Haken Continuum as a Theremin Effect

https://soundcloud.com/rk53-1/theremin-through-continuum-vocal-formant-bachmarcello

So a bit more experimenting this weekend with the female voice configuration of a BiqMouth module to see if this may be a way to get a nice higher voice simulation through a theremin - where the talking machine kind of falls flat on its face. Playing a Wavefront Theremin through the Haken Continuum's AES port (preamp set to output 24 bits, 96Khz) into an EaganMatrix BiqMouth module configured to a female "Ah" parameter set I came up with (see graphic below).

I arranged the slow movement of Bach's keyboard transcription (BWV 891) of Allesandro Marcello's Op.1 #2 concerto here for strings and theremin (added a number of melodic lines and harmonic changes not in the original to get a melodic line going throughout - the original has large sections with just chords). Changed the chord patterns a bit too to get to the melody sooner (but improvisation was done all the time in the period so this kind of treatment is not totally without precedent).

Here's the EaganMatrix setting for the female formant setting I used if anyone is interested.

Hopefully it's a bit more in tune than the last go round. You can hear the highest note was just a tad above the formant's ability to keep it vocally in tact. In fact I had to transpose the original version of this down a fourth as too many notes playing it in the original key were breaking up. It needs to be down another step or two (but the accompaniment would get too muddy without redoing it). I still need to play around with the settings a bit more and maybe throw a Shape Generator at it too. There is a certain range of about a fifth near the upper breaking point that is really is amplified by the module. 

I found that part of the issues I had was trying to get more realistic filter settings - but a big part is simulating vocal vibrato to make it sound more realistic. That's perhaps more difficult than playing in tune. The right vibrato in this kind of emulation is really key to making it sound more like a voice. Wish I could do that better, but I think with more practice and experimenting more with the BiqMouth parameters, the continuum indeed is a much more useful tool for imposing a higher vocal emulation than the Talking Machine.

Let’s forget the fact that the performance is “pitchy” - you already know that. There are a couple of other things it might be worthwhile pointing out. These are my personal opinions, and do not necessarily reflect the views of this station, WTFF.

If you want to convincingly emulate a soprano (or for that matter, any traditional instrument) you have to play your sound with the soul, and all the constraints and limitations, of the real McCoy. There are two things you should consider about your Eagan Matrix singer: she doesn’t breathe, and she sings at the same volume level throughout the piece. This tends to make her seem robotic and artificial.

As for the timbre, any device you use for your vocal formant will be limited by the sound you feed into it. You soprano gets a bit squeaky and thin when she sings her G5 (the highest note in the piece) but I suspect this is due to the theremin you are playing.

Love the orchestra.

Yes - I was quite ready for your habit of pointing out the musically obvious. At least it's not horrendously out of tune. I can live with a bit of pitchiness (I just wish it was your kind). And the formant filter does detract from the ability to use a controlled wide dynamic range (which I need all I can get of) as it imposes its own dynamics emphasizing some frequencies and attenuating others due to the nature of the cascaded filters. I suspect I need to tweak the parameters to overcome this more.

I'm sure the sqeakiness at the top note is the format filter breaking down at that frequency. Like the Talking Machine, above a certain range things break apart. But don't take my word for it. Try giving it a shot. Would be interesting for you to do up a tenor/soprano aria using the Talking machine and this technique.

And if you thought learning the theremin was hard - try mastering the EaganMatrix. There's a second lifetime of learning I fear to be had there. On to the next thing. I'm thinking of using the theremin as a fixed oscillator feeding a modular amplifier under the CV control of the continuum to better help me learn the ins and out of EaganMatrix W, X, Y, Z control.

Rich, I too very much love your orchestral accompaniments!  And your vocal synthesis explorations are pretty much what I'm looking into this point as well.  I don't have tons of experience yet, but, as Peter points out, the harmonics of the driving waveform can strongly influence the final result, particularly in the upper registers associated with the female voice (perhaps the reason the tenor Ah works so well).  I've noticed the low pass filtering of reverb (along with the general mushiness of reverb) can mask a lot of the weird harmonics of the excitation that manage to squeak through the formant filters, and therefore dramatically heighten the realism.  For instance, I'm playing around with the "SillyChoir" code for SuperCollider, and sans reverb the vowels sound artificial, but with reverb (freeverb2 - not the best IMO, but OKish for vocals - looking a free reverb in the mouth!) I almost can't tell it's synthesized.

rkram, I have a confession to make. My knowledge of the Continuum is practically nil. I don’t even have an AES converter, and everything I have done on the instrument was done with the built-in presets using the analog outputs from the stereo headphone jack.

