Is one great voice enough?

Wasnt sure which forum to post this, or even whether to post it at all.. Its another of these questions which I have often asked, got great feedback on, and done nothing with..

So please, understand that this is probably hypothetical and will probably come to nothing..

I have been playing with heterodyning mixers, pre-mixer wave-shaping, and formant "crafting" for a long while now, and have hit against something of a brick wall -

My problem is cost and complexity. I wanted to create an instrument where the player had control over the sound to a far greater level than on any present theremin.. In my head there's an instrument that can be adjusted to sound like a Claramin through to a MiniMoog..

The MiniMoog is the easy part ;-) I now believe that to replicate Lev's theremins, one cannot use subtractive synthesis - I was convinced that additive synthesis was the way forward - but my latest simulations convince me that the sound can only be achieved with a complex combination of the mixer waveform, formant filters, and seperate tone equalizers for each formant.

To do this in a user adjustable way one needs a massive control panel - for example:

6 waveform controls [waveform 1, Waveform 2, Morph, FM, Tilt 1, Tilt 2]

6 Formants each needing 4 controls (= 24 controls total) [Frq1,BW, Tilt,OP Lev]

And thats just for the formant section, doesnt take account of the other voicing stuff, tuning, preview etc.. Id be looking at about 50 potentiometers!

I have been looking at small finger-adjustable preset potentiometers (I used these on my H1's), they could be fitted under an accessible lid, so could be adjusted by the player - but they are not robust enough for frequent adjustment, and would be quite tightly packed together..

A few (perhaps 10) high quality potentiometers would be on the front panel, to allow some adjustment of the tone (and adjustment of tuning, preview, output etc) but the primary sound would be "programmed" by the controls under the lid.

If you had one great theremin voice (and im talking about one particular class of sound - the vocal type sounds from Lev's instruments - it would be a waste if you want the sound of modern theremins) you could program and forget, would the theremin appeal to you ? - Oh, this theremin will probably be about twice the price of an EW+ - or an add-on for the EW costing about the same price as an EW+ I would guess.. (would require a multi-pin socket on the back of the EW, small board in the EW, and an 'expander' box containing the voice circuitry)

I am hoping that a few (perhaps 2) controls on the front panel will give quite a lot of variation in the tonal qualities (probably more than the EW controls do) but the sound will primarily be determined by the settings of the formant resonators - and these are designed to provide two complete formants (as in, at least two formant filters) which morph into each other (one increasing in level as the other decreases) as the pitch changes - These formant filters are configured as independant individual formants, so can be arranged in any configuration - a 2 formant + 4 formant, a 3 + 3, or flipped as a 4 + 2 (the best results have been with 4 formants at the bass end and 2 at the treble)

I have been working on this with a view to incorporating it in my ribbon instrument - Its a lot easier with this because its 1.2m long and the front panel can accomodate more knobs - also, its modular, and its possible to have  add-on units with different mixer and waveshaping stuff in these -

But I think (must emphasis that its only simulation) this scheme MIGHT produce results closer to the sound from Lev's theremins than anything available now.. It just feels that this stuff should be tried with a theremin first.

Oh - The above wont replicate a Mini-Moog as well ;-) .. I kind of got focussed on the Claramin.. Its easy enough to add a seperate "synth" voice later.. So this voice is focussed on adjustment of the waveform into the formant filters, and adjustment of the frequencies and levels from these filters, and adjustment of the levels from each filter as a function of the audio frequency from a seperate Ambler 'tilt' equalizer that follows each filter - then there are also some global filters and distorters to try to replicate the behaviour of tubes and transformers - all of which are adjustable.. But it doesnt do much else! ;-)

And like I say - its in my head and in my "virtual lab" but nowhere else - With my present situation the best realistic hope I have is of making one prototype, and publishing a video / sound and the design - but who knows...

I think I have a better chance of getting this "out" because im not thinking about production anymore - just plan to knock it together on breadboard without my usual agonising about every component.. If I did ever want to produce it I would probably need to do a major redesign..

Fred.

I like multitudes of voices. the 3 theremins I have all sing differently...

"I like multitudes of voices. the 3 theremins I have all sing differently..." - Amey

Interesting comment that! ;-)

Your 3 theremins all sing differently - But do you adjust their voices, or do you have a favorite setting on each?

Also, when (if) you adjust them, do they still each have a vocal "identity" that is unique to them?

You see, this was my original target - A single instrument that could sing like any / every other instrument, so you only needed to buy one theremin...

