MIDI Guitars: Who needs them?
Despite the variety of MIDI guitar systems around, nobody has got it right yet. Dove Lockwood offers his views on where manufacturers are going wrong.
Even my best friends would agree that, for a guitarist, I'm not much of a keyboard player! So why then have I spent much of the last two years stabbing away with one finger of each hand at a keyboard? Why? To participate in the great 'MIDI Revolution' that has changed the lives of so many people working in music, of course.
We have had MIDI guitar controllers almost as long as we have had MIDI; some of them very sophisticated, some of them very expensive, yet all of them profoundly unsatisfactory, I would contend.
The infamous conversion delay problem of the commonly encountered pitch-to-MIDI systems can, I know, be quite effectively counteracted by monitoring the normal guitar sound along with the MIDI device, so that you always hear the attack of the guitar note to keep you in time. This does work pretty well for live performance, but it is a bit limiting for MIDI recording. Hearing the guitar all the time doesn't exactly help you play in sympathy with the more subtle and delicate synth voices, and it does preclude the possibility of using any form of real-time note transposition, apart from straight octaves.
The quantisation and negative delay features available on most sequencers would appear to provide at least a partial solution to timing problems, once you've got the information into your system, but this too proves unsatisfactory in practice. At faster tempos, triplet quantisation (and sometimes even straight sixteenths) will shift quite a few notes further 'out' rather than 'in', whilst negative delay can only be used where all notes need to be offset by the same amount. With pitch-to-MIDI conversion, the amount of delay is directly related to pitch; the lower the note, the longer the delay. So the notes are all different, and all require a different amount of offset for optimum timing - a nightmare editing task on any sequencer.
More recently, some manufacturers have sought to refine the pitch-to-MIDI concept somewhat, by offering guitars with built-in synths or systems with dedicated voice modules that do not rely on MIDI to communicate with the guitar. They are relying on the principle that it is not merely pitch analysis that takes time, but that a substantial part of the delay factor between triggering a note and hearing it is caused by the subsequent conversion into MIDI data. They are right. These systems are significantly faster, and much more playable, but the MIDI data that they eventually output - the bit that you can record - is no different to any other pitch-to-MIDI device.
In the search for a dedicated MIDI guitar controller that really works, I feel these designs will ultimately prove to be something of a blind alley. So what of the 'radical' designs - Stepp, SynthAxe et al? The SynthAxe, I am told, works rather well once you get used to it, and if it was five hundred quid I am sure they'd sell thousands of them, but as it costs thousands of pounds they are probably unlikely to sell five hundred. It contrives to do everything imaginable, and is surely something of a sledgehammer to crack the MIDI nut.
Stepp, alas, are no longer with us, although their original design (DG1) only worked well on its internal voices, and didn't perform well over MIDI. Their dedicated controller (DGX) came too late to save the day.
There is insufficient space within the confines of this article to touch on the lesser known hybrid designs, such as the Quantar, Passac, Zeta Systems etc, but Yamaha's G10 deserves a special mention. The G10 is a proper, dedicated controller, using revolutionary ultrasound technology for super-fast tracking, with special bend sensors, velocity sensors, MIDI whammy bar, on-board assignable controllers, and surely the most extensive MIDI implementation. So what's wrong with it?
Well, not that much actually. Played through one of the recommended sound modules (TX81Z, TX802) in true MIDI Mode 4 (Omni Off, Mono), you can sound quite convincing on it. But Mode 4 requires a separate MIDI channel for each string, and for the average MIDI recording setup that is far from ideal. How many of your voice modules implement Mode 4? How many MIDI channels have you got? Your sequencer software might have 96, but have you actually got the necessary hardware to support them?
Even the most sophisticated sequencers are simply not configured to handle a multi-channel transmitting device efficiently, and it all ends up being a bit of a pain in the neck. One look at your edit screen, on any MIDI guitar system, will have you diving for the Chromatic Mode to get rid of the torrent of spurious Pitch Bend messages, and whatever your program's version of 'Delete Short Notes' is, to remove the proliferation of unintentional notes generated simply by lifting a finger off the fretboard to move it elsewhere - a reasonable enough thing to do on a guitar!
Of course, you can always operate the device on a single MIDI channel... and then you'll find out why they are so keen for you to run in Mode 4. Apart from making possible individual string bend messages, unfortunately, Mode 4 is the only way to prevent a 'glissed' note (a new note started, on the same string, before the previous one has finished, ie. the sort generated by lifting-off) from restarting its envelope, and therefore re-triggering its attack. If you can lose the attack you can get away with spurious lift-off notes, but with the attack they are all too evident, even at low velocity.
There is no denying that the ideal MIDI controller is a keyboard; effectively a collection of switches to turn notes on and off, with a few extra sensors and controllers thrown in to add expression. It works. The MIDI language simply lacks the necessary vocabulary to describe every nuance of an instrument like the guitar - the complex pitch envelopes, the inner development of sounds over time, the sympathetic resonance. Very little of this can be properly detected, and what can be detected cannot always be transmitted; and what can be transmitted cannot always be properly interpreted by the receiving device. The G10, and others, are elegant answers to the wrong question! As programming tools they are blunt instruments.
I would contend that what is required of a MIDI guitar controller is that it should electronically emulate a keyboard, whilst physically allowing the guitarist to feel instantly at home on it, ie. it would need to bear some resemblance to a guitar in layout. Under the left hand should be 'switches', simply made to feel somewhat like strings, and laid out like a normal fretboard, while the right hand has the triggering 'strings' to sense velocity and contribute to Note On information. I say 'contribute to' because obviously they could work intelligently in conjunction with the left hand to interpret hammering-on and pulling-off. If the left hand sensors could be made pressure sensitive, then Channel Pressure messages (aftertouch) would be available to add expressiveness, and to read Release Velocity, thus making it possible for the operating software to take care of the lift-off problem.
The device should transmit on a single channel, with standard Pitch Bend and Mod Wheel controllers placed close to the right hand position, with the option of 'mapping' either Modulation or Bend to the left hand Channel Pressure messages for fingertip control if desired. Shouldn't be too difficult should it?
Well no, I don't think it should actually. In fact, the basis for such a device has been in existence for some time. There are already two cheap plastic 'MIDI guitars', each an offshoot of what can only be described as some sort of 'toy' guitar, with risible built-in sounds, and each, in its own way, so compromised in operation as to make it virtually useless for any serious work.
Yet if some of the characteristics of each could be combined, and with the addition of some rather more developed software, they would actually constitute the beginnings of the device I have just specified.
I am referring to the Casio DG20 and Suzuki XG1M; the former transmits at a fixed velocity and has horrible 'strings', while the latter has a ludicrous Note Off arrangement. Horrid though they are as 'instruments', each is actually MIDI usable in its own way. They get the notes out on time, they don't fill your sequencer's edit screen with spurious data, and with a little more development either could have been an outstanding answer to a fundamentally less demanding question.
We shouldn't be seeking a guitar that transmits MIDI, rather a 'keyboard' that plays like a guitar, to let guitar players exploit the power of MIDI for what it can do, rather than what it can't.