings (palm or other) you don't want sounding - being exact with the picking, change sensitivity, and changing the volume of some strings on certain instruments helps also.