Modelling the theremin wave in software.

Hi everyone.

I'm trying to model the theremin sound in software (using SuperCollider or Csound).

It's really surprising how hard it is to find any analysis of the theremin timbre, you find everywhere that it's just a sine wave when it doesn't sound anything like it.

I found this post http://theremin.tf/category/projets-2/open_theremin/ that explains a way to generate a waveform with brightness and waveform controls.
Unfortunately it doesn't take frequency and amplitude into account.
If someone can explain how the waveform changes with frequency and amplitude, that would be great. (if possible using simple mathematical expressions like the ones from that post, my electronics knowledge is a bit weak and most of the stuff you guys discuss here goes way over my head)
My intention is to create a sound for the Linnstrument (a 3d controller like the continuum), and I like the theremin timbre. I'm aware that, this won't replicate a theremin (and I don't want it too, I'm playing it polyphonically for one thing)

I tried contacting the author of that post but he hasn't answered yet.
Thank you.

Look at Thierry's last post in the Open.Theremin thread - he references just what you are looking for there.

http://www.thereminworld.com/Forums/T/30942/is-openthereminuno-worth-it-cannot-it-be-drastically-simplified?post=209706#209706

"Look at Thierry's last post in the Open.Theremin thread - he references just what you are looking for there. http://www.thereminworld.com/Forums/T/30942/is-openthereminuno-worth-it-cannot-it-be-drastically-simplified?post=209706#209706"

He references the same article that I linked which doesn't take frequency and amplitude into account (because it generates a static wavetable for the open theremin).

EDIT: I see that he is the author of that article, I hope he sees this  . To be clear I liked that article a lot, it was the only usable model that I could find on the internet that I could understand.

What exactly are you after, Fourier plots in different ranges of the theremin so you can write a program to add up the sine waves at the right amplitudes? There's not much of that on the web. I use an iphone app to get the plots but they are likely not super accurate.

Hello Rampoina,

When I was active in theremin research, the classic sound is what I was always looking for.

http://www.thereminworld.com/Forums/T/30784/finding-claras-voice-she-is-vocal

Most at TW were into the digital approach and stated that any theremin sound could be generated this way. Ten years later I have to conclude these people are mislead or delusional. I got older and may not live long enough to find out if my simple analog method was the better approach.

Sometimes I think my friend Paul Tanner and Brian Wilson mislead everyone in what a theremin should sound like, surely not a cheap whistle. Paul with a tear in his eye did apologize to me before his passing that theremin lovers may have been mislead.

http://www.oldtemecula.com/theremin/tanner/index.htm

Imagine being part of a number one recording in all the world to only be reminded constantly that it is not a theremin when Paul himself never said it was.

Christopher

@rkram53

I know there's not much of that on the web, believe me, I scoured the web for weeks. Yes, fourier plots (or rather the data from those) would be helpful, and I thought of buying a theremin and doing that myself before finding Thierry's article. Ideally I'd like him or anyone to tell me how the timbre varies with frequency and amplitude.

Quoting his article:

The filtering can not be taken into account when generating simple wavetables since these force us using the same waveform for whatever audio frequency, thus we’ll have to rely on the Open.Theremin’s post DAC low pass filter.


So I know there's some kind of filtering but not how to model that (or what frequency it's filtering, etc).


@oldtemecula


I hope you can achieve what you're looking for!, best of luck. The classic sound is beautiful, although I have to say that I like the modern sound and I don't think it sounds like a cheap whistle at all.

OT wrote: “Sometimes I think my friend Paul Tanner and Brian Wilson mislead everyone in what a theremin should sound like, surely not a cheap whistle. Paul with a tear in his eye did apologize to me before his passing that theremin lovers may have been mislead.”

Paul Tanner called his instrument the “Electro-Theremin”, which I think is a misnomer since it seems to imply that a regular theremin is not “electro”. I coined the word “Tannerin” in the late 90’s because I thought it might clear up the confusion. 

I think I may have added to it!

The instrument is essentially the “ruban” of an ondes martenot, and Mr. Tanner could have more appropriately called it an “Electronde”, but that name had already been used by Martin Taubman for that other theremin ripoff that replaces the loop antenna with a cut-off switch held in the left hand, while a foot pedal controls volume. I call Taubman’s instrument a “ripoff” because it had already been invented and demonstrated by Leon Theremin in the 1920’s, and called the “Etherphone”. 

OT, you can’t go anywhere until your work is done. WHICH IT ISN’T! 

Ever wonder why Stephen Hawking is still alive when he should have died decades ago?

Yes, unless Tanner modified the Heathkit A0-1 audio oscillator on which it was based, the Electro-Theremin could only do sine or square waves.

Rampoina, you bring up a very good point in terms of how to "do" Theremin tone synthesis. I suspect a lot of what we consider to be The "Theremin Sound" is due to oscillator coupling.  (But too much coupling can cause non-linear pitch field problems in the real (analog) Theremin.)

I doubt if anyone here will be able to help you more than yourself in this endeavor.  There are a number of oscillators used among the various Theremins, most produce a sinusoid at the antenna (a notable exception here is the Thierrymin) but the internal pick-off point for mixing isn't necessarily sinusoidal.  Then you have the mixer doing things to the waveform in terms of coupling, distortion, and filtering (low pass to select the audible lower heterodyned image from frequency subtraction).  And finally the amplitude modulation circuit has its way with the waveform.  So it's a gauntlet of distortion and filtering, performed by quite crude and simplistic electronics, that rather predictably doesn't always yield musical voodoo magic IMO. 

You might be better off playing around in SuperCollider with various traditional additive and subtractive synthesis methods to see if you can replicate that buzzy Clara sound (if indeed you are after that buzzy Clara sound - I find it to be fairly harsh).

This is actually a pretty decent free spectral analysis plugin for PC or MAC:

http://www.bluecataudio.com/Products/Product_FreqAnalyst/

BlueCat has suite a few free plugins and their other products are reasonable. I've been meaning to run a number of spectral plots of my theremins over low mid and high freq ranges to see how the harmonic content changes with range. 

Rampoina - If you think this will be useful to you, I'll give it a shot. Though I expect Thierry will be able to give you more valuable information here. For sure it's no sine wave.

Even just looking at the the Burn's pro - which has a more sine-like sound - it's obvious it's not at all just a sine wave (this is just from my iphone sampling a fixed theremin tone at ca. 220 Hz - just a fixed object in front of the antenna so there is no vibrato on it)

I suspect you could get a decent thermein-like sound just by adding up the first 8 or so harmonics in a relationship like this. Maybe throw a random number generator at slightly changing the harmonic amplitudes over time. But to get it to really sound like a theremin you will have to add the right envelope, vibrato, etc. As Thierry points out the spectral shape is dynamic, but frankly I think you could create many waveforms that might sound very much like a theremin if you played them correctly - and that's more your csound problem. How do you get that slow but yet distinctive ADSR shape? How do you impose a variable and musical vibrato that makese sense. Without that it's just an oscillator without much character if you ask me.

But the time it will take to program the intangibles that make the sound sound like a theremin, frankly are prohibitive - at least to me.

SynthMaker has a couple of FFT imager plug-ins though I haven't played with them yet. 

My go-to FFT is the one in Adobe Audition 3 (which you can find "free" on the web - released into the wild by Adobe due to the retirement of the license server).  Audition 3 is quite nice for all sorts of things audio.