To the Patent Office I go?

August 10, 2008 3:32 PM

DoesorwhydoesntthisexistFilter: A very simple guitar synth.

So, it's quite simple.

You have a microcontroller.

And each of the six strings on a normal guitar are isolated electronically. So, like most acoustic guitars, the bridge does not conduct between them.

You have grounds on each of the frets. You run a very small charge through each string in quick succession - let's say, taking a tenth of a second to run through each string. Charging each string individually eliminates the guesswork of which string is pressed to each fret.

After that, it's straightforward to figure out where each string is fretted. This will not deal with bending strings, etc, but it could way exceed the tracking time of a Roland guitar synth, etc. To avoid a nasty line of wires grounding each fret, you could do it all within the neck, next to the truss rod, and connect to internal electronics.

Does this not exist? I've been killing myself over whether it does or does not, or could not, etc. If not, it seems incredibly straightforward to implement it with Arduino with a multiplexer. I'm really considering getting a semi-junker guitar to try this out: the possibilities seem quite cool: for instance, combining this with a footpedal to sustain a chord with a softsynth while playing other parts with the guitar.

Eh eh? Too simple? Huh?

Should I copyright patent this? Make it? Buy it?

posted by tmcw to technology (19 comments total)

Searching for "midi guitar" gets me Guitar synth and this: It seems like they're called hexaphonic pickups, or divided pickups. This Roland GK-3 with a GI-20 seems to provide a live midi interface with note details for each string...
posted by suedehead at 3:45 PM on August 10, 2008

(Granted, that isn't the specific technology you mention, but the end result is similar.)
posted by suedehead at 3:46 PM on August 10, 2008

Well, at first read, it sounds really over-engineered. The first thing that comes to mid is that changing strings would suddenly become a much more difficult endeavor, since you'd have to hook each one up to the charging & reading mechanisms.

The second thing that comes to mind is that the strings would ground whenever you touched them, either with your right or left hand. That would mean that picking or plucking the strings would read the same as fretting the string.

The third thing that comes to mind is that the decoding & interpreting mechanism would probably take a more complex hardware set than the current hardware available, something closer to a small computer than the rather simple hardware that Roland offers.

That said, my GR-20 is pretty piss-poor at tracking. You build it, and despite my nay-saying, I'll buy it.
posted by lekvar at 4:15 PM on August 10, 2008

Yeah - the big difference is that this isn't a pickup and it doesn't involve any pitch-detection, which seems to make it a whole lot cheaper and easier?
posted by tmcw at 4:15 PM on August 10, 2008

"Arguably the first production "guitar synthesizer," the Guitorgan featured electronic organ tone generating circuitry contained within its body. Wired frets, each split into six segments, were the basis of the pitch detection system. A MIDI output was optionally available in the late 1980s."
posted by bunnytricks at 4:17 PM on August 10, 2008

The Guitorgan is pretty cool stuff but the operating principle is a lot different - seems like the frets were pressure switches ala Guitar Hero.
posted by tmcw at 4:26 PM on August 10, 2008

Should I copyright patent this?

No, you just publicly disclosed your idea, so you can't patent it.

Try to build it. It probably won't work out the way you expect it to, but it still might do something interesting.
posted by b1tr0t at 4:43 PM on August 10, 2008

I think the answer to your question is that anyone who wants to use a synth to make guitar sounds is probably already fluent with a keyboard and would use that instead. If they're fluent with guitar frets, why bother with a synth?

