Jump to content

Realtime Guitar to MIDI Converter Software


ReaktorFreak

Recommended Posts

  • Members

Mr. Anderton, i was wondering why you never make a Realtime Guitar to MIDI app/plugin ? I'm a guitarist who want to play synth but from what i've experienced i could never get the real guitar feel when playing MIDI guitar. So i think with today tech it is possible to write app/plugin that go far beyond MIDI guitar, that doesn't sacrifice our guitar playing feel, that makes synth really sounded like a string being plucked, bend, and slided. It will help a lot of guitarist to get into electronic zone that was always dominated by DJ's and keyboard players all this time. Thanks, and i really love your Quadrafuzz, its sound is so original that i could do crazy stuff with it.

Link to comment
Share on other sites

  • Replies 63
  • Created
  • Last Reply
  • Members

The Axon is the best of the guitar/MIDI systems I've seen. As far as something without a synth pickup, I imagine that such things can only work with monophonic material. Many DAWs have pitch to MIDI options. I've never used them so I can't speak to how well they work.

Link to comment
Share on other sites

  • Members

There are basic limitations (due to things like physics) in any guitar-midi system that uses string vibrations to determine pitch, that results in a more-or-less noticeable delay. Personally, it doesn't bother me, but I am not a fast guitarist. I can see the day when we can use an 'active fretboard' to determine pitch, and will use the string vibrations as a sort of 'aftertouch'.

 

Personally, I would like to see something that would take the separate string signals into the computer, where one could do as you like with them....

Link to comment
Share on other sites

  • Members

Well, i did try build one myself in the end, but sadly it works only inside Reaktor, i couldn't get it outside because from what i found out, MIDI's been not too kind with guitar. But at least Reaktor has good synth instruments, and the instrument i build really works the way i want it, sort of monophonic synth built into electric guitar (a software version of the ARP Avatar that works maybe ?). Maybe MIDI isn't for guitarist ? If you'd like to check, it can be found here :

 

http://electro-music.com/forum/topic-20775.html

Link to comment
Share on other sites

  • Members

 

There are basic limitations (due to things like physics) in any guitar-midi system that uses string vibrations to determine pitch, that results in a more-or-less noticeable delay. Personally, it doesn't bother me, but I am not a fast guitarist. I can see the day when we can use an 'active fretboard' to determine pitch, and will use the string vibrations as a sort of 'aftertouch'.


Personally, I would like to see something that would take the separate string signals into the computer, where one could do as you like with them....

 

 

The Axon, reputedly, uses neural network processors to 'learn' the note from the attack of the note, which is why it's much faster than those that time the period (length of time for the fundamental note to complete a cycle) or attempt to do an FFT (which also takes longer to do lower frequencies) to determine pitch.

 

The Roland, which I presume uses the 'period' method, is VERY playable in general, but does have perceptible lag on the low E string. Consider that the A note (5th string, open) is 220 Hertz. The period of that frequency is 1/220th of a second, which is about 4.5 mS. This is the time needed for sound to move about 4 feet. It's not terrible, but is perceptible.

 

It means you sometimes have to play a bit ahead of the beat on the 5th & 6th strings, especially when doing drum parts on your guitar. Alternatively, since the Roland allows mixing the straight guitar signal (from the normal pickups) into the synth sound, I generally do that if I'm using the synth to do something like bass guitar parts so I don't have to do tricks with timing. This is especially effective with palm-muted strings, as the muted 'thunk' sound from the pickup acts as the attack for the bass note (which the synth will play even with the string muted).

 

Both types will output 6 channels of midi (one channel for each string) to feed a PC or an outboard synth. Neither system is perfect, but perfection is very expensive...

 

I did once have an idea to embed wires in the neck, one wire per fret, and make a bridge that would leave the 6 strings electrically separated from each other. Then, each time a string is fretted, the contact between the string and fret would be sensed by electronics and immediately assigned a midi note. Pitch analysis would still need to happen, though, so one could translate string bends into pitch bends, for example. Also, the amplitude of the note would need to be used to 'gate' the midi note, since many players will fret notes but not actually play them (like arpegiating a chord). It could be done with maybe 3 or 4 chips for the electronics, methinks. Maybe a few more. Let's see.. a small ASIC to encode the fret signals into high speed serial binary, a D/A to sense the string amplitudes, a PIC to generate the MIDI & squirt it to the synth, probably a voltage regulator to run the others... Hmmm...

