Today I decided to go back to a technique I’ve rarely used, and on a much grander scale. I don’t use noise very often, and when I do it tends to be for the obvious use cases. Hit hats, wind and ocean sounds, sprays, etc. I seldomly use it for modulation, and only once have I used noise of any flavor to amplitude modulate an oscillators wave. Today I would do it again, times eight.
I conceived of using noise to modulate all eight harmonics of the Verbos Harmonic Oscillator this morning as my wife was talking to me. I even popped up a bit at the idea, and she took notice.
Wife: “What?”
Me: “Nothing. Just had a thought occur to me. Not even sure if it’s worth a shit.”
I spent the better part of the morning and early afternoon thinking about how I wanted to do this patch. I knew that just noise into each harmonic’s VCA wasn’t it. Then it occurred to me: Chaos! As soon as this though hit my brain I knew what to do, and immediately went to the synth to start patching.
I ran blue noise from Sapel to input 1 of the Intellijel Amps. Amps is a special sort of VCA. Everything cascades. All inputs cascade, as do CV inputs, and there are mixing outputs as well. It’s incredibly flexible. I have four of them chained together to be an eight channel “super VCA/submixer” and it’s been a great choice. Since each input cascades, I only needed one noise input to run this entire section of the patch. Every other channel received that same blue noise input as well. Into each channel’s CV input I patched one of the eight outputs from Nonlinearcircuits The Hypster to chaotically modulate the noise levels of all eight channels independently. Once that was patched, I ran each Amps output to its own Harmonic Oscillator VCA input at random. The only part of this patch that was planned were the first and fifth harmonics, which received their noise modulation from the U and -U outputs on The Hypster as they’re the outputs with the highest amplitude. Each harmonic was slowly brought in by slowly adjusting each CV attenuator individually at random until they were all playing. The nature of chaos means that cycles, even if semi-regular at times, don’t repeat exactly the same, and the harmonics never played the same twice, which kept movement interesting. There were often pauses or redirections in motion for each harmonic. Wonderful.
The mixed HO output was patched to the Multi-Delay Processor. I’ve been taken in by the earthy sound of the Harmonic Oscillator. Each harmonic sine wave has a little hair on it once you give them a little push. The drive in the MultixDelay Processor, both on the input and on each tap output, accentuates that hair in all the right ways. This Verbos ecosystem is warm and inviting, but it can also roar. Taps four and eight were patched to the Verbos Scan & Pan, hard panned left and right, and the output of the MDP, which only had the dry signal, was patched to be in the middle of the mix. This mix created a strong signal with some subtle stereo movement which ended up being fantastic. This stereo signal was then patched to the stereo matrix mixer to be spread around to different effects.
The Rossum Panharmonium fed the Holocene Electronics Non-Linear Memory Machine, which was set with a fairly slow delay and full clockwise smearing, which really smoothed out the Panharmonium’s output for an accompanying drone that floats along beside the ever moving Harmonic Oscillator. This output then fed the Dradd(s), which did its thing in Grain Mode (although I think I forgot to turn on the modulation to both P1 and P2 on both Dradds 😬 – I’m also not convinced it isn’t lost in the mix).
I’m very pleased with how this patch turned out and was a great success at using this technique which I’ll be sure to use more often.
I like wavetables and I like drones. Jamuary 6 saw both. A modulated quad wavetable LFO modulating the levels of a quad wavetable oscillator. It’s a patch I’ve been wanting to try for a while, and Jamuary is the perfect time to experiment with new techniques.
There are a lot of moving parts in this patch. The base creating the ever-changing drone is four unsynced LFOs from Kermit Mk3 controlling the levels of the four outputs from the E370 in the Intellijel Amps. These four outputs were patched to the ST Modular Sum Mix & Pan and then to the mixer. The pan CV inputs of the latter three channels were modulated by the Addac506 Stochastic Function Generator. The stereo signal from Sum Mix & Pan went to both the output as well as to a second mixer for summing to mono before being routed to the Verbos Multi-Delay Processor. The mono output of the MDP, which only has the dry signal,1 went to the output mixer, while two of the taps were routed to the Verbos Scan & Pan for hard panning left and right. All three of these signals were mixed and sent to the output.