And I believe this is possible for modern (Non Lev, and probably Non-tube) theremins - But when it came to getting a voice like Lev's theremins, well - I cant do it with simple waveshaping, or even complex waveshaping.. I cant do it with voltage control, and although I believe its possible to do it with additive synthesis, the complexity is huge.

Purely hypothetical - If you could buy a theremin which sounded like your favorite old-time theremin, but had limited ability to sound like anything else, would you be interested? ;-)

Technical:

The circuitry which I think MIGHT do it, uses formant filters - its what I now believe happens in the Lev (and possibly some other tube) theremins due probably more to the mixer and audio-side transformers and resonant stuff there, than anything else.

Ao lets IMAGINE that you could get a theremin which couldnt sound like a EW, or an EW-Pro, or a Subscope, or a Tvox, but had a couple of controls on the front which you could adjust and it could sound like a RCA or a Claramin or be a reasonable replication of any Lev built theremin and some other tube theremins..

You would be able to fine-tune the sound for your one particular favorite instrument, but this would not be an easy task, so you wouldnt want to do it often.. This chosen sound would be the main "charactaristic" voice of your theremin, and whilst there would be quite a wide tonal variation on this voice available, the thing which gives it its "identity" would be its formants - and you would not be able to adjust these easily.. (*Just thinking about it, it would be possible to have duplicate sets of adjustments to say allow 3 different setups, or 'presets' one could select.. but I aint even going to think about that right now ;-) .. If I got this thing built, and it was liked, then the whole matter of engineering it would perhaps be on the table IF I could get anyone to finance such a project)

I suppose what im thinking comes down to giving the musician the ability to "engineer" the sound they want (IF they want the vocal charactaristics of the early theremins)

To me, the ideal would be to have this as an add-on to existing theremins - and I am working hard on this.. I could do it as an add-on, but at present it requires getting into the existing theremin and adding a circuit to tap the required signals, and bring these to a connector... But I am hoping that I can evolve this to a box that only needs to take the audio as an input.. But I have a way to go yet on that challenge ;-) .. I also know that any such box would be inferior to doing it at the heterodyning mixer, even if it gave some passable 'replication' of the formants (creating the changing harmonics at the mixer, to drive the formants is the only way I found that looks / sounds convincing - so I need access to the HF oscillator waveforms)

And once again, I must caution - The above may never happen, has not been proved, and may be completely delusional bullshit!

Fred.

I suppose the real question is this -

IF there was a modern theremin that REALLY sounded like a Lev theremin (RCA, Claramin) BUT wasnt a Lev theremin, didnt have the history or value or romance of the Lev theremins, and was twice the price of an EW+ or more (say £600 to £1000) would there be any market? Do people REALLY want the sound of the old theremins, or do they just pay lip service to this? Is it actually worth persuing this direction... I actually need to ask myself whether I think the sound of the Claramin is REALLY that great.... I am inclined to say it is.. But how much of this is really down to its sound? Would I buy a Claramin for say £800 if it wasnt one built by Lev ?

Hi Fred,

Claramin sound is definitely great, but I would't say the sound of Subscope or TVox is worse. It's just different, and Clara's tremendous playing skills are of great importance here...

Since Claramin sound is rich of harmonics, in some cases at least another one timbre is needed (closer to pure sine wave).

There is no need in 50+ potentiometers anyway: you can add a touch screen or radio module to control these parameters from the tablet pc through, for example, bluetooth. I can help here.

For £1000 it should be a professional-level theremin with the corresponding stability and linearity parameters (otherwise, the demand will be very slow). However, in this case, it will compete with much more expensive second-hand professional theremins, so there's no need to pay so much attention to the tone characteristics (the price, the fact it's brand new, and the ability to order some customization from you would be killer features).

Ruslan

Hi Ruslan,

Yes, a digital interface to NV digital potentiometers would allow for programming of multiple presets.. If there was demand, that would probably be the route to go.

I also completely agree with you on the issues of linearity and stability etc.

And I also fully and wholeheartedly agree that there are other theremins (Tvox that I have played, and Subscope I have heard good reports about) which have lovely tonal qualities.. This isnt about "better" - thats almost entirely subjective..

I am also not really looking at production at this time - To be damn honest, I think that ship left the dock a while back - Oh, it may come back, but I have my doubts.

My only tiny hope is in my ribbon instrument I think - and much of the market for this probably isnt from those who frequent TW.

BUT - I have all this research and development and time and money that I have spent chasing the "Lev" sound.. And even though it wont make one iota of difference to "me" after my annoying brain finally goes to oblivion, wasting all this stuff is annoying me now!

