Playing the D-Lev Digital Theremin

With the completion of my second prototype D-Lev I was able to move it out of the shop and into my (sound insulated) practice room, which is really just a DIY home theater room that's been taken over by junk.  I tend to practice early in the morning (really early) before anyone can hear me, because even though I love playing the theremin, I'm not all that fond of listening to one and I'm guessing that others with any sense of musical pitch may feel the same way.

So despite having this unit working in various states of assembly in the shop for most of the last year, this is really the first chance I've had to play it hard and really give it a road test.  Things like pitch preview that I never use when just playing for fun are getting a workout, and overall the theremin is performing admirably.  There are a few issues that have become apparent, but nothing more serious than I have seen on the Etherwave and Subscope, and the only real difference here is that something can generally be done to fix problems more easily than on analog theremins.  In particular I am having a long-standing issue with setting up the volume contour, and this can be heard in some of the upcoming recordings.  Either I need to get educated on the proper setup, or there is a genuine need to make the volume contour a little more foolproof.

The more distinct voices of some of the presets (in particular the strings) really make pitch errors stand out compared to more mellow timbres, and I have to work hard to hit the pitches, and they are missed all too often.  My timing is the first casualty when trying to land on pitches, which will be heard in all of my upcoming demos, but to me, pitch comes first, at least when playing solo.  Playing without accompaniment is a little like the dream where you are performing or speaking before a large audience and you find that you forgot to put any clothes on - there's nothing to hide the ugly parts.

But these videos (with many more coming) are not about my playing; they are intended to demonstrate a few of the nearly infinite variety the D-Lev sounds. The limited number of presets that you will hear in these and other upcoming videos have lost some of their refinement in the software changes that have evolved, and are all due for some rework. Also my recording setup is not very well refined yet, and in a few areas I overshot the threshold for my camera's audio. I was also unable to get my Lexicon Pantheon reverb plugin working in Vegas, so I used the Sonitus reverb that seems to have a characteristic speaker-in-a-barrel sound. I hate recording myself for a number of reasons, which is why I gave up recording guitar after just a few videos.  Part of it is all of the setup required, but most of it is rarely liking what I hear when finished.

No audio processing was used for these videos other than the addition of reverb in the video editing process.  Pitch correction was never used.

With the requisite pre-apologies aside, here are a few recordings appropriate to the holidays that jumped the queue to get posted in time.  I'll probably fix them up later with a better reverb and when I post more demos.

Enjoy!  And best wishes to all on TW this time of year .  2020 can only get better. 

Silent Night (Cello preset)

What Child Is This? / Greensleeves (Female Vocal preset)

O Holy Night (Male Tenor preset)

Amazing Grace (Cello preset)

White Christmas 2 (Male Vocal preset)

It Came Upon a Midnight Clear (Flutey preset)

Off with their heads!

Fantastic playing / videoing Roger!  Your pitch hand technique seems rather like Katica's?  Making videos is really time consuming, particularly so if producing anything of quality as you clearly have here, so I can appreciate all the effort you've put into this.  I know the tuner LEDs don't generally expose very well in videos, but you might try a bit of quantization on the display to help visually resolve the note being played (I use Qant = 1).

We'll have to put our heads together on the volume response, I'm intrigued by how you might have it set.  It's too bad our D-Levs aren't identical, as I could then just dial in your settings and see what's up.  When you switch to the latest SW could you capture preset -128 (the system preset) and send it to me?  Is there any way to quantify the Subscope volume response?  I assume the volume response you use on the Subscope doesn't involve anything dynamic (that getting closer to the Subscope response would only involve scaling the distance, and not velocity)?

Setting up the axes is probably the most difficult adjustment on the D-Lev.  I've endeavored to keep the setting count to an absolute minimum, but they necessarily interact to some degree, and the volume knee introduces even more complexity.  Luckily though, after the initial adjustment, I don't find myself messing with them at all (other than the pitch field null Pcal).  Documentation here would probably help you!

Eric,

It seems that I am either fighting with a response that doesn't have a clean cutoff as the hand dips into the antenna, or a volume that increases too rapidly near the loudest end. We talked about this in the past, and to a certain extent I've found settings that can fix either the low or high volume characteristics, but not both at the same time.

I think this all came home to me again when I saw the audio clips of "O Holy Night".  The circled areas show how the volume is increasing as it nears full amplitude, and I can't think of any situation, except perhaps for a sound effect, where this should be allowed to happen.

The Subscope to me has the most ideal volume response, and that is where you have a full cutoff near the antenna but away from the antenna the volume approaches maximum gracefully, sort of like the first curve in the image below.  At least for the song above I feel like I was struggling with a contour like the third one down or the bottom one, where the behavior on the louder side was too abrupt.

