I'm a happy YouRock Guitar user. For what it is, it works nicely - but it has its limitations. Looking just at the fretboard side of things, there's no way of doing string bends or vibrato, it's not possible to tap two notes on the same string like a Ztar, you can't do pull-offs to an open string, and it's limited to 22 frets and 6 strings. And although the strings-as-rubber-lines thing works pretty well, I'm sure real strings would be nicer. So I got to thinking...

First of all, we keep the fretboard bit and strummy bit separate. That means we can keep the principle of making the fret spacing whatever we want, which feels nice on the YRG. We also don't need to worry about string vibrations from picking, which means we can set up a near-perfect action. But how to detect strings pressed? The easy way would be to check for electrical contact between a string and a fret, of course, but this basically turns the guitar into a keyboard matrix which then has the problem of "ghost keys". Then an idea came to me.

Suppose we make our left-hand strings out of Constantin resistance wire instead. To feel like a guitar, each string should be a different gauge of wire, which would naturally give a different resistivity for each string. Because each string has a different resistivity, we can measure the resistance between adjacent frets and tell exactly which strings are pressed down at that fret position, because every combination of strings pressed will give a unique resistance. Of course you'd need to measure resistance from either end to the frets too, in case only one fret is contacted; and that'd let the thing auto-calibrate the resistances when the strings aren't fretted. There are lots of different gauges of wire, so you could easily expand this to more than 6 strings.

Add a strain gauge at either end of each string, and now you can detect string bends and vibrato perfectly too. Or you could repurpose the bend/pressure for volume and get proper dynamics in two-handed tapping. With a bit of smarts to look at strain gauge transients when "unfretting", pull-offs will work properly too. Since string tension is no longer a function of pitch, you can easily use thin strings with low tension and make it nice and easy to play, or tighten them up if you prefer more feel.

It all seems pretty logical - so much so in fact that I'm suspicious, bcos I don't usually have original ideas! Can anyone see the flaw in this? And more importantly, can anyone point me at someone having done this already?

The devil is in the detail, of course. These resistances are pretty small, and measuring low resistances very accurately is a bit of a game - and in this case it's a game which needs to be played across 24 frets! However we do have the guitar neck as a convenient place to embed a load of fretboard-related gubbins, so it's not all bad. And although I haven't really researched it yet, I'm sure there are going to be a fair few ICs out there these days which would be very handy, such as mixed-signal ICs like Cypress's PSoC.

Any thoughts...?

The old Synthaxe did the separate strings thing:



--------------------
An Eagle for an Emperor, A Kestrel for a Knave.

Thanks Elf - not heard of that. (And since they lived and died in the 80s, and only sold a hundred or so at £10k a time, it's not a huge surprise I've not heard of it!) The YRG and Ztar do the same kind of thing, of course, except that they use buttons instead of string contacts.

The idea of diagonal cuts on the frets is a clever way of dealing with string-bends and stopping "ghost keys". I imagine that would have been difficult to assemble though.

The whole thing is well overdue a revival though. A few grand of complicated electronics in the mid-80s could likely be done for peanuts these days.

I think the main reason guitar synths are not that popular is that it just isn't as good as a keyboard. By that I mean if you are using it to say play strings or horns it's still 6 strings versus 10 fingers, plus how awkward certain combinations of notes are, esp. if doing two things at once. Also drum sounds are better played on keys. So from a band point of view what's the point when the keyboard player has it covered? In the studio, again it would be redundant as soon as there's a keyboard around. Thing is with sequencing it's easy even for a guitarist with limited keys skills to use the keyboard.

I know some people love exploring the sounds and have made it the main part of their thing, but I think they are the minority. Most guitarists like playing the guitar as a guitar, after all it's the one thing a midi controller can't do, and leaving the synth stuff to the keyboard players.

Just my opinion on why nobody is making them, I don't think there's much of a demand for them. I'm not against them or anything like that.

Yeah, for sure there's less demand. I guess I'm just approaching it from what *I* would like a guitar synth to do.

... which is what most inventors do, isn't it? once they're happy with it they find out if anyone else needs the same thing...

--------------------
hourglass | random thoughts | doubledotdash!? collective

As a guitar synth user for over 25 years, I think there are lots of reasons guitar synths have never taken off...