There is a growing army of synthesizer enthusiasts who seem to be focussed exclusively on exotic interfaces and coming up with new sounds. There’s nothing wrong with that, but it is essentially a technical activity - not an artistic or musical one. These people are often unable to do anything with the sounds they develop other than demo them for others who share their interest. Understanding the technical side of things, and keeping up with the never ending tsunami of updates and innovations, takes so much time and concentration that nothing is left over for either the study of music, or the development of the skills needed to play the devices that are the focus of their attention.

There are many videos on YT of interesting new interfaces that are demo’d by their inventors, who haven’t the slightest idea how to play their own inventions! The more enlightened among these individuals will sometimes preface their demonstration by apologizing for the fact that they can’t actually PLAY the thing. 

…..now where the hell was I? ….oh yeah….I have just ordered a Behringer ULTRAMATCH PRO (entirely on YOUR recommendation) and Lippold Haken has just come up with a new fingerboard playing surface that reduces friction and allows the hand of the player to move more freely. You may have noticed that the silkscreening technique used to print the notes on the neoprene surface, leaves a slightly raised image (only a tiny fraction of a mm) with a different texture. I understand the new printing technique is absolutely smooth throughout. I had tried making a new playing surface of my own by printing onto heavy satin but it didn’t work. Lippold experimented extensively with surfaces during his R&D period and I doubt I could come up with an idea that hasn’t already been tried and rejected. 

For years, I have watched people do this very thing with the theremin, experimenting with all sorts of ideas and “improvements” to the instrument, unaware of the fact that the early thereminists experimented with these things decades ago, and for one reason or another, rejected them. These ideas include such things as placing the rod so that pitch is controlled by the left hand instead of the right, reversing the action of the antennas so that volume is increased by moving the hand closer to the loop rather than farther away, etc. etc.

I’m all for experimentation, but it is often undertaken by newcomers to the instrument who have not acquired a sufficiently advanced understanding of the theremin to know why their brilliant ideas may not be so brilliant after all.

And therein lies the rub, Peter!

If I had your talent, I likely would move in the same direction. But the time it would take me to play like a true professional (assuming I have the innate ability) is prohibitive in relation to all the other musical things I need to do after I get home from the day job figuring out how to connect the world with undersea fiber optic cable (people have no idea what is on the bottom of the world's oceans).

But the techie in me compels me to experiment. And perhaps if I come up with any great sounds I will of course share them in the process. Lippold says that few actually try and use the EaganMatrix - which takes literally years to fully master - as its taken them decades to refine all this.

I need to experiment a bit more with the UltraMatch Pro. Lippold says the continuum should be able to support 24 bits at any rate and it does in fact from my preamp's AES output. I assume it will too from the ultraMatch Pro. I'll report on that as you want to set it more than 16 bits. 

As for experiments - some succeed, some fail. None are brilliant. Hopefully some are of interest. Maybe part of the reason I like Gordon's music so much.

The really amazing thing is after a year and a half, my interest in the instrument has not waned a bit. If I was really smart - I would stop buying all this crap and just practice!!

Rich

I retested it and Ultramatch Pro works fine through the Continuum's AES input at 24 bits, 96KHz setting (Ultramax's max setting). No dithering.

I only connect the UltraMatch AES out to the continuum running my theremin into Ultramatch's Analog in (need XLR for that - and I also run through a DI box) and then I run the continuum's analog stereo out to my mixer. But the UltraMatch is a full A/D and D/A converter so you can also run the continuum's AES out through it as well if you have a need for that.

In the case you are connecting the continuum's AES outputs, according to Lippold you should "slave to incoming" rather than fixed sample rate, unless your device can auto-detect and adjust sample rates. But I don't see a need to run the  audio through another round of conversion unless there us some reason you want to use the AES out (I can't think of any right now as I don't record to DAT or anything like that that might take the digital signal).

I think I might have that new surface as I can't feel a bit of difference anywhere on the fingerboard. Actually I would not mind a bit of a raised edge on the lines that is barely perceptible with a little white guide line between adjacent "red notes". For melodies could care less but a little bit of a guide for playing chords might be helpful. It's so easy to fudge up on chords if things are moving at any speed. You really have to get used to setting your fingers on the narrower grid vs a piano for multi-part playing. Of course, when I play piano I don't necessarily look at the keys. Hopefully I'll get to a similar point on the fingerboard. 

The difficulty with trying to play the Continuum without looking at your hands is that unlike the keyboard of a piano, there is no physical/geographical reference. You cannot FEEL your position, you can only SEE it. This makes playing polyphony almost impossible without looking at your hands.

The theremin cannot be felt or seen, but it is monophonic so you only have one note to deal with (and I hardly need to tell you how difficult THAT is).

You can quantize your Continuum to minimize the possibility of off-pitch playing but this defeats the very thing that makes the Continuum what it is (as quantizing does with the theremin). Control is no longer continuous.