(Also, 100-millisecond timing granularity is much too coarse to be satisfactory.)
posted by Class Goat at 5:11 PM on August 10, 2008

I've thought of this too, and the main problem as I see it is you only get information about which frets are pressed, not which ones are struck and when, and with what strength it's done. You'd still need to extract information from the acoustic signal to get that somewhat right. If you manage to do it, the idea might work reasonably, but I'm not sure it would be better than extracting it all from the signal. The biggest appeal of that common method is that you can use your regular guitar. As the instrument morphs into a computer game controller or a keytar, people will think it's less fun to play, in particular if the transcription isn't perfect anyway.
posted by springload at 5:26 PM on August 10, 2008

These days, with DSP technology being what it is, pitch tracking is so much easier.
posted by phrontist at 6:02 PM on August 10, 2008

@Class Goat: I'm looking at the other way around, a guitar making synthish noises (sidenote: I hate most synths... I'm not spending time making 80ish phaser sounds).

I'm somewhat interested in this idea because it would not be just using the guitar as a synth - it would kind of add an electronic element which isn't dependent on strumming, etc. I hate the idea of the guitar becoming as lifeless as the keytar, but love the idea of it seeing a progressive future instead of constantly restating its love for 1959's design.
posted by tmcw at 7:49 PM on August 10, 2008

Also: does anyone know if such a tiny pulsating current would make an appreciable difference on electric guitar pickups? (I'm aiming for the answer to be no?)
posted by tmcw at 7:51 PM on August 10, 2008

I'm looking at the other way around, a guitar making synthish noises (sidenote: I hate most synths... I'm not spending time making 80ish phaser sounds).

Then you have a solution in search of a problem. No one needs a new way to control synths; they already have one which is satisfactory.
posted by Class Goat at 7:53 PM on August 10, 2008 [1 favorite has favorites]

I disagree with Class Goat. There are lots of people interested in finding new ways to control synths, and people with strong guitar (or flute or violin...) skills who would like to be able to use those skills to control a synth in a fluent way.

Also, I agree with Class Goat: a tenth of a second is way too slow. You should scan each string a few hundred times per second.

If you do it at 3.3 or 5 volts, the grounding and shorting effect of your body and fingers won't be enough to swamp the signal in the frets. The catch is that these pulses probably WILL cause interference in your pickups. So try this out on an acoustic guitar instead, with a piezo or microphone pickup that won't be affected.
posted by moonmilk at 8:38 PM on August 10, 2008

Use a metal pick with a long wire running back to your microcontroller, and you may be able to tell exactly which strings you're picking/strumming! You'll have to scan your strings quite fast for that, since the pick is in contact with each string for a very short time.
posted by moonmilk at 8:43 PM on August 10, 2008

Nerding out a bit here: You don't have to run a separate wire to each individual fret. What you do is to connect adjacent frets with resistors. Then you use an A/D to convert the voltage, and deduce which fret made contact from that. With as many frets as there are, you'd need a 16-bit A/D and probably a 12 volt rail, but you could use a simpler circuit to check yes/no is there a connection at all, and do the A/D conversion only if there is.

Having said that, I don't fully agree with moonmilk: There are lots of people interested in finding new ways to control synths...

In a world of nearly 7 billion people, you can find someone interested in nearly anything. But that doesn't mean there's a market, or that this is a marketable product idea. That takes a lot more.
posted by Class Goat at 9:02 PM on August 10, 2008

If you are in it for the love of the music or hacking, then go for it. If you are in it for the money, learn to love the 80s synthpop sound and get back to producing boy-bands.

In other words, what Class Goat said.
posted by b1tr0t at 10:25 PM on August 10, 2008

Ah, right, I wasn't thinking about marketability. Class Goat, I concur!
posted by moonmilk at 7:27 AM on August 11, 2008

How you you detect when a string is played, vs just fretted. The note shouldn't trigger on being fretted, but should trigger when plucked. This also needs to capture dynamics - you don't everything to be at the same volume.

But really, its pretty important to capture actual pitch information. Being able to control pitch through bending and vibrato is vital to guitar technique, and without this it would sound flat and lifeless, and defeat the whole point of playing something on the guitar in the first place (as opposed to a synth keyboard for example)

No, you just publicly disclosed your idea, so you can't patent it.
Actually you have 1 year after public disclosure to file a patent.
posted by jpdoane at 10:58 AM on August 11, 2008