* they're expensive. and then you have to buy some synths...
* we're luddites when it comes to tech - lots of people just want a combo and FX pedals. even rack amps failed to shift large quantities. and you want them to learn about synths too?
* we're incredibly conservative - we're playing the same guitars as we were in the 50s
* a large amount of modern music already revolves around the guitar - and most synth stuff is 'just about button pushing'
* you end up taking two rigs with you when you use them

additionally for pitch to midi
* they shine a spotlight on your technique
* they're inconsistent even when accurate - while modern gear tracks relatively well, the spread of response is too wide

So what we really need is a cheap, easy to use and reliable system that can be installed reversibly on a favourite guitar and is reasonably forgiving for use with cheap, high quality soft synths. It could look like a pedal, or it should be easy to work with cheap controller pedals cause guitarists like to stomp on stuff

For key-based triggering, the YouRock guitar is pretty much ideal at £130. The GR55 is pretty good: the tracking is not all that you would want, but given the VG synth waveforms you're pretty much good-to-go. Shame it isn't half the price...

I'm waiting on the Fishman triple play unit. This fixes the inconsistent response problem while being faster still and plugs right into your laptop: being designed by the guy responsible for the Axon units - the best pitch to midi available - and with main stage costing just £21 it looks like it could be pretty much perfect. And if it's cheap enough then it might just take off

Steve

--------------------
arK music

Quote onesecondglance:

---

... which is what most inventors do, isn't it? once they're happy with it they find out if anyone else needs the same thing...

---

Yep - if I ever get enough free time, I may well start hacking things around and seeing what works.

Cost is certainly a consideration for guitar synths - it's why stuff like the Ztar is never going to be mainstream. The YRG is cheap enough that it could be mainstream though.

I don't find 6 notes a limit, since I play two-handed. Generally I find I'm using one hand on the lower three strings near the nut, and the other hand on the upper three strings around the 12th fret, so each hand covers three notes over an octave, and the whole chord from both hands spans three octaves. Which is pretty much what you'd be doing on a keyboard, right? Except that with a keyboard you need a keyboard, where I've got something the size of a mandolin that I can run around the stage with. (Please don't say that you could use a keytar. The 80s are over.

The YRG only lets you fret one note per string, of course. The Ztar solves that problem, so with a Ztar you can be fretting 12-note chords across 4 octaves. Try that on a keyboard!

Of course a guitar synth will only span 4 octaves, where an 88-key keyboard has 7 octaves. I think there's scope for guitar synths with more strings - especially with "virtual" strings like the YRG or Ztar. This could be configured like a regular guitar with extra strings top and bottom, or could be pitched like a Chapman stick. If we assume 10 strings (as per Chapman stick) is the optimum two-handed multi-string arrangement, that gets the guitar up to 5 1/2 octaves - more than a 61-key keyboard.

I guess also it's where you no longer have a guitar but a new stringed instrument (even if it's a midi controller). I mean why bother with frets? Just have 88 strings and you could use your thumbs as well. A midi harp

Seriously though do you really want to run around the stage playing string quartet and trumpet sounds, although maybe there would be a good guitar patch in there somewhere........

I'm only joking, I wouldn't mind one myself as long as it's not to expensive. It would have to play like a guitar though - if I had to learn a new instrument I might as well learn piano.

I tell you what, there's always a great reaction when I start the piano intro from "Don't stop believing". And no-one asks the other guitarists about their instruments, but I regularly get people coming up to me and saying "what the hell is that? that's amazing!" and variations on that theme. Which is nice.

I'll just get my coat...

Personnally I think the success depends on the sound engine rather that the 'interface' alone. I love Allan Holdsworths music for the reason that he made the unique sound work.

No serious keyboard player buys a synth for the oboe sounds... so why would a guitarist? Unless of course the instrument could be used to play the sound better than a keyboard..?

I'm a long time user of the Roland GK solution and really would say the VG8-88-99 series due to the DSP (COSM & HRM) engine is the most 'natural' way to go for stringed instruments.

For sample based sounds, a postive trigger point is required; not a stringed instruments strongest selling point..

A very low cost and yet powerful option is...
http://usa.yamaha.com/products/musical-instruments/entertainment/lighted_k ey_fret_instruments/ez_series/ez-ag/?mode=model

--------------------
C.LYDE
http://soundcloud.com/c-lyde

Quote The Elf:

---

The old Synthaxe did the separate strings thing

---

Shucks, you just reminded me I need a Synthaxe to prop against the Fairlight. When I get the Fairlight.

Someone needs to come up with a couple of absurdly overpriced and crazy looking electronic instruments for the current age that future retrophiles with more money than sense can hanker after. I guess the Reactable is a good start.

Clyde, I'm guessing you've not seen a YouRock. It's similar to that EZ-AG, except that the "fretboard" has continuous silicone-rubber "strings" instead of discrete buttons. And unlike the EZ-AG you can play the YouRock with two-hand tapping.