Fiber optic technology was scavenged from the UFO that crashed in Roswell, NM, in 1947 and backwards engineered by Pentagon scientists. What you need to do is get ET to lay your damn transoceanic fiber optic cable. They're the ones who know how to do it! ☺

Good idea. And I bet ET can play the continuum without looking.

I never turn on quantization, though when connecting to Omnisphere, Ed makes the point that you may want to do that for mass sampled sounds like a string section or chorus where there are already a mix of voices recorded that are microtonally interacting. 

I can't believe I was thinking of getting a continuum without them. But when you are playing 4-6 part polyphony, your hands can somewhat obscure the guides. You do still have to feel your way around a bit. It's quite a wonderful invention. I rate it way above fiber optic cable.

But I venture after playing it enough you just might be able to play by feel alone.

I'll review what I did to connect my theremin to the AES input port of the continuum (though this will work for any analog input signal in a similar way).

Using the Berhringer UltraMatch Pro (other converters will be similar):

1. First turned down all volume in the Continuum's Editor (Gain) which should be connected and showing the blue connection light. So assume continuum is connected by MIDI to the PC as usual. AES Input volume should be set at the minimum by default here.

2. Also turn down the (mixer) inputs connected to the continuum (this is all just in case we type something by mistake in the matrix that causes a sonic explosion - very easy to do if you type in the wrong cell). The continuum defaults to overload protection mode itself so I’m not worrying about that.

3. Now I need to get my 1/4" mono analog theremin input into the Behringer converter. I use a DI box that I can go 1/4 in to XLR out - needed to connect to the Behringer L analog in. Of course I could have converted to a stereo signal and fed both L and R into the Behringer’s analog inputs as well. This will slightly affect how I program the EaganMatrix. I also connect the Behringer AES output to the continuum’s AES input using a XLR cable that is specifically made for digital signal connections.

4. Now I set the Behringer to convert analog to AES (if you are just using it for AES IN now – note those settings in case this changes operation).

Source: Analog In (there is another source next to that that gets defaulted to s/PDIF that I can’t change)

Mode: A/D D/A

Clock: PLL Clock / Internal

Sample Rate: 96 KHz

Format: AES/EBU

Word Length: 24 its

Dither: Off

5. Now go to Presets in the top Left of the Continuum Editor and select the Empty patch. This will display a blank Matrix. If you only see the top part and not matrix, click the little red diamond on the upper left. If you’ve never programmed the EaganMatrix – you are about to begin!

6. We’re going to test this with the simplest thing we can do – just route the AES input right back to Stereo Out. On the Left of the Matrix you’ll see an AESIn option. If you are just using a mono input, (in this case I connected by theremin to the Left input of the Behringer converter) then route that Left input signal to both the main Left (SL) and main Right (SR) outputs by typing a 1 in the row/column intersection (which routes inputs on the left column to outputs on the top row). If you make a mistake just put your cursor on the value and press delete. The 1 means a unity signal – in this case we are routing the full signal to output.

However, if you input a stereo signal to the Behringer converter (went into left and right analog inputs). You will put a 1 in the intersection of the AESIn and SR column and keep the 1 in the intersection of the AES In L and SL column. This sends AES Right in to Right out, etc.

You can now turn up the gain on the Behringer input (physical known on front of the converter) say half way and then slowly turn up the AesIn dial (circled in the drawing). If the main Gain editor dial is down you should turn that up to. You should be hearing the AES Input from the Behringer now when you play.

Not very exciting, as you are just basically converting the input to digital and the continuum is converting it back – unless you are feeding the AES output of the continuum to a converter – maybe back to the Behringer’s AES input and that is converting to analog.

But there is one big plus about doing this. The SL/SR output is at the end of the continuum’s reverb chain. So you can add reverb and delay this way without using a pedal. You can play around with that by adjusting the Recirc editor dial to the right of the AESIn gain editor dial (this might say Reverb or with older firmware version). You can then go to the bottom right Reverb/Recirc dials and adjust delay time and other parameters and also change reverb/delay type by clicking on the reverb type name in red.

Note that depending on the analog signal, you may want to feed an amplified signal into the Behringer. My Etherwave is so hot that’s not needed. I bet if I used the Burns I would get better results by amplifying it first.

Finally if you want to save this setting, go to presets and use the Save option to save the current matrix patch to your disk that you can then read in / Open instead of one of the default editor preset patches in the long list.

If you are successful - I can start posting some patches you can download and load in.

Here's another thing to try. This just routes AES Mono Input to the Voice Delay module and turns it (1 in Direct row) on and then the outputs of the voice delay (from bank C) are then routed to SL and SR. You can click the red number under voice delay to set different delay times. If you are inputting stereo AES, you will set the AES R row to a 1 under the Voice Delay input column as well as the L.

This is about as easy as EaganMatrix programming gets. I'll post some more interesting things to do with the theremin soon - maybe with the .mid files you can just load as Presets into the editor.