I can probably get one prototype knocked together without having to spend more than another couple of days on design (wheras if I did the sensible thing and went to a digital input like you suggest, I would be spending another 6 months+ on design).. Depending on my lab access and other domestic factors, I could probably build the instrument in a month of work.. I could then video it, record it, document it, publish it, and be rid of this damn obsession with these bloody theremins!

IF there was a load of people screaming that they want to buy one after having heard / seen it, well, the matrix might change - and at that stage I could look at going the sensible route and get help from people like you ..

(Im not going to be messing with the front-end on this instrument at first - Ive got a new EW board, and will be making an interface board for that.. I have the idea that this will give me a simple option to supply this as an expander module at first If demand was really there - with the option to do my own front end and digital interface later .. This instrument is intended as a proof-of-concept for the sound, and not for much else.. But with a small dedicated interface board, almost any theremin could have this expander)

But I have wasted too many peoples time to date, started way too many things that have ended up going nowhere, and raised peoples hopes and expectations but delivered NOTHING -

This enquiry was to test the water a bit about the issue of tone choices / availability and how people felt about it.. You and Amey have so far come back with useful guidance -

A few more controls and circuits is all it would take to add a couple of "modern" theremin voices to this instrument.. One could perhaps have 3 'volume' controls so these voices could be blended - 1:) Preset "Formant / Lev" voice, 2;) Softer "Sine" voice 3:) Mixed signal "Synth" voice, and a few tone controls on the front panel for each.. Oh, Ive been forgetting that I also have the normal (except better because the oscillators are buffered by the interface board) EW voice - Perhaps thats all I need - a user presettable "Lev" voice and an EW voice... perhaps blendable and an Ambler on the EW voice..

I had sort of thought that people would be after a realistic "Lev" clone more than anything else, but I did have a niggling doubt about this - My hunch has always been that a multi-voice theremin is what people really want - I think my obsession with including  Lev "cloning" in my scheme was a big mistake - A '91 with a few more synth controls is I think what most of the world wants! ;-)

Fred.

Not being a player, and not being a huge fan of the Claramin sound (too nasal IMO) I probably shouldn't chime in here (but that unfortunately never seems to stop me).  And I realize you're talking one-off / prototype, and not production, which are entirely different things.

A multitude of controls seems to almost require some kind of preset mechanism just so you can get back to a certain voice in a reasonable amount of time and effort.  A preset mechanism in this case seems to almost require a partial / full digital implementation.

Going digital will take more up-front work, but would allow you explore voicing issues more at your leisure, and with a lot less layout, board stuffing, and lab work.  You could leverage your new-found DSP knowledge in other projects like your ribbon.  If you keep doing this stuff you know you're going to have to go digital eventually.  I'm thinking there are a lot of DSP constructs (oscillators, filters) that don't require a lot of math, particularly if you aren't doing mixing board EQ lab quality filtering with super low noise floors and such.  Inexpensive yet powerful enough ARM boards with lots of memory and digital and analog I/O are beginning to litter the design space.  A few trivial adapted Clapp oscillators, an ARM board, and maybe a tiny FPGA and it's all over (except for the fighting - the enclosure & SW!).

(Semi-OT, but I'm aiming for more of a human vocal sound initially because that really interests me.  It's almost as spooky as the Theremin itself!  I particularly want to make a "ghostly" voice that goes from 1. 100% breathy to 2. semi-random excitation to 3. periodic excitation, but does so continuously.  And then proceed to creep everyone out!)

Eeeeeek, I know I am crazy, but I am not really in love with Clara's theremin voice either.

Fred: On my 3 theremins, I adjust the voice on them and I usually do not really change them. On my Subscope, I like it on a full bell/flute sound like. My wavefront I like it on the muffled brassy tone (kinda hard to explain the sound) and on the Epro, I love it on the 3rd setting from the left - its more of a toned down flute sound. I don't like to change the settings, when I do, I usually go back cause it doesn't please me at all. :P

"Do people REALLY want the sound of the old theremins, or do they just pay lip service to this? Is it actually worth persuing this direction... I actually need to ask myself whether I think the sound of the Claramin is REALLY that great" -Fred

"Since Claramin sound is rich of harmonics, in some cases at least another one timbre is needed (closer to pure sine wave)." - Ruslan

"Not being a player, and not being a huge fan of the Claramin sound (too nasal IMO)" - Dewster

"Eeeeeek, I know I am crazy, but I am not really in love with Clara's theremin voice either." - Amey

------------------------------------------

This is real interesting to me, and I would really love to do a poll on peoples liking / disliking of the sound from Lev's original theremins - but not just a "like / dislike" split, more a "I like this aspect but dislike that aspect" type thing..

