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

Too Close Encounters

I often think about what voices might work best with Theremin, particularly in a performance setting.  It's really interesting to have a varied assortment at your fingertips, but how useful are they really?  They force you to play quite differently if you want them to sound as realistic as possible, which seems like a positive "stretching" sort of thing.  And it's instructive to craft, say, a bass clarinet sound within a given synth architecture.  But is ultra-realism all that useful?  I mean, we've all seen/heard keyboardists do distorted guitar and it looks like fun, but it almost always comes across as kinda pathetic - what's the point?  What is the use case other than noodling around or goofing off?  Is there any way to genuinely use it musically?  If I want to hear an electric guitar, please play that, there are too many "tells" when a keyboard is used as the controller of a guitar sound - the voice itself may be perfect, but the performance will always be a poor approximation of the real thing.

The Haken Continuum is faced with this conundrum too.  The synth lashed to it is absolutely incredible, but is the best use of it to exactly produce squeaking strings?  Honking woodwinds?  I'm not faulting the creators for working on these kinds of voices, I do the same thing (but with necessarily inferior results).

Does this account for some the attraction to analog synths, that they aren't usually trying to ape traditional acoustic instruments?

Some Theremins offer a middle ground between the boring oscillator sound and hyper-realistic traditional sounds by being string-like in some registers and female-like in others.  But it's a tough row to hoe via the vagaries of heterodyning and coupling, and some enhanced form of this might be more ideal.  I do think formants and resonators offer a lot of promise in this direction, and the analog synth approach not so much.  The suggestion of a natural sound, or a blend of them, might make for the most versatile type of Theremin voice.  Something that obviously isn't trying too hard to emulate a specific instrument.

That said, I can't stop playing human vocals - the more human the better! - on the D-Lev.  The magic of C fields coupled to the magic of a simulated vocal tract go together perfectly.

"Simply lowing all the formants yields a similarly very pleasant and expressive tenor voice" - Dewster

Yes, that's probably first idea I would follow by intuition: Greater vocal cavities on anatomic side, slightly lower formant on synthesis side.
Larger vocal string means playing in a lower register (and maybe adapt picth offset to do that comfortably / not to much close to the body).

All is, of course, a question a measure whether you want to get something realistic, artificial or maybe just funny (really easy to fall into the third case by lack of measure).

Voices are not interesting just because they are realistic, they are because they are expressive. Expressivity is the real quest for a synth.
And particularly of a synth commanded by a theremin: Sometime we might feel like singing while playing it.
(When I was a beginer, I even found myself trying to correct false pitch with my pharinx before realising that the problem was in my hand... )

Expressivity means that I can control the sound in the direction I want to go, vmod and pmod are the tools for that.
It could be also interesting to explore more the synthetic side of the D-Lev, the Phase Modulation oriented structure of the oscillator allows that (try some "vmod" on "harm" level... ).

"Yes, that's probably first idea I would follow by intuition: Greater vocal cavities on anatomic side, slightly lower formant on synthesis side." - Mr_Dham

It's pretty easy too, IIRC just 4 or so clicks down on every filter.

"Voices are not interesting just because they are realistic, they are because they are expressive. Expressivity is the real quest for a synth."

Very nicely stated, and I totally agree.  I think this gets overlooked a lot with keyboard type controllers because, other than aftertouch, they generally rely on automation to get "movement" in the sound, like complex ADSR and arpeggiation.  But it usually sounds mechanical and boring after a very short while.

"And particularly of a synth commanded by a theremin: Sometime we might feel like singing while playing it."

I've definitely felt that too, it's very close to singing.  My wife tells me I sometimes breathe funny when I'm playing.

"Expressivity means that I can control the sound in the direction I want to go, vmod and pmod are the tools for that."

Glad to see you're digging into the synth some!

"It could be also interesting to explore more the synthetic side of the D-Lev, the Phase Modulation oriented structure of the oscillator allows that (try some "vmod" on "harm" level... )."

Yes, almost all of the presets utilize this as it is such a natural thing (increasing harmonic content with increasing volume / decreasing pitch).  And many of the FM type presets use it to radically change the sound.

Organic!

Trying to work on tVox voices, so of course am off on a total tangent.  Octave above and below with a bit of harmonics gives a pretty passable roller rink organ sound:

https://d-lev.com/audio/2022-01-06_organic.mp3

Lots of knee and some velocity to give an attack sound, with some volume expression pedal effect above the knee.  Fairly tight pitch correction, but it still lets vibrato through - it's fun to fake the rotating speaker with it.

Our Betters

I read a saying the other day: "better necessarily implies different."  If people are ready to accept something improved then they have to be ready to accept a certain level of change along with it.  Different isn't always better, but better is always different.  This doesn't give you carte blanche to make drastic changes for little reason though.  I guess you can tell I'm talking about non-traditional features on a Theremin here.

And then there's: "better is the enemy of good" which I can't help but interpret as lazy corporate speak for "get it out the door already (bugs and all)!"  Never pass up an opportunity to improve your product, polish it like a gem and it will polish your skills and hone your craft like nothing else.

"If you knew how much work went into it, you wouldn't call it genius." - Michelangelo

Organic!

Very realistic!

I tried to reproduce it based on the instructions given here. I got a pretty close result but it is very difficult to play with the same realism.
I also have an organ, so maybe I don't need this sound so much, but playing with it is a worthwhile exercise by itself.

How did you set up the pitch correction ?
I have the feeling that the key is there.

"How did you set up the pitch correction?"  - Mr_Dham

So sorry, I should be sharing these.   I just added a presets link table on the support page:

https://d-lev.com/support.html

I made the pitch correction a little tighter after I recorded it.  The patsy tenor preset is there too.

Much better than what I could get. Thanks for sharing !

A slow and not too much wide vibrato (slow rotary speaker) is very hard to achieve.
Acceleration and deceleration are even more complicate.
Very interesting exercise for the player !

"A slow and not too much wide vibrato (slow rotary speaker) is very hard to achieve." - Mr_Dham

Maybe try increasing PITCH:slew, which will let more slow movement through.  This is the knob I tightened up post recording.

"Very interesting exercise for the player !"

I think so too!  Very different voices often require very different playing styles (pitch stepping & vibrato, volume envelope & articulation) to maximize the realism, which stretches one's playing technique.

tVox Tour Samples & Presets

Lydia Kavina kindly supplied me with five different voice samples of her tVox tour Theremin (which was designed by her husband, George Pavlov).  The samples consist of a slow gliss from very low to very high frequency, followed by two or three fixed notes per octave.  Lydia also kindly gave me permission to share them (mono, converted from *.wav to *.mp3):

https://d-lev.com/audio/Lydia_tvox_samples.zip

I've spent more time on these than most other D-Lev presets, and I wish I could say they were perfect, but they're not.  The synth in the D-Lev is designed more for vocals and strings, and so can't model certain things going on in the tVox.  For the non-sharp switch samples I found a suppressed second harmonic to be the main characteristic of the timbre, which can be accomplished with the oscillator tracking filter set to 880Hz (+1 octave) notch.  For the sharp switch samples the raw non-filtered oscillator sound is a better fit.

https://d-lev.com/presets/tvox_presets.zip