Hi all,

I'm wondering what experience all of you have with Stick melody side to MIDI conversion via the SE MIDI pickups (modified Roland GK3, GK-2A, etc). Curious about the tracking accuracy of different units and what settings you might have found work best.

I'll go first... For decades now I've been using the old Roland GI-10 as the interface to go from GK to MIDI (released circa 1995). I actually own three of them, and I think the oldest predates me even owning a Stick. It's been the most bullet proof piece of gear I've ever owned. Primitive but simple interface. I keep the settings pretty thinned out, with bend data off (everything is chromatic) and the touch sensitivity is pretty much middle of the road, or maybe a slight bump on the thinner strings.

When Roland introduced the GI-20 as its successor in about 2002, I think they boasted it had a faster processor which at least implied faster tracking, so I jumped on board eagerly. A good deal of experimenting later though I felt like it just had more false triggers/wrong notes than the GI-10 on the Stick. I think I probably tried it on guitar too, and in any case opted to stay with the GI-10.

Currently, I'm experimenting with the pitch to MIDI function on the Boss SY-1000. Interestingly it has more detailed settings than anything else I've used for the GK pickup. They ask you to input your instrument scale length, and the distance each of the six divided pickups are from their string's bridge piece. This creates a bit of confusion on the melody side though.. since in my case I have a 36" scale instrument. The longest scale length you can enter in "guitar mode" is I think about 30... and some of my bridge pieces are set a little farther from the pickup than you can input (maybe 1-2mm short). It has a separate "bass mode" I haven't tried yet, and might, but my impression from tech support is that bass mode cuts off higher notes at some point. The result so far though... is pretty bad. Lots of triggered notes that I can't even imagine a source for. But... there are many playing settings I've yet to experiment with.

Any thoughts? Thanks!

Tom Griesgraber
www.tomgriesgraber.com

My own findings in regards to GI-10 vs GI-20 were very similar to yours.

Truth be told I just never managed to get a setup that felt natural to me and I've since forgone MIDI entirely. My newest instrument doesn't even have the pickup on it. I really only used it to feed some pads and riffs into some of the loops I was doing. Since my entire setup has gone computer based, I just have a small Arturia Keystep keyboard that I USB into the computer for those tunes where I want some synth.

I pulled my GK pickups off years ago.(melody side only).using the Roland gr55 was decent for recording because you can taper.I was prone to using the electric piano patch.did experience latency on my 10 string.nice for a lot of strange sound effects , but I am a meat and potato guy. Before you know it I was taking off focus and was neglecting what I like most (Bass). That’s me I’m sure others disagree.

I took mine off and gave someone here a great deal. Never got on with it.

My assumption of your posting this topic is that you’d like your Stick(s) to control any given MIDI module on any given day. And the main question being has anyone in Stick-land had any better luck with anything that anyone has offered in the last 25 years?
Seems like every 5 years or so I write up another update on my take, so here ya go…
(I too have a SY-1000, so maybe it is time)

I’ll try to break the components down quickly on what I’ve experienced. I’ve only used the Stick Enterprise supplied GK pickups. I’ve never used RMC or ‘standard’ Roland guitar pickups on a Stick. I have no idea if the literal physical pickups that sit under the strings are that much different than what what Roland supplies for guitar or bass, i.e. whether there is any difference to their output signals. Of course there is a Roland GK housing on the Stick supplied units.

I have tried using noise filtering circuitry in-line. It made no difference. I think tapping and plucking have different needs and tapping doesn’t seem to need that.

I have 3 different Sticks with double GK’s. Your GI-10 settings are what I would agree are the best for tracking - no pitchbend, maximum or middle of the road sensitivity, and thinner strings. The GI-20, for all intents and purposes outputs the same as the GI-10. I have two of each GI-10 and GI-20. I have Y’d out to both GI-10 and GI-20 for a simultaneous comparison - within error, they are the same - one triggers as well or poorly as the other, in random fashion.

I do like to use PitchBend on the melody side. Not on the bass, which helps the tracking situ. Use the widest PB range settings (24 (finer resolution), not 2 (less fine)). Not all external synth modules can match that, but with Roli and MPE becoming more mainstream, that wider PB width is more commonly achievable.

I never bought into a dedicated Roland guitar synth modules (like a GR-55) until I purchased the SY-1000. I do have VG-99’s and a VB-99 (and older VG’s but they don’t output MIDI), and I have GP-10’s. All of these boxes have MIDI output capability but they literally output MIDI crap. I use them solely for the COSM sound conversion. I haven’t tried a GK specifically from guitar with these for MIDI output, but for the Stick they are useless in outputting decently tracked MIDI notes.

The situ with the SY-1000 being different from the GP-10’s is that there are some (just a few) internal MIDI synth ‘modules’ (the simpler waveform ones) that trigger or track somewhat ok, but that the other (‘dynamic synth’) module portions trigger terribly - which basically overall reflects on how the unit triggers externally. The GP-10’s internal dynamic synth modules track poorly, reflecting its own MIDI output as well.

