Theremin Oscillator Frequency Questions

Thank you both for the wealth of info.

If, as you suggest, heterodyning is not necessarily the key to the theremin sound... what can you suggest?

I totally understand that each individual instrument has its own variation of the sound, and I also understand that I won't be able to replicate with 100% accuracy even a "generic" sound. All I'm trying to do is produce something that is close. Obviously, just sinewaves doesn't cut it.

I had read about the heterodyning in some other places and that seemed like something I thought I could emulate. But apparently, it's not as easy as I thought.

Thierry, you mentioned "post mixer wave shaping by adding controlled distortion." Can you elaborate on this?

Other than that, I am open to any suggestions...

Just study the schematics of any theremin and you will see what I mean:

The Theremax uses a comparator to transform the sine wave in a symmetrical square wave. After that there is a pot which allows panning between pure sine and pure square with an infinite number of sound variations between the both.

The Moog Etherwave Theremins use clipping of the signal by overdriving OTA circuits where the degree of clipping and its symmetry is adjustable. An adjustable low pass filter allows still more variation at the end.

The B3 theremins start with sawtooth oscillators (inverse exponential ramp). After the mixer and a fixed low pass filter you'll find a slightly distorted triangle wave.

The tVox Tour theremin allows panning between sine and linear ramp triangle wave.

As we found in earlier discussions the output waveform is only one and not the most important factor of the theremin sound. It depends much more on the player how he/she shapes the sound by making transitions between tones more or less quick and audible and by adding vibrato, how the envelope of the tones is modulated with the left hand, and so on...

That's only one of the reasons why you will never be able to perfectly emulate a theremin - the human factor is much more important than technical tone generating aspects!

If you hear theremin professionals like Lydia Kavina or Peter Pringle you may be able to identify them clearly, mostly independent of the theremin they are playing on.

I would hazard a guess that if you want to simulate theremin sound in software you will need to something similar to "amp modeling" used in effects pedals.

Rather than try to synthesize 2 not-completely-linear RF oscillators, a non-linear mixer stage and a filter stage to strip off everything except the fundamental and harmonics of interest.

Unfortunately, you would end up with about the same thing as an analog synthesizer ... which is not the goal here.

The electronic "charm" to theremins is that they use fundamental hardware blocks to create a desired musical timbre by interaction of their own imperfections. They are not "models of ideal components" interconnected to calculate expected result. And as Thierry's discussion alludes to ... it's much harder to simulate all these imperfections than to just create them in the first place.

Don

Thierry and djpb, thanks very much for the additional insight.

I can appreciate that discussion of emulation of the theremin sound within a community of thereminists is probably akin to asking virtuoso violinists how I can replicate the sound of a Stradivarius with a Roland Juno 106.

I completely appreciate and understand the theremin sound is owed as much to the technique and prowess of the player as well as the individual unit itself.

What I am after is merely an approximation.

I truly admire the instrument, the history of it, and completely respect the dedication and passion of its base of players. Had I the money I would own one myself and already be devoted to learning the art; and this is still a dream attaainable at some point in my life.

For now, I am working on fulfilling my educational requirements and emulating the theremin is my short term goal. I am merely hoping to achieve as "authentic" a sound as possible in my software; which is why I seek, and am grateful for, your guidance.

One thing that is very conspicuously lacking in the continuous pitch instruments on my iPod touch is the ability to do a fast smooth glissando. They just don't poll the sensors - be they the touch screen or the accelerometers - often enough, and a smooth gliss turns into a stepped progression.

Probably the resolution on the Ipod screen does not allow enough space between notes to produce a good glissando.

I had this very issue myself on my application. I am using screen X-Y coordinates and pixels to represent points in space with which to register notes. When I was using Integers, I was getting stepped notes, but when I switched to double data types, I was able to get the "in-betweens" and therefore produce a good gliss.

I will be posting more information about my project once I am at a point where I feel more confident of a reliably functional state.

The interface, I hope, will allow for some of the technical nuances you folks have described here in terms of the way in which it is played.
Aside from struggling to design some good sounding "patches," I am working out some kinks with getting MIDI out to work, which I think would be a cool addition to allow for incorporation of a wider sonic palette.