Link to comment
Share on other sites

  • Members

 

I did once have an idea to embed wires in the neck, one wire per fret, and make a bridge that would leave the 6 strings electrically separated from each other. Then, each time a string is fretted, the contact between the string and fret would be sensed by electronics and immediately assigned a midi note.

 

 

 

Similar idea to the old Guitorgan (they may still make those... hmmm). They used a divided fret, wired to the organ circuits, to determine the note. Of course, it would play whether you strummed or not, and pitch bends didn't work with the split frets....

Link to comment
Share on other sites

  • Members

 

Nice funky groove!!


Need to add some soulacious drums & bass now, eh?

 

 

Thanks. BTW, i've tried Lateral Solution Guitar Synth 2.0 and it didn't work in Windows XP. Is ver 2.2 works on XP ? I guess that programs really cool because it's polyphonic. Too bad it's discontinued...

Link to comment
Share on other sites

  • Members

Philbo:

I found your comments on the ROland very interesting. In particular: It means you sometimes have to play a bit ahead of the beat on the 5th & 6th strings, especially when doing drum parts on your guitar

 

I am just starting out on the Roland; my goal was to use it to work on my rhythm, staying on the beat stuff like that. Where can I find more information about how to use midi guitar and its limitations?

Thanks

 

John

Link to comment
Share on other sites

  • Members

The best way (and maybe only way) is to plug it in and play some parts, especially playing against inflexible beat sources: CDs, radio, recording in an overdub situation (rather than playing with other live musicians, who will hear what you do and adjust their playing to sound good with the timing you play with).

 

The interesting thing is how much you have to adjust your playing technique for other various instruments - - piano, organ, sax, cello, flute all require you to change your playing technique in different ways to get realistic sounding parts.

 

For example, on piano nobody can play a 6 note chord spanning 2 octaves with the left hand, unless they have 10 inch fingers and an extra thumb... ;)

Link to comment
Share on other sites

  • Members

I've got the Roland hex pickup and floor synth unit. It's not in front of me at the moment and I'm spacing out on the model names, but it's recent vintage stuff.

 

I found that when used as a system, as in the pick up driving the synth and then using those built in sounds, outcomes are pretty good. The problem? The built in sounds suck so badly. It's just the standard Roland ROMPler set with little in the way of tweakability. I have used the system extensively, indeed primarily, for MIDI work. Latency is noticeable and glitches abound. You've got to play super clean and even then, I find myself constantly having to edit every performance. I'm a fan of unexpected and unusual outcomes so in a sense none of this bothers me, but for a guitarist looking to do things in real time, the Roland stuff is hit or miss.

Link to comment
Share on other sites

  • Members

You are right - you do have to play cleanly to get clean results. You also have to have your guitar neck relief, action, intonation, hex pickup height set extremely well.

 

And, if you're talking about the GR-20 synth, I do agree that many of the built-in patches are simply ridiculous - - for example, there is a only one steel drum patch I can find, and it is buried in the percussion section, and the low E string triggers a drone rhythm that makes that string completely unusable for that patch. Out of the 485 patches in the pedal, I've found about 25 or so that are really quite good, and the rest are rather junky - - not much different than many other synths.

 

I'd really like to have it set up with at least 100 distinct, separate, faithfully recorded instruments, (I KNOW there's at least that many out there!!) - a dobro would be nice; so would a harp, and there's really no reason they couldn't add 20 or 30 REAL stringed instruments to the guitar bank (dulcimer, balilaka, & similar) instead of the miscelaneous crap that's there.

 

I'd also like to have it possible to assign a different (or same) patch to each string. Having to be stuck with these canned patches is pretty awful. I guess I could just drag along a rack mount midi sound generator and program that... or buy a GR-33, which is a much better synth.

