Searching the Evolver's Digital Wave Shapes

[Sacred Synthesis]

One of the challenges of the Poly Evolver Keyboard is that you have this immense musical monster whose potential is only too easy to neglect or overlook.  This is especially true of the 96 digital wave shapes.  How on earth do you explore the whole lot of them?  Randomly, by chance or luck?  But this is sure to miss much of their potential. Because I'm very much rooted in the fundamentals of musical synthesis - rather than such things as "Hack," Distortion," and so on - the oscillators are one place I spend much of my Evolver designing time.  So, I've devised a method of exploring those oscillators that works well for me.

First, I decide what category of sound I'm interested in designing.  Will it be an ethereal pad type (the PEK is superb at these), percussive piano/harpsichord/bell type, sustained organ type, instrument imitation type (such as flute or oboe), or some other type?  Because this approach is so time-consuming, I generally can do only one category per day or night.  Once I've decided on a sound category, I then set up the envelopes and filter to match it.  I really like the 2-pole filter setting for choir, strings, and all sorts of pads, so that has to be considered as well, as does the setting of Keyboard Amount, which is essential in choir type sounds.  For the sake of making the experimentation thorough, I also program the modulation wheel to open the filter (between 20-40), since the filter has some remarkable effects on the digital wave shapes that I don't want to miss.  Next is the modulation - usually a vibrato depth of 2-3 and a rate of 65-72.  If the filter will be modulated, then I usually set both the LFO's depth and rate at 6-12.  If I'm exploring melodic type monophonic sounds, then I usually delay the vibrato with the third envelope.  Finally, presuming that the digital oscillators will need a little reinforcing, I add two analog triangle wave forms at about half the volume of the digital wave shapes.  This adds a little warmth and depth without obscuring the unique characteristics of the digital oscillators.  After adding a moderate amount of reverb, I then very slowly dial through each of the 96 digital wave shapes, taking the time to play each one for perhaps several minutes in order to find its strengths and weaknesses.  This has to be done with great care and thoroughness, and it can go on easily for two hours.  Some patches will be absolutely gorgeous, but only in the narrowest range - perhaps within only two-octaves of the keyboard or with one very precise filter setting from the modulation wheel.  The synthesist's great challenge is to find this precise range or place.

This method works exceptionally well for me.  It's quite time-consuming, but the results are worth it.  Best of all, it prevents me from haphazardly passing over the many musical gems hidden in that big blue blinking monster.  I've discovered many sounds that a random method certainly would have missed.  It's been especially useful lately in designing ethereal pads.  As a result, I've now got more patches stored in the PEK's memory than I have time to record.

[chysn]

Great topic.

For me, sound design approaches vary wildly based on the instrument. For example, with the Little Phatty, I work sort of deliberately. It's beautiful in its minimalism, and I can think it and make it. It's got a huge forgiving sweet spot.

The Evolver, for me, is all about turning knobs to see what happens. But it's not "random." My way of working with the Evolver is more improvisational than random. The emergence of the sound is the music-making experience itself, rather than preparation for something more formal. Recording it would either embalm it, or else pop it like a bubble. So I seldom record, and the sound goes out into the universe, like a meditation; and all I can say is that it once existed before I start something new.

[Sacred Synthesis]

I can fully appreciate the merits of spontaneity and experimentation.  They seem to be at the heart of your approach which can produce unexpected moments of beauty.  I do use that approach at times, too.  But my usual method is meant to tame the beast and master it, so as to draw pure music from it.

[chysn]

I did sort of miss the point of your original post, which was about using the digital waves. I audition them a lot, because they can be shaped in so many ways; things like filters and FM can completely change the character of a digital wave.

I tend to sequence digital waveforms quite often, as it's a great source of discovery, so it's not uncommon for me to choose six or eight digital waves for a single sound. In these cases, I'm auditioning waves for how they accent specific beats.

I also use custom digital waveforms liberally. I have single-cycle snippets of sounds that have personal meaning, like my own acoustic piano, or my sons playing their instruments (trumpet, and clarinet).