But so far, it seems that even if one was to clone a ClaraMin to exactly produce its sound (on Clara's one favoured 'preset') of that instrument, there wouldnt be much interest. Is this a case of the "Emperor's old clothes" ;-) - Is this instrument "revered" because Lev designed / built it and Clara played it, much more than because it really sounds great ?

Since my last posting some hours ago, I have been listening to Lev theremins (including the Claramin) trying to be honest with myself, and trying to calm my nerves ;-) .. I couldnt find any examples of a badly played solo Claramin (if anyone knows where I can find such a sample, please let me know) But I did find some examples of other Lev theremins which were horribly played - I think I remember once hearing a vidiocam recording of the Claramin being "unplayed", and at the time being impressed by its sound - but I cannot find this.

I wanted to hear the instruments without being "conned" by the skillful playing - to hear their "raw" sounds without perfect vibrato etc - which is why I want to hear the ClaraMin without Clara playing it.

But what I listened to led me to an even deeper belief that there is a special quality in Lev's instruments that no other theremin I have heard comes close to - Yes, there are "nasal" qualities in some which could be a bit unpleasant, but they all had that "something" IMO .. A "Quality" which I love and which I believe (I may be delusional ;-) makes these instruments (despite them sometimes having an almost completely different sound to each other) "identifiable".. (and of them all, Clara Rockmores theremin stands out as the "best" to my ears - but how much of the reason for this is because she is playing it.. Well, that I cant say)

IF this "quality" is formant shaping, as I believe (and hope) it is, then it should be possible to replicate this quality even on other designs where the mixer can produce less "harsh" harmonics if desired.. You see, I dont believe the formants were "deliberately" added in a controlled manner - I suspect that Lev just fiddled with components until he got the sound to the best he was able to manage THEN**. Things have changed - we cant easily get the components he used, but we have a load of new parts and circuits he didnt have.. And these should enable us to replicate any "magic" in the sound, and perhaps enhance this IF we know what the "magic" consisted of (all assuming there was / is "magic" there in the first place.) and simultaniously get rid of any "harshness" in ways that were not available to Lev, and which he would probably have used if they had beed available to him.

** When I say "fiddled" I do not mean that he didnt know what he was doing! - I think he knew more about what he was doing than most of us could ever aspire to.. But I think that with things like transformers and audio capacitors of the time, it would have been impossible for him to have "designed" exactly what he "wanted" - The final crafting I think MUST have been done by ear.

"A multitude of controls seems to almost require some kind of preset mechanism just so you can get back to a certain voice in a reasonable amount of time and effort.  A preset mechanism in this case seems to almost require a partial / full digital implementation.

Going digital will x take more up-front work, but would allow you explore voicing issues more at your leisure, and with a lot less layout, board stuffing, and lab work.  You could leverage your new-found DSP knowledge in other projects like your ribbon.  If you keep doing this stuff you know you're going to have to go digital eventually.  I'm thinking there are a lot of DSP constructs (oscillators, filters) that don't require a lot of math, particularly if you aren't doing mixing board EQ lab quality filtering with super low noise floors and such.  Inexpensive yet powerful enough ARM boards with lots of memory and digital and analog I/O are beginning to litter the design space.  A few trivial adapted Clapp oscillators, an ARM board, and maybe a tiny FPGA and it's all over (except for the fighting - the enclosure & SW!)." - Dewster

I agree with everything you say above, Dewster - The killer statement  though is "Going digital will take more up-front work" .. And the truth is that I havent got the time or money or inclination for that. Also the "but would allow you explore voicing issues more at your leisure" wouldnt apply to me - I can go to a strange large analogue modular and create exactly the sound I want in minutes - But trying to create a similar sound on a DX7 or D50 or my Karma could take me days (and also be a frustrating boring irritating expierience, whereas twiddling knobs on an analogue is instant and fun)... Actually having to fiddle with parameters at a code level, without having first developed an intuitive UI (itself a major task) lould probably keep me frustrated for the rest of my life... ;-)

IF this goes anywhere, I suspect the "path" migh be something like this.. I twiddle the presets to get the sound I like, mess about and eliminate any redundancy and/or add any extra processing I find or think I need.. "Expose" the results and hopefully get "real" musicians and thereminists to play and assess the instrument -or more particularly the sound/s from the instrument..

