Let's Design and Build a (mostly) Digital Theremin!

Rich, what are your feelings regarding pitch quantization (ala the Theremini) versus pitch correction as they apply to real-time Theremin playing?  I'm thinking correction might be too disorienting, but the AC high-pass possibilities of letting vibrato and portamento through seem desirable.

==============

I ordered a bunch of stuff for the prototype, most from Hong Kong and China, so a couple of weeks before things really get going. Currently looking at the screwed up way older LCD controllers are talked to.  Thank goodness the SPDIF TX component is behind me.

Dewster,

If you try and correct pitch (or just quantize) but in the process portamento and vibrato get drastically affected, musicality goes out the window. Pitch correction may well be a useful and desired effect (like the Autotune vocal quantization effect you hear all over now) but if the goal is to create a theremin for "precision" playing, it’s crucial that vibrato and portamento come through sounding and feeling effortless. And this all comes down to the setting. Quantization at some level turns into pitch correction (at least will sound that way). I don't really want pitch correction so much as varied pitch response that possibly gravitates slightly to a tone (but it must have a defeat option for when you want totally smooth “analog-like” response).

So if you are just quantizing without regard to gravitating towards a pitch, that will likely just feel like the instrument can be set to be more or less responsive to hand motion.  In other words it will maybe do what I am suggesting – let you turn the quantization dial and have an Etherwave feel turn into a Burns feel, etc. But turn it too much and it's an effect.

But I also think a small bit of pitch correction will be helpful. Frankly at the lower settings I think Moog did a very respectable job there on the Theremini (their problem is horsepower to handle vibrato). But there are a lot of issues to work out here vs. simple quantization if vibrato and portamento have to be considered in the processing and it could get tricky - for example you might only want it to kick in pitch correction when you sense hand motion (or intended vibrato) has stopped so the players ear can just deal with vibrato as they would normally use it but you will let them more easily hold a desired pitch steady. And I think likely it should not correct to 0 cents but dynamically correct randomly around 0 cents around some offset to sound more musical.

First I suggest you see where variable quantization gets you - hey - its only SW! You can always change it. I get that a lot at work.

Rich

Whatever I end up implementing will be 100% defeatable, and before that certainly somewhat customizable to whatever degree makes sense.  As you, I think for the Theremin some blend of quantization with high pass influence over pitch modulation is important, though AFAIK this isn't implemented in the Theremini beyond simple quantization.  The solution is probably somewhere between how quantization and correction are currently implemented.

The Theremini likely has the horsepower to do all kinds of fancy real-time stuff, but my guess is they didn't properly research their choice of front end, and it ended up owning them in the bandwidth department.  Not their fault I suppose, it takes someone (Bob M.) thinking about this stuff all the time to make any headway at all, and modern companies simply don't work that way anymore (lunch is over, down on your knees).

Alternate Tuner Layout

There are two ways to do the note display (above right): (1) the way I had it in the original video, and (2) the way shown above, where it is rotated 30 degrees and the notes moved over one.  I think I'm partial to the above.  I don't like the way it doesn't have C on the outer ring, but the "house" pattern of C maj / A min scales are perhaps a bit more contiguous looking to my eyes, and the A note is directly above the C note which seems mildly fortuitous as well.  Given rotation, there are only two major / minor scale patterns regardless of the starting note.  I would of course prefer one pattern, but I have a feeling that isn't possible to do and still get an easily recognizable pattern to the eye (I'm all ears to suggestions of any sort).  With both (1) and (2) the B/C note transition (i.e. octave switch point) is located on the left horizontal axis, so that's a wash.

The octave display is a 7-segment deal on the left.  It only changes on the B/C note boundary, so a single number may make the most sense here.  Numbers are fairly universal when it come to Earth-based languages.  

All LEDs are 10mm diffused high brightness through-hole types, and I believe there is enough empty room on the board for the driver ICs and associated passives.  [EDIT] Will most likely use red LEDs for the octave digit as the 1.7V forward voltage drop can support two in series with the driver IC @ 5V.  The others are singles so they can be any color (all other colors have higher forward voltage drops).  May make the inner ring a different color than the outer, with the center LED white or something.  I ordered a large assortment of many colors for experimentation along these lines.

==============

My wife teaches private piano, and one of the materials she uses a lot lately is the "Pattern Play" series by Akiko and Forrest Kinney.  Being a guitarist I'm a huge believer in pattern-based learning, and it's interesting seeing it applied to keyboard.  (As an instrument designer, what might be the most pattern-centric and most playable configuration of keys, what would those keys look like and how would they function?)

Can I just toss in my idea for Pitch Guidance. (Would require stereo pitch preview earpieces.)

With Pitch Guidance a tone that is exactly in tune would be located centrally in the stereo pitch preview field. A note that is ≈50 cents sharp would be located purely in the right hand earpiece (or left hand for left handed players)  and a note that is ≈50 cents flat would be located purely in the left hand earpiece (or right hand for left handed players). The effect would not be audible to anyone but the player.

I have not tried this, it is just an idea. The idea is that to correct the pitch you would move your hand in the direction required to locate the sound in the centre of the stereo field.

GordonC, isn't stereo image perception based on phase difference rather than pitch difference?

GordonC, nice idea!

Bringing the idea to its logical conclusion: the apparent location of the sound source might be not at the right/left, but around the head.

Dewster,

What posting page was the original description on? I forget how this all works.

A stereo field to indicate pitch? Some of us don't hear the same way in both ears, some may not hear at all in one ear. And once you start playing with other instruments you are never going to be able to hear that stereo field right, plus stereo earpieces for pitch preview will not let you properly hear yourself. 

A nice inventive idea - but in practice I think it will have serious shortcomings.

Rich, I posted a drawing of the original (and a link to the video) one page back in this thread.  Let's try out the spankin' new post link addressing feature!

http://www.thereminworld.com/forums/T/28554?post=209081#209081

Here is the original note part redrawn on the same background as the proposed new:

I think I'm back to liking the original better - gaaa!  

C is on the outer ring, A is in a direct line to C, and somehow it just looks cooler.  And there is more useful spare room for the driver ICs and cable connector.