I could try to measure the Subscope to see exactly what it looks like on the low end, but I know that the high end looks like the first or fourth curves.

I agree that we need to have similar setups soon to help with the communication.  I'll be working on a box next week that may become yours.

PS:

I have always had the tuner quantization set to 0 so that it shows a nicely graded transition between notes, assuming a quiet environment.  But it doesn't much matter now because I've developed a habit of playing with my eyes closed.  It seems to help.

Roger, I'm watching your rendition of "Silent Night" with particular attention to your volume hand.  I don't see anything obviously out of the ordinary going on (which of course means nothing in terms of how it feels to play!), but if the volume past the knee feels too "jumpy" and hard to control then it might benefit from some adjustment.

00. Write down your volume axis settings so you can easily revert back to them if necessary.

0. Use preset 0 (which includes the system preset) because we'll be writing to it.  If your antennae are physically very similar (yours aren't!) and you've got the pitch axis already properly set up, set Vlin on the V_AXIS page to the same value as Plin on the P_AXIS page.  Otherwise, set Vlin=0.  Linearity adjustment is a subtle thing for the pitch side, and sooo subtle for the volume side you probably won't notice any difference between settings (other than sensitivity differences, which unfortunately is a side effect of adjusting the linearity).  Set Vrev=1 if you like closer=louder (non-traditional) volume axis response (you don't!).

1. Do an acal.  I do this with my body in playing position but my hands away from the antennas.  It don't believe it really matters where your body, hands, etc. are, as long as you are consistent with where they are positioned during the acal (and have Vcal and Pcal set accordingly to compensate).

2. Turn off the knee (knee=0).

3. Try these V_AXIS settings to start with: Ofs-=20, Sens=65, Ofs+=49.

4. Adjust Vcal so that the far field sensitivity is the same as the near field sensitivity (i.e. linearize the far field).  You can use the same open/closed hand gesture here that you use to calibrate the pitch axis.  With the settings in step 3 I get ~0.5 led change per open/closed hand, and a total field size (no LEDs => all LEDs) of ~12".

5. Adjust Sens to get the total field size around 12".

6. Adjust Ofs+ (and/or Ofs- if necessary) to move the field around.  Use this to set where the final full volume point should be in space.

After doing the above I get no LEDs = ~2", all LEDs = 15", the field seems subjectively linear, and it responds more or less the same to an open/closed hand anywhere in the field.

7. Save to preset 0.

8. Now you can set up the knee:
- vloc sets the location of the knee ending point, which is the point past which knee expansion isn't happening (the zone where you would normally "play" an audible note).  vloc=0 sets the end point to -48dB, vloc=60 sets the end point to 0dB, and this is linear in dB so a 16 count change of vloc results in an ending point change of 12dB.  For example: vloc=32 gives you a playing range past the knee of 24dB before hitting full volume.  vloc=48 gives you a smaller playing range of 12dB before hitting full volume.  Adjusting vloc obviously also moves the "note starting (in earnest)" position around in the volume field.  I tend to use a values anywhere from vloc=32 to vloc=48 for normal type sounds.
- knee sets the strength of the knee gain, which is downward expansion of lower volumes.  This is "expansion" in an audio expander sense, where lower volumes below the knee go to zero faster (faster in a slope sense, not a dynamic sense), with the knee knob setting the increased rate (slope) at which they go to zero.  I usually use values between knee=4 and knee=16 for normal type sounds.

Traditional volume response (farther=louder) is just trickier in general to set up for any Theremin.  This is due to the interaction of one's body, drift, etc. rather strongly influencing the size and linearity of the far field (as is their wont).  These confounding influences are "expanded down" and out of the way with non-traditional reverse velocity sensing by the knee.  My suspicion as to what's going on with your "jumpy volume" is a non-linear far-field, which must be tended to more fastidiously with the traditional volume axis response.

Of course the above steps are just a suggested starting point.  Once you have it all working more or less linearly and OK you can tweak the playing field size and location via Sens and Ofs+, respectively.  And if you want the far field to have a droopier response (poop out as you get farther from the antenna, the top contour in your illustration above), you can certainly accomplish this by increasing Vcal.

Also, moving the whole audible thing around via the various volume controls scattered throughout the D-Lev and one's amplification can make a big difference on how it "feels" as well.  Getting the knee & max volume locations lined up with the right audible dB points seems fairly critical.

I think the theremin playing is good but the theremin tone sounds too unnatural, without passion or emotion. This will limit its professional interest.