Although I like the drive the MDP created, it largely defeated the panning of the three E370 outputs. I’d have been better served to run the stereo signal from the Sum Mix & Pan to a stereo distortion for some added drive so as not to throw a cover over subtle movement in the stereo space, but I do like the overall result. The MDP is a fantastic source for overdrive, and is a different sort than a full blown distortion like Mimosa. It’s deep and warm, like a fuzzy blanket on a cold winter day.
There’s lots of modulation too, with the bulk of it being supplied by the Nonlinearcircuits The Hypster fed a Let’s Splosh, with its outputs modulating the waves in both the E370, and three of the four channels of Kermit. Let’s Splosh self-modulated both Gain and Damping to keep its outputs in constant flux.
To accompany the drone proper, I sent the panning wavetables from the Sum Mix & Pan and stereo delay taps from the MDP to a pair of unmodulated Dradds. Both are in Grain Mode, and both are time stretching, with the left channel in reverse at about 20% speed and the and the right channel in forward at a slow crawl. It’s a bit difficult to pick out in the mix, but the overall sound is very different without it, and the final result benefits greatly with it.
The final touch to the patch are the drips and crickets. This oscillator and envelope are both from Falistri, sent through the Holocene Non-Linear Memory Machine. Pitch and random triggering was provided by Sapel. The last free channel of the Addac506 controlled both the Freeze and Scanning of the buffer. Its EOF trigger turned Freeze on and off, with the EOR gate output gating the function output in a VCA, which was scanning the buffer. This little sub-patch took a little while to figure out, but the results are rewarding. I’d been wanting to mess with scanning the buffer of the NLMM ever since I heard this patch by Ras Thavas, and today seemed like the day. It was a fun patch that I’ll be sure to explore further in the future.
As a result of acquiring both the Synthesis Technology E370 and the Flame Instruments 4VOX, after also getting the Humble Audio Quad Operator and RYK Modular Algo earlier in the year, I’ve been stringing together a series of chord-based polyphonic patches using various forms of slow modulation to control the volume of each chord tone. From standard LFOs to chaos, and stochastic functions to ocean wave simulations, I’ve tried at least a dozen of this style of patching over the last several months. Some of these have used static chords that don’t really move anywhere. Different notes of a chord come in and out chaotically (in most cases), but the chord itself doesn’t change. Others are based on the harmonic series, where only one pitch change of the master oscillator affects all of the individual harmonics resulting in chord changes. All of those were composed using chaos or random as a pitch source. But, with one exception, it wasn’t until this patch that I used the NOH-Modular Pianist with real intent and composed a chord progression to move the piece along. To set a mood and provide some tension and relief with harmonic motion in addition to volume and timbre changes. And this time I went big with using all eight CV outputs, rather than just four.
The NOH-Modular Pianist is an interesting module. It promises a world of harmonic movement in an environment where using chords isn’t a simple proposition. Polyphony in Eurorack is equipment and labor intensive. Each separate note of a chord requires its own separate oscillator, function generator, and VCA, at minimum. and requires its own discrete signal path. That’s a lot of patching for what is an easy task in a DAW or by using keyboard-based synths. It’s a lot of tuning (and re-tuning); lots of signals to tweak, and lots of modulation to account for. Before the Pianist, ways to get this sort of advanced polyphony was hard to come by. You could use a MIDI > CV converter, which has its own challenges, or else by painstakingly programming a pitch sequencer note by note, which requires a level of music theory knowledge that most don’t possess.1 MIDI > CV converters require careful calibration, and there are few sequencers with more than just four channels. But the Pianist is different.