The idea of the YouRock (and the SynthAxe, and the EZ-AG and my idea) is that you *do* get a positive trigger point. It doesn't have to measure a waveform and calculate a pitch based on that - it's effectively just pressing a switch like a keyboard, so you have true zero latency.

And I'm not thinking about this making sounds itself. Keyboard players spend serious money on nice keyboards that only produce MIDI signals, and then they plug them into the synth module or laptop-with-software of their choice. So I don't see any difference with having a guitar-shaped MIDI controller, except of course for the shape.

You're right I'm not familiar with the 'YouRock' ... (awful name IMO..ha).. but my point is ...why?

Why create a guitar controller to better emulate a keyboard, the more interesting view would be to take the strengths of the natural guitar action, especially the string movement and create a living sound, unlike a dead midi message..?

Ever see the Yamaha G10..
http://www.synthony.com/vintage/g10g10c.html

--------------------
C.LYDE
http://soundcloud.com/c-lyde

Erm, that was what I was talking about in the original post, Clyde!

All current guitar synths are either just pushbutton (Ztar, YouRock) or rely on guessing the pitch from hearing the string (Roland, the G10 you just linked to, and so on). The pushbutton method gets you latency-free accurate tracking and potentially lets you play more than one note on the same "string", but at the cost of no pitch-bend, vibrato or pull-offs. The pitch-recognition method gets you guitar-style playing, but at the cost of latency, occasionally variable pitch recognition, and only one note per string. (I don't agree that there's such a thing as a "dead MIDI message", any more than a guitar string is "alive". Talk to Herbie Hancock about MIDI being "dead"; talk to beginners in an open-mic about guitars being "alive".)

So in the original post, I was suggesting a way of combining the best of both. Turns out the old SynthAxe had a similar idea (although implemented slightly differently), but was too expensive to succeed. No-one's come up with obstacles to the basic concept, although there's the perfectly valid argument that if I want to sell it, I can't just build it and they will come. But if I'm just doing it for my own use initially, no big deal there.

Having now recapped the thread...

On the subject of guitar synths and the pitfalls vs pleasures, have you seen this:-

http://evenharmonic.com/products/gtak5

I have an Axon AX100 and a Godin LGX-SA and Kontakt 5. I'm starting to think it's rude not to have bought this yet....

Grab,

Actually the G10 used radar to detect finger position, if memory serves.

The MIDI message is limited to 127 discrete values - as opposed to the dynamic range of a vibrating string.. hundreds, thousands? That's why the Roland VG method makes sense - no triggering, rather DSP is used..

Do you now follow my train of thought... why buttons/triggering of any sort?

--------------------
C.LYDE
http://soundcloud.com/c-lyde

c.lyde

Midi quantisation isn't a reason I would put forward as being important when comparing them. For me it is all about the accuracy of the pitch detection: you can always nashville string your guitar to improve detection speeds, but if low C sharp on the 3rd string sometimes takes 20ms to emerge and others 30ms, then its difficult to play. If every one takes 20ms then you can cope. Of course, the VG unit - by avoiding the limitations of information theory - has a more-or-less constant processing delay so after a while you ignore it.

<pedant>
On midi quantisation: MIDI velocity is indeed restricted to 127 values and - with a suitable modular style synth - you could devise scenarios where you could hear the difference between a note of velocity 126 and 127 all other things being equal. But in the context of pop, rock and even jazz music those scenarios wouldn't be realistic. If it was a problem we would have had MIDI2 by now. Pitch bend - which is much more important - is represented by over 16,000 levels so if you set your range to be 2 octaves - common in guitar synths - then the difference between successive pitches is 0.0014 of a semitone, which is surely good enough.

On expression, if Roland weren't so risk adverse, we might see pick envelope and harmonic content rendered as control signals - they are certainly measurable with modern processing speeds. Axon units can read the distance from the bridge to the plectrum and this is a very useful source of control signals.
</pedant>

But for me I could never replace pitch to midi because synth-like sounds created by waveshaping are audibly less pure than those created by circuitry or calculated mathematically inside a suitably powerful computer. Synthesisers are - in some respects - 'cartoon' instruments. They are abstract in the same way that a drawing is abstract when compared to a photograph. The stylisation and reduction of detail is part of their enduring charm and thus the wave shaping artefacts in the 'wave' and 'brass' processes reduce their impact.

Although I'd love to find out how good waveshaping could get in under 10ms if you threw some serious processing power at it ...

TBF I wouldn't be without either my Axon or my VG88....