 

Another thing that torques my trigger is that there is no good way to change banks without stopping playing, bending down to turn a knob, then standing up again to play. Just stupid design! The only way to make it usable is to copy the patches you want into the User bank, and then scroll through them to get to what you want. I hope the patch change switch on my guitar is robust enough to take 200000 operations or so before it wears out!

 

Even worse, if you try to use your foot to turn the knob, you better be REAL careful not to accidentally step on the 'set string sensitivity' button, because once you spin the knob CCW, it'll set the sensitivity of that string to OFF. And you have to twang the sh!t out of that string to even get it to register again so you can turn it back up. If you're lucky! If not, there's no way I can find to make that string trigger the synth again. (This happened to me on gigs a couple of times till I finally figured out what was happening) - - it has identical symptoms to having a dead pickup for one string! I wonder how many RMAs Roland gets for this problem? Why on earth would anyone want to set string sensitivity to ZERO?

 

I can't figure out what Roland could have been thinking when they dumped the GR-33 and went with the 'dumbed down' GR-20 - - maybe they are assuming guitar players are not synth-literate? Maybe I'm too picky - - I do design work for a living, and recognize and appreciate decent ergonomics and elegant design... IMHO, Roland really should consider doing some field tests of their stuff by real guitar players (non-Roland employees) before they commit themselves to their design mistakes.

 

Link to comment
Share on other sites

  • Members

So I justed checked - it's the GR-20 I have. My guess is that they felt fine putting all those {censored} sounds in the unit and giving so little room to tweak because they figured most guitarist would be overwhelmed with anything else.

 

Truth be told, I really just use the set up for recording. Laying down midi tracks to generate other higher quality sound sources, plus drum programming, are the main things I do with it. I'll never take it on stage.

 

On the matter of drum programming, I have to say using the guitar as a controller was the first time I could effectively program tom fills and rolls on the snare. I pick so much faster than I can pound on a controller pad, so I was finally able to program really realistic stuff. Again, it's glitchy, but quantize it and it works quite well. I never program drums any other way now.

Link to comment
Share on other sites

  • Members

 

From what everyone's saying i guess even hardware converters are still have lots of problem. No wonder software ones never worked well.

 

 

guitar has some inherent problems as a synth controller -- both

 

technically :

they are polyphonic - but there is also a decent amt of resonant "crosstalk" between the strings

They have a percussive profile (so they have a noisey, pitch indeterminate initial attack, the decay is such that keeping a readable signal in the later stages can be difficult)

etc

 

AND

 

playing style :

A lot of modern guitar playing uses more inicidental noises -- damped/muted strings/sounds , 'ghost' notes, fingering noise, taps, scrapes, weird harmonic..."dirty" (not perjoratively, just stylistically looser in terms of articulation, use of those incidental noises to artistic effect) playing in general

 

SO there are the twin problems of "tehnical tracking" -- can the machine extract pitch, trigger, etc information in an accurate (glithless) and timely (negligible latency) manner

 

 

Then there is the problem you mentioned in the topic post..."guitar feel" -- all the little techniques a guitarist uses, often

1) don't have a directly analogous parameter to be assigned to in a particular patch

[conversely, as was iscussed above, when we try to imitate another instrument, THEIR techniques and approach can differ from our guitaristic approach]

 

2) the guitar isn't so much a an instrument with discrete parameters like a synthesizer...so the overall "parsing" of the guitar sound into

pitch, trigger, dynamics, some other forms of "performance expression" (relative spectral profile - ie "tone/timbre", the old sentient 6 had upstroke/downstroke detection) can fall flat

 

 

Back in the 24-pin days -- I often found that guitarists would say "this thing doesn't track well"

While there was certainly the technical "pitch to glitch" type problems, I also started to realize what they meant is that, even with very good tracking, much of the "guitar feel" was stripped -- both as a function of parsing the guitar's signal into discrete parameters (leaving some important info behind) AND as a function of applying that data to un-guitarlike" sounds

 

 

We've certainly seem to have come a long way -- but there are still some sort of inherent problems, in our approach and the physics of the instrument itself

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...