Rather than programming chords note by note, Pianist uses standard western music shorthand for identifying chords, and the module does the rest. When you program it to play a CM7 chord, for instance, it knows to send out pitch data for C E G and Bb. It’ll even repeat chord notes in a different octave if no color tones are used. You can add two chord extensions beyond the 7th, called Colours in the Pianist, or use chord inversions to designate the third or fifth as the bass note in the chord. If a up to six note chord can be played on a piano, it can be played by the Pianist.
Users can freely enter chords from scratch in Free mode, or, to make the job even easier, set it to Scale mode and choose only from chords within your chosen key. The scale can be set to Major, Minor, or any of the modes2 and Pianist does the rest. So, for example, if a user in Scale mode were choose A Major as the scale, Pianist would present you with only AMaj, Bmin, C#min, DMaj, EMaj, F#min, G#dim, the diatonic chords in A Major, in order to facilitate easier chord progressions for theory novices. As long as your oscillators are tuned, your chords will be in key. Nifty. For those who want to use chords outside of a key, or if your composition isn’t really in a specific key, Free mode allows for creating chords from scratch. Virtually any chord is possible (up to six notes). In both modes, harmonic complexity is simple, with up to two color tones available, and made even simpler in Random Gate mode where each gate received will add random colors automatically, and choose colors that make harmonic sense within that chord. The workflow in creating chord progressions is intuitive. I was quickly making fairly complex progressions with repeats and skipped chords with ease.
Though Pianist is a boon to those of us seeking access to polyphonic 12TET harmonic movement in our Eurorack patches, it does have its weaknesses. Though you can move notes up and down in octaves to create chord depth, it’s done in a haphazard way. Rather than setting each note for the exact voicing you’re looking for, you have to rely on functions Pianist calls Shift and Spread in order to get full, rich chords that don’t clutter a particular part of the audio spectrum, but it’s not exactly clear how that affects the chord as a whole. I can hear changes, but can’t always identify them. Easy variety, however, can be achieved when the Gate mode is set to Spread. No chord will be voiced exactly the same which creates intrigue.
The calibration for the module, at least in Version 1.0, is straight funky. This patch uses eight discrete oscillators. While tuning I sent a C from Pianist to set a baseline. But in order for the oscillators to play the C being sent, they each had to be tuned to G, which I found odd. The newest firmware, 1.2, addresses tuning and scales in a way that version 1.0 does not, which is a great improvement by all accounts, even if I haven’t used it yet to note any changes. Since I’m using Pianist in Free mode in this patch, however, there wasn’t a compelling reason for me to upgrade, though I certainly will now that I’ve finished recording it, even if I have an aversion to the upgrade processes of most digital modules.
The screen has a lot of information, and not a lot of room. However, navigation is still reasonably simple and the information on the screen laid out such that it’s not hard to read. It’s easier to read and use than many far more established modules like the Disting Ex, Kermit Mk3, or uO_C, even if there isn’t a lot of screen real estate. The interface is super easy to navigate using the mini joystick/push button. Version 1.2 is reported to have an even more streamlined navigation and menu system. Though altering global settings like the Scale, Gate or Spread behavior requires some menu diving which is never fun, programming chords decidedly does not. It’s a point and click operation made easy with the joystick, all done on one level. Move the cursor to what you want to change, click, move the joystick to the desired value, and click. Done.
A major issue with version 1.0, which may have been changed, is that it always boots up with the first saved sequence. Unless you save your progression to one of the user slots, you will lose your work if the module power cycles. If you don’t have much in your progression, or it’s a super simple that’s no problem. But if it’s long or has a lot of direction you might be losing a lot. Ask me how I know. 😕
Pianist has its own clock that will change on each beat, along with a clock output to trigger envelopes or some other event as chords change. But it also has a clock input, which will move along the chord sequence with every rising edge like any standard step sequencer. Being that I rarely use a steady clock, I haven’t tried the internal clock, and have instead used clocks created by chaos or some other irregular source. This patch used a fairly complicated sub-patch in order to derive the chord changes. I didn’t want haphazard pitch changes in the midst of notes actively being played, but only when nothing was being heard. Finding an approach for this was time consuming, and although there are probably (certainly?) other methods that would work as well, I settled upon an approach using two comparators, one analog and one digital.