Steve

--------------------
arK music

Edited by arkieboy (20/04/12 02:19 PM)

arkieboy

I think we're talking apples and banana's here. Dynamic velocity is unique feature of stringed instruments, its what makes the sound 'alive' ... by using our fingers on both hands we have infinite control on velocity of just the string, even playing a single note pitch.

The standard approach to A/D conversion for the pitch to MIDI units is sample and hold, quantize and trigger closest pitch. This is required for velocity and pitch (frequency) info.

The VG unit did/does not convert to MIDI and therefore is not concerned with correlating to set table of pitches and samples as one would find on a traditional synth engine.

"we might see pick envelope and harmonic content rendered as control signal"

To understand this request one has to have an idea of the amount of data being generated through DSP techniques, else the discussion is not very meaningful. Stated further -- how many points of the pick (volume) envelope must be plotted and made available - 300, 3000?

"But for me I could never replace pitch to midi because synth-like sounds created by waveshaping are audibly less pure than those created by circuitry or calculated mathematically inside a suitably powerful computer. Synthesisers are - in some respects - 'cartoon' instruments. They are abstract in the same way that a drawing is abstract when compared to a photograph. The stylisation and reduction of detail is part of their enduring charm and thus the wave shaping artefacts in the 'wave' and 'brass' processes reduce their impact. "

All very subjective and personal - I respect this, but this has little to do with my train of thought ...

- pitch to MIDI is useful for triggering sample or modeled instrument -- the key word being 'trigger'
- through performing FFT analysis and modification of the string in motion, triggering and set sound is not even a consideration - its as natural as me playing my acoustic and loving it (pull-offs, hammer-ons, flamenco, sweeps, slides included)

-------------BTW----------
The Yamaha G10 actually had a very unique approach to pitch, fret position and velocity detection -- check it out.
I still wish they were available.

Edited by C.LYDE (21/04/12 12:08 PM)

Quote C.LYDE:

---

Dynamic velocity is unique feature of stringed instruments ...

To understand this request one has to have an idea of the amount of data being generated through DSP techniques, else the discussion is not very meaningful. Stated further -- how many points of the pick (volume) envelope must be plotted and made available - 300, 3000?

---

I personally think if we had 127 levels of each we'd be in clover, but midi supports 14 bit controllers so you could have 14,000 if you wanted - if the target synth could interpret it. Pitch to midi gives you a processing window to calculate the Fourier Transforms and smooth the volume envelope sampling: do you want a continuous controller for each of the levels of each of the first three harmonics? Or should we calculate some distance function from an 'ideal' string decay? All of these calculations are simple physics, suitable for parallel processing and since Andras asserts that the Fishman Triple Play only uses a fraction of the DSP power available in the unit ... we could have some of this next year

(I worry that by the time you did this the result would sound like a guitar through a funny fuzzbox! But it would be nice to find out!)

Quote C.LYDE:

---

by using our fingers on both hands we have infinite control on velocity of just the string ...

---

I think we're using the same words to mean different things. I'd say 'expression' because for me 'velocity' is a concept that only applies to a keyboard. You can't account for pick angle, how much flesh off the thumb and position of the plectrum on the string in the single term 'velocity'. I also use 'infinite' differently (so I'm a scientist, I can't help it! ). 'Continuous' is a good word because the guitar - in absolute terms - has limitations, and our control over it is less than perfect. I think of a musical instrument as a complex, semi chaotic system - the tiny differences that occur even when you try to do the same thing ten times in a row excite the ear. THAT is the difference between real and synthetic instruments, and because they are familiar we are attuned to those tiny subtleties.

Quote C.LYDE:

---

The standard approach to A/D conversion for the pitch to MIDI units is sample and hold, quantize and trigger closest pitch. This is required for velocity and pitch (frequency) info ...

The VG unit did/does not convert to MIDI and therefore is not concerned with correlating to set table of pitches and samples as one would find on a traditional synth engine.

---

You are right in poly/single mode. In mono/separate mode the pitch of the string is modelled as the nearest semitone + a correction sent as a bender amount. It then sends more pitch bend messages to track the vibrato, and if you hammer on then it does not trigger a new note, it steps the pitch bend message to emulate legato.

This makes editing midi from a guitar synth really hard!! But it also makes mono/separate mode as expressive as a VG in terms of pitch - in no way is pitch quantised although there are some vids of me on youtube where I wish it was!

And you are absolutely right. It's not as immediate or as intimate as a guitar. But we have one (ok, several!) of those already!

Steve

--------------------
arK music