I could "tweek" the presets to quickly  craft the sound for the musicians taste, and run a VM down a row of test points to "capture" this preset.. Then at a future time have a full analysis of the processes required to create the sounds that I could hand to a digital wizard to implement in a FPGA or DSP or whatever.

But thats all quite far from where I am now ;-)

"On my 3 theremins, I adjust the voice on them and I usually do not really change them. On my Subscope, I like it on a full bell/flute sound like. My wavefront I like it on the muffled brassy tone (kinda hard to explain the sound) and on the Epro, I love it on the 3rd setting from the left - its more of a toned down flute sound. I don't like to change the settings, when I do, I usually go back cause it doesn't please me at all. :P" - Amey

Thanks Amey ;-)  ... Its real interesting, but it seems that a lot of thereminists use their instruments in much the same way - Each instrument having a special "character" and adjustment being mainly to hone this "character" to the thereminists taste.. Rarely it seems do thereminists radically alter their instruments sound in the way that synth players do .. Possibly because there arent the facilities to do so - But I think more likely because there really is one setting which the thereminist likes far above any other setting.. Even Clara, I believe, stayed with one of the (6?) available presets, and when this sound changed wanted Bob Moog to restore it..

You have 3 different instruments to give you 3 different sounds! "full bell/flute sound" "muffled brassy tone" and "toned down flute" "presets"! ;-) .. IMO its likely going to be  the mixers in these instruments which define their character more than anything else - not sure about the Subscope, but the "muffled brassy tone" is probably a ramp like waveform with both odd and even harmonics, and the "toned down flute" is likely to be predominantly or exclusively odd harmonics, and not many of them..

Technical:

If the theremins oscillators are well buffered, its possible to feed multiple different mixers from the same HF oscillators - Doing this allows the completely different characters of different mixer topologies to be combined in a single instrument, and switched or mixed to taste..

I believe it would be possible for example to take the SubScope mixer and audio stage, the Wavefront mixer and audio stage, and the E-Pro mixer (complete with register switching) and audio stage, and combine this in an instrument which would than have the character of all 3 instruments.

I have done this kind of "multivoice" stuff a lot - even did this to a limited extent on my H1.

Fred.

No one should feel apologetic for not liking the sound of Clara Rockmore's custom theremin. I know a couple of professional, highly respected, classical musicians who have told me they find the tone a bit too "throaty" for their personal taste.  

The most popular theremin sound that has come out in the last few years, and one which many thereminists seem to prefer, is the open 'AH' sound of the Electro-Harmonix TALKING MACHINE. 

Perhaps it is not worth trying to build a large timbre menu into a theremin - thereby increasing the retail price of the instrument. Instead, builders could make a theremin with one voice equipped with the usual filter, brightness, envelope knobs, and make a peripheral device with dozens of different formants, similar to the T.M. but specifically for the theremin, that could be sold separately for those who want to experiment with a wider variety of sounds.

"Perhaps it is not worth trying to build a large timbre menu into a theremin - thereby increasing the retail price of the instrument. Instead, builders could make a theremin with one voice equipped with the usual filter, brightness, envelope knobs, and make a peripheral device with dozens of different formants, similar to the T.M. but specifically for the theremin, that could be sold separately for those who want to experiment with a wider variety of sounds." - Coalport

Yes, I agree that that would be the best solution - an external processor taking the theremin audio and processing it..

And this is probably doable if one isnt trying to make something which will turn the 'source' signal into a near perfect replication of some Lev theremin, but just want formant manipulation.*

Many thanks to you all for your opinions.. I will hold back on building this monster theremin, and build a much simpler analogue formant / filter box first .. This wont be able to turn a theremin sound into a "true" RCA ir Claramin sound - but hopefully will modify rhe sound in ways people like.

There is absolutely no point in creating a Claramin or RCA clone at high cost if hardly anyone would want it, and if a simple box at perhaps 1/3rd of the price or less gives people what they really want.

I will probably still build the original idea at some future time though, just because doing so has become an obsession ;-)

Fred.

*the only real aspect that makes this idea intrinsicly inferior to the "monster" is that one is limited in terms of shaping the input waveform, which one can craft perfectly if one has access to the HF oscillators - I will need to distort the input waveform to get the required harmonics, and this box is likely to sound better with an input having higher (particularly even) harmonic content to start with.. So some theremins will sound better than others through it - This area (the input signal shaper) needs careful design, and there are real difficulties particularly at the bass end and particularly at low volume.