So the Stick’s melody side generally tracks MIDI ok, as it is guitar range and guitar tuned mostly - which the Roland modules expect pitch-wise, per string. Not so for the Stick bass, as the Roland modules don’t expect 5th’s.
So my work-around for all this is to split the GK signals from the bass side. As you mention with 5’ths in the bass, a single GI-20 module does not offer full range coverage - i.e. if the GI-20 is set to ‘bass’, it will not track the top string all the way. If it isn’t set to ‘bass’ then it won’t track the lowest pitched string at all. There is no bass setting on the GI-10. So using two GI-20’s - one set to ‘bass’ and the other set to ‘guitar’, use the combined outputs, and filter out the strings that don’t work. There is also some poor tracking on the lower notes of some of the middle Stick bass strings, (unfortunately) but i think that this is the nature of using a unit that is calibrated to expect what is 4’ths separated string ranges, rather than 5’ths.

And in general as a reminder - each string needs its own channel. It especially is helpful for the open string filtering when a ‘tap off’ occurs. Even the most careful players will have a tap off trigger, no matter how much the string is dampened (via velcro, or the literal finger itself leaving the string) - under any given normal playing circumstance. So e.g. if the lowest melody is string is a ‘B’, then the receiving module is set for a bottom range of at least one note higher (C or C#), so that the B note doesn’t trigger during the tap-off.
If 6 strings all have the same MIDI channel output, then this isn’t possible to set up on the receiving MIDI module. And also pitch bend would affect all of the strings. All of this multi-timbral capability is very (and early) MPE.

Summarily, over the years I’ve recommended to SE to drop the MIDI label for GK’s. The GK’s are great for their literal audio output for COSM conversion, but generally -not- for MIDI output. This write-up doesn’t cover everything I’ve experienced, and believe me if Stick MIDI was simple, I’d use it all the time (which I do with synths, MPE, drums, percussion, etc.), but I don’t. Caveat - I think it’s more useful in a laboratory situation, where there’s a lot of post production clean—up in the DAW, or if you don’t really care what comes out in a haphazard manner, which can be ok. And - using a double GK equipped rig requires extra gear (cabling, modules, splitters, mixer channels, etc) to haul around.

Hopefully this isn't off-topic, but do any of you have experience with the SY-300?

here's Emmett introducing the SY-300

I haven't used any of the other SY-'s, sorry. Not off topic at all. I think anything Roland GK (or not) related is worth hashing over. I'd love to hear about positive experiences. I really dig all of the VG and GP gear, but not for MIDI triggering, which I believe is what Tom was asking about, in general. The dedicated GI-10 and GI-20 still are the best performers IMHO.

Some of the SY-1000 internal synth sounds track well. And sound good. It's like a next generation re-packaged VG-99/GP-10 combo, with more muscle. Functionally eh not so different.

btw in regards to all of the GK set-up stuff - i.e. scale length, level setting, pickup distance from bridge, etc - tweaking is pretty straight forward. It's kind of a 'take your best shot at it, play it for awhile, then set various things at the other end of the scale to see how bad things won't work, then dial it back in'.
I suggest downloading and reading some of the other product related manuals too - sometimes there are insights gained from one manual to the next, or better explained in later generations.

Also - regarding sensitivity settings for 'Per string level' - that I've found to work best - set peaks to around 95%, otherwise crosstalk can occur (with note triggering across adjacent strings) if the levels are too hot. The fatter strings need a lot less gain.

btw - I think this looks pretty amazing - and is probably/maybe covered elsewhere in the forum(?):
https://www.jamorigin.com

May I double check my understanding of that.
If I'm I getting it correctly these devices can be taught to identify which string the note is being played on just from analysis of the audio feed, thus allowing 'per-string' MIDI processing and removing the 'per-string' advantage that hardware MIDI pickups used to enjoy? Is so, then WOW!

It sounds like the tracking has been improving constantly from polyphonic interpretations. I would need to get ahold of the software to see how splits, layers and such are laid out, but I would would assume that the old school method of per string midi channeling would be more easily accomplished (for splits/layers) on a multi-timbral setup - eh TBD.

Tracking is still the key issue to me. Tracking is not a problem in general with polyphonic audio (GK type split pick-ups) to (some) internal 'synth' or other COSM conversions on the Roland modules, but is a problem with audio (split pick-ups) to pitch to MIDI, and so on (whatever the MIDI triggers).

One thing I have not mentioned is that MIDI bandwidth can be and still is an issue towards glitching, if you are trying to blend too many streams of CC (continuous controllers) and pitch bend messages in with notes from the source. e.g. the GI-10 or GI-20 generate notes and pitch bend as the main messages, but there is a limit of only one type of CC at a given time sent out simultaneously (like a mod wheel, breath, or filter cutoff message) - because when you start to add in other messages, it overwhelms the receiving object, and buffering glitches occur - which can be as disruptive as poor tracking. I use a Roland Integra-7 and regardless of whether its input is USB or MIDI DIN, this CC type of overload happens in both/either type of input.

Whether this newer tech is capable of 'dealing' with any kind of bandwidth issues for anything streaming in (beyond simple note, pitch bend and sustain pedaling) will tbd. Sounds like something to check out though.

I am also curious as to whether it would be useful to output separate strings from a split pickup on different audio channels, with waveforms that are more conducive to tracking - into this newer software. Like I do with the bass side GK, split the bass strings and more melody range into two different audio paths, to make the bass more 'readable' for tracking sake. I haven't totally explored this yet but it seems like there is a separate package for bass, so melding the two streams (of what works for a bass, and of what works for a guitar range, then filtering out the unwanted) in whatever manner, may require some experimentation to say the least. The filtering out may be the hardest part, because I doubt that the software knows where on the neck your finger is.