[dslsynth]

I've discovered many sounds that a random method certainly would have missed.

Well careful sound design like you do it and random methods are two different methods to approach sound design on a synthesizer. The papers I have read about random methods suggest combination with manual editing. I would expect a useful implementation of random method to support and integrate manual methods.

Some search by knowledge. Some search by random. Smart people use the methods that works best for them.

[Sacred Synthesis]

My method is meant primarily to be thorough, to avoid missing a tone that's hiding in there, that I would definitely want to find and use.

[dslsynth]

Well that is the core of sound design so you are on the right track! In fact I learn a lot from reading your sound design posts so please keep making them. What random methods can do is to search for new unexpected sounds. Combination of both methods can give interesting results which is my main point in my responses to your post.

[Sacred Synthesis]

Absolutely.  I agree with both you and Chysn on the advantages of using both approaches.  And up until a certain time, I was using only the random approach.  The problem was that, of the Poly Evolver's 96 unique digital waveshapes, I was using only so many and passing over many others.  I needed a method that wouldn't pass over a single one; I wanted to squeeze every drop out of the instrument.  By choosing one sound type per session, setting up the envelopes, filters, and modulation to serve that sound type, and then painstakingly working my way through each and every waveshape, I've been able to maximize the usefulness of the Poly Evolver.

[chysn]

What random methods can do is to search for new unexpected sounds. Combination of both methods can give interesting results which is my main point in my responses to your post.

Absolutely.  I agree with both you and Chysn on the advantages of using both approaches.

I think my position has been misinterpreted, and I'm sure I'm to blame. To be clear, I don't see much value in random sound design techniques. I've read John Cage. I've tried randomness. I didn't like the results at all. It seems like you'd theoretically find something new and interesting, but in practice you don't.

Wendy Carlos said something along the lines of, "What is full of redundancy or formula is predictably boring. What is free of all structure or discipline is randomly boring. In between lies art." I don't claim to produce any art, but I agree with that.

[Sacred Synthesis]

I think we all from time to time sit down at our instruments, start randomly fiddling around, and unexpectedly come up with something we very much like.  Of course I have no problem with this.  That's what I thought you were saying, Chysn.

[chysn]

I think we all from time to time sit down at our instruments, start randomly fiddling around, and unexpectedly come up with something we very much like.  Of course I have no problem with this.  That's what I thought you were saying, Chysn.

I suppose that my internal definition of "random" might be a bit technical, is the communication gap here. There are composition techniques that, instead of the composer choosing sounds (note, timbre, rhythm, etc.), he or she designs a way of determining sounds by chance operations, decoupling judgment from the process.

Hidden deep within this is the idea that humans are incapable of randomness without the aid of some process. (I'm sure that some drugs are able to decouple judgment from action, but I'm not interested in that.) There's very little "randomness" to improvisation.

And that's what it is. The matrix-layout of a Desktop Evolver is an excellent improvisational tool. The eight knobs in a row make the relationship of beats very clear. The hands don't need to travel far, so once you learn where the parameters lie, everything is right at hand. A Desktop Evolver is a self-contained musical instrument, with no need for a keyboard. The tools for exploration being right there make it brilliant.

(Ultimately, the weakness of the Mopho line was that the sequencer was well out of reach in every single model. The keyboard models had a fiddly three-knob system where by you chose the step, the destination, and the value. The Tetra/MoBrick's whole interface was a disaster. My complaint with the X4 was that there was plenty of room on the right side for sixteen sequencer value encoders. I think, after the Evolver, DSI lost sight of what sort of instrument a desktop synth can be. A Mopho with a matrix-layout interface like the Evolver's would have been amazing.)

[dslsynth]

To be clear, I don't see much value in random sound design techniques. I've read John Cage. I've tried randomness. I didn't like the results at all. It seems like you'd theoretically find something new and interesting, but in practice you don't.

The fun thing is that we are talking of two to three different kinds of random sound design in the discussion above. The one I talk about is a mixture of random parameter values, random mutations and interpolation in parameter space between such sounds. Some people prefer to start with random sounds whereas other want to start from specific good sounds they want to explore. It can indeed be useful.