Now that one song is sentimental. It was Spring of 1975 when Bo was fiddling with his little AM Radio looking for something to take our minds off of being waist deep in muck with leaches sucking the life out of our legs. The forest canopy was so thick you could not tell if it was day or night. Totally tired, but we could not sit down because of the muck, it was safer out there. One or two weak AM radio stations, one was stronger than the other and surprisingly both were playing “I’m dreaming of a White Christmas” in the jungle no less. It gives me chills to this day think about it, we were Ordered to get home NOW anyway we could.

Edit: Bo was Black and the kids had never seen this before, they would dare each other to run up and touch him, they thought he had been burnt.

Happy Holidays

Eric,

I printed out your instructions so that I could refer to them at the theremin, and with some deviations from your numbers I was able to get to the subjective linearity from 2" to 15". Next I started working only on the KNEE and VLOC settings, and I think I was starting to remember how those parameters work.  I was still having a hard time finding a satisfactory combination of sharp response near cutoff with very soft response near full volume, at least the way I have been playing so far.  

I'll admit that from there I pretty much went into a full Monte Carlo knob-spinning approach until I realized that I was asking a lot for any theremin to give me the volume non-linearity that I want at the relatively long distance that my hand is located from the loop on this particular case design and with the way I've been playing recently.  I think that the ability to obtain pitch field linearity so far away from the theremin is perhaps fooling me into thinking that I should be able to set up a non-linear volume field as far away from the volume loop as I like, and that may not be the case.  On the Subscope, which I am holding up as my gold-standard for volume control,  I am playing much closer to both of the antennas, simply because the smaller pitch field requires it. But the huge pitch field of my D-Lev (which is by choice) is setting me back away from the theremin so the comfortable tendency is to play farther from the volume antenna as well.  When I moved closer to the volume loop and then experimented with all of the parameters I was able to make some good progress.

I know we are getting into personalization stuff here, and my left hand technique is probably suited more toward a folded plate element than a loop, but I should maybe explain a little more about what you are seeing in the videos. For normal playing I like to have my left wrist almost stationary, with the fingers going from fully extended in the open position to a relaxed closed position with the first finger and second against the thumb, as if holding a sheet of paper to read.  Over that limited physical range I would like the volume to go from essentially full cutoff, well below where the first led goes dark, to a maximum of about 3+ leds lit.  Beyond that, a wrist rotation with further drawing of the fingers into the palm, resulting in at least 4-6" additional movement away from the antenna, adds the dynamic peak to nearly fully illuminate the fourth led.  This is the region where the dV/dl (V volume, l distance) needs to approach zero smoothly. 

I started thinking that moving my rest hand position closer to the antenna would let the ~ 1/distance capacitance relationship help with the non-linear volume response that I was looking for.  Doing this, I found some settings that are getting very close to where I want to be:

I started with these settings:

Approximate hand movements 3-10" from antenna edge
vcal 4
knee 3
vloc 18
lin 0
ofs- 35
sens 100
ofs+ 49

to a better feel:

Movements 1.5 to 6" from antenna edge
vcal 22
knee 3
vloc 37
lin 0
ofs- 33
sens 70
ofs+ 61

Don't worry if some of these parameter changes don't make sense to you just yet.  This is far from a controlled experiment, but I'm seeing a glimmer of hope that a combination of settings exists, at least if I alter my hand position.  If the sharp non-linearity that I want can't be achieved starting at some arbitrary distance from the antenna (and I don't see how that would be practical), I can certainly adapt to a new hand position one way or another.  And this might further bolster the idea of that cabinet design with the closer volume element.

Looking at this from a consumer standpoint I started to think that maybe all of these interactive parameters should ideally be pulled into some canned volume contour presets, something like velocity curves for a keyboard.  Could you for example have a setup screen with easily understandable user settings such as full cutoff distance, full volume distance, linearity curve shape. etc, and manage the translation to the more obscure parameters behind the curtain (while still allowing access to these for advanced tweaks)?
I realize that some of these cross the line between global and preset parameters, but maybe that is okay?

Anyway, thanks for writing the detailed instructions... which further points to the need for that PDF guide so you don't have to keep repeating this stuff over and over, and over.  If I can't figure it out after all of this, think of the hand holding that 2, or 10, or millions and millions of D-Lev customers are going to need.  Oh wait, I forgot that it has been declared dead on arrival.  Never mind .

"I'll admit that from there I pretty much went into a full Monte Carlo knob-spinning approach..."  - pitts8rh

Ha ha!  I know from whence you cometh...

"...until I realized that I was asking a lot for any theremin to give me the volume non-linearity that I want at the relatively long distance that my hand is located from the loop on this particular case design and with the way I've been playing recently. 

I think that the ability to obtain pitch field linearity so far away from the theremin is perhaps fooling me into thinking that I should be able to set up a non-linear volume field as far away from the volume loop as I like, and that may not be the case."