The four waves from Swell Physics first went to the Xaoc Devices Samara II. Samara compares all four inputs, and outputs the Maximum signal (AKA Analog OR). Being that these four waves were controlling the volume of the individual chord tones, it occurred to me that once the Maximum signal went below 0v meant that all four parent signals were below 0v, which meant no volume at all from the chord voice. This is exactly when I want to trigger the next chord in the sequence. I then sent that Maximum signal from Samara II to a digital comparator, the Joranalogue Compare 2, with its compare window set to anything below 0v. So once that Maximum signal went below 0v, it would spit out a gate that would trigger a chord change in Pianist.
The eight chord tones created by the Pianist went to eight different oscillators. The root, third, fifth, and seventh (or fifth if there is no seventh) form the base of the chord and all go to one of the four Flame Instruments 4VOX oscillators, while the color notes and two additional root notes, one that follows chord inversions and one that does not, all go to a self-frequency modulated Frap Tools CUNSA, where each filter is set to self oscillate, and pinged in a Natural Gate.
The Flame 4VOX has been around a long time. My brother, a house sound engineer, producer, and DJ who’s been into Eurorack a long time, lusted for one long before I even knew what Eurorack was. It’s a fully polyphonic, wavetable oscillator beast, split into four sections of up to four oscillators each. Each oscillator can create detuned swarms, chords, or be unison. Each oscillator can be controlled by v/oct CV or midi, and is fully polyphonic with its own output. It really was a very advanced piece of gear for its time. It still is, even if it hasn’t been updated in several years and is showing its age. There are two pots and two CV inputs per oscillator that can control several parameters including scanning the wavetable, detuning, amplification, and more. It has internal VCAs to control volume, but I did not like how they functioned at all, and opted to use external VCAs, which worked to my benefit allowing me to modulate two wavetable parameters rather than the volume and only one parameter. There are also separate FM and reset/sync inputs per oscillator, along with its individual output. Even if CV-able options seem to be limited, virtually every facet of the 4VOX can be addressed via midi, although I haven’t used it with midi at all. It’s a very powerful oscillator bank that can cover lots of ground.
Although I wouldn’t say programming the 4VOX is difficult, it’s not as easy as most more modern interfaces. The screen is bare bones with low resolution and a slim viewing radius. The encoder is a little weird. You have to push it down and turn CCW to move downward in menus, while you simply turn it CW to change parameter values inside the menu. As a unit, it’s impressive. There are lots of options, plenty of stock wavetables to choose from, and it sounds good, but it shows its age. Upgrading firmware is a laborious process with modern computers. Although you can install your own wavetables, the processes to convert them to the right format and get them loaded can be a nightmare, particularly if you’re a Mac user. All of the computer-side software is a decade or more old, and workarounds are sometimes needed. I’m not a “I need to load my own wavetables” kind of guy, and my unit came to me with the latest update, but if I were that guy or my unit hadn’t already had the latest firmware, it would not be an easy task. I’ve had similar problems with older gear before3, and they’re no fun.
The 4VOX forms the base of the chords, brought in and out by the four waves from the Addac508 Swell Physics. The sound is both powerful and delicate, with each quadrant set to four slightly detuned, unison oscillators, each one being slightly modulated by the Nonlinearcircuits Frisson. Although I was pleased with the 4VOX’s performance, the Synthesis Technology E370 is a better overall option. Although the E370 is also based on nearly decade-old technology, it’s still a better user experience. The screen is in color, fully customizable, bigger, and gives more