Of cause its a personal preference what suits ones style of work. Nothing beats careful exploration of sound spaces backed by intricate knowledge of the instrument as original poster describes it. On the other hand random methods can supplement the search for new sounds.

[Sacred Synthesis]

As a footnote to this very interesting discussion (and it's good to be having these again), in my original post I was making a suggestion to anyone else who felt a little overwhelmed at the sonic potential of the Evolver.  How do you explore the instrument without passing over most of it?  My method obviously concerned only one parameter - the Shape/PW - under which lies nearly a hundred unique digital wave shapes.  In the art of designing sounds, my method is one approach that will well serve anyone pursuing the sorts of sounds I use. 

In my opinion, the Poly Evolver Keyboard is an extremely under-rated musical instrument, and I suppose this is due to its limited voice count.  Still, it is a masterful brush, palette, and canvas set for creating tonal paintings, but one that is used too exclusively in the sound/noise domain.  I find precious few recordings on YouTube or Soundcloud that apply its potential to traditional types of music, and I suspect this is due in part to its rather mysterious and nearly undocumented digital oscillators.  This is not the case with the much simpler Prophet '08, for which countless musical recordings can be found. 

I've found that, without a proper method, it's only too easy while tweaking on the PEK's control panel to drift away into sonic lah-lah land, without achieving much other than a pleasant daydream.  I've wasted too many late nights doing just this, with nothing to show for it the next day.  I realize this sort of daydream is just what some folks want and enjoy, but I'm usually after another objective.  I've suggested a method for reaching a specific end - the discovery of a maximum number of musically useful sounds.  It works well for me; who knows if it might work well for some one else?  But if this end is not of interest to you, then perhaps neither will be this method. 

[Razmo]

I've not used the waveshapes much either... The main reason is that the two user waveshapes you could potentialy choose to use in a preset, is NOT saved with the patch itself... if that had been the case, I can almost guarantie you, that you would have a bunch of user waveshapes used in publicly available Evolver patches.... the fact that it's circumstancial to get them into the Evolver, and the extra work you'd have when you wanted to compile your own banks of sounds is what keeps people from using them... and that is a shame... had there been just two slots for user waveshapes, and these was part of the patch, then it would have been much more flexible.

So how do you use the waveshapes? ... well... one major usage is the sequencers... that is still one of Evolvers main unique things, that no other DSI synth can touch... not even the P12. If you like to use rhythmic timbral changes in your work, Evolver is it... it's something I'm going to explore A LOT more in the future, now that I decided to keep my PER anyways... Especialy for long evolving sounds, that just keep changing and changing rhythmicaly, the waveshape sequencing could be a huge plus, especialy if you start to use Combos, where the sequences can run at different divisions and lengths... combine this with very slow LFO's and long envelopes... it is potential drone heaven!

Other uses of the waveshapes are probably not so obvious... but if you program percussion sounds, and you use the waveshapes to create very short clicky transients, then the different waveshapes become a treasure throve of different click-types... another not so obvious use, is when you engange stuff like Ringmodulation and Frequency Modulation ... here the different waveshapes becomes everything in creating weird new textural sounds... even when the oscillators are running extremely low in pitch, they can be utilized as LFO's for pitch and volume using RM and FM, and then the weird shapes can generate some rather complex modulations.

Also... I've been messing with speech synthesis using the wavesequencing earlier, which it could also do... unfortunately, only 5 of the user waveshapes can be used in the wavesequencing, which is a real shame.

Another use is if you want to try and replicate some Waldorf Wavetable synth patches... this of course is not very doable, but you can in some cases do it, if there is not much wavetable modulation going on... I had a sound on my MW that I liked very much, which I duplicated on Evolver... it did not modulate much, but it used a specific waveshape that was essential, so I sampled that waveform, dumped it into the Evolver, and that helped me imensely recreating the sound on Evolver.

Another use would be to recreate patches from the earlier SCI instrument from which they were taken, which is probably the main goal Dave had with it I guess, since it's the exact same waveforms.