Ah, yes, that's an excellent point: the size of the field probably influences any non-linear gradient that you might be trying to intentionally impose on it.

I've done everything I can to make both fields linear, but I have very little idea of how most analog Theremins behave on the volume side.  It took me a while to understand that a large linear volume field isn't all that useful, and from there I developed the two step linear knee, which solved things for me at least.  But others may be looking for a gradient (rather than a step gain change of the knee) which is how most analog Theremins probably behave.  Before the introduction of the knee, I was "bending the field" via rather extreme settings of the knobs, but the result wouldn't behave when reversing the field sense without a ton of tweaking.  With the knee I can reverse the field sense, do an acal, and a bit of minor tweaking later it pretty much works fine.

So, yes, maybe (as you are obviously doing!) try a smaller field (lower setting of Sens) closer to the antenna (Offs-, Offs+) and warp it via Vcal (like you would with an analog Theremin pitch field via the null).  This way you might even be able to get what you want with the knee turned off (knee=0), though if sharpness (abrupt change in response) is what you're after then some application of the knee will likely help.  I like the knee as it can produce percussive envelopes when turned up and combined with velocity, with the "hit point" way out in space (I don't like playing too near either antenna, particularly when banging a virtual drum).

"Looking at this from a consumer standpoint I started to think that maybe all of these interactive parameters should ideally be pulled into some canned volume contour presets, something like velocity curves for a keyboard.  Could you for example have a setup screen with easily understandable user settings such as full cutoff distance, full volume distance, linearity curve shape. etc, and manage the translation to the more obscure parameters behind the curtain (while still allowing access to these for advanced tweaks)?

I realize that some of these cross the line between global and preset parameters, but maybe that is okay?"

Certainly something to think about!  I imagine once the physical design and choice of inductors / inductance is more or less a constant, then system level presets become more broadly applicable / useful / doable.  I don't like having Ofs- and Ofs+ doing essentially the same thing from the user standpoint, and endeavored to reduce them to one knob, but the behind the scenes math requires both to accommodate all situations.  Ofs- could probably go to the SYSTEM page to effectively die, and perhaps it should.  The knee parameters strongly affect playing but not the basic field itself per se, so they are per patch.  I figure most people will have one volume field they like and feel comfortable with, and will then tweak the knee to get more or less abruptness.

But beyond that, I'm not sure how one might combine axis parameters to make them fewer / more basic from a user perspective.  There's a lot of non-linear interaction between them, and the axis "operating points" are rather vague things in the first place.  Lord knows calibrating an EWS is a can of worms even if you know exactly what you're trying to do as an EE, musician, Thereminist, etc.  The Theremini tried to nail this stuff down via their calibration routine, but ended up interposing a rigid algorithm and removing too much control IMO. I wake up every day and thank the universe that linearity boiled down to one simple, rather subtle knob that never seems to require touch-up (thank you, universe!).

"...which further points to the need for that PDF guide..."

Yes, I feel pretty bad that you don't have better documentation and end up having to go Monte Carlo on its ass.  The Hive doc is years old now and so also crying out for attention.  Technical docs are an activity I really enjoy doing, so it's not procrastination holding me back.

Very impressive, i saw a big musical potential on this Theremin. In my modest opinion the left hand is too far away from the loop antenna. Please share the next results!

"In my modest opinion the left hand is too far away from the loop antenna." -robonil

I absolutely agree.  I was working my left hand much more than I would like.  If you read the last few posts above you will see that this has been a problem with my unit for some time. Most of the problem turned out to be in the settings, although when I play off the edge of the loop like I did here it is inherently less sensitive.

There are something like seven parameters dedicated to the volume contours alone, but after making a few changes the volume range is closer to the loop and mostly within the range of finger jumps instead of larger hand movements. The D-Lev is can be set up to work with very small or large hand movements for the full dynamic range, but the parameters that affect the volume curves require some experimentation. Fortunately now it's set up to be very similar to my Subscope, which I've always thought was just about perfect for me.

The next batch of videos that will be coming out next week were already finished before I changed the settings though, so I will still be waving my left hand like a fool .

This is the roll-over count so it will bury itself

You should stop what you are posting as the professionals are not being kind but honest. Work on your sound. For me I think you both are unable to hear accurately, this why you have so much emphasis on artificial pitch control. This is what I have mentioned to dew over the years, what does it sound like. Get the opinion of someone you trust that has some theremin listening experience. Why call it a Cello sound, when it is so far off the mark it makes someone scratch their head.

Remember the first time you put on corrective eye glasses and the world became brighter, so it could be with your sound. At concerts did you stand too close to the giant speakers and went home deaf? Actually this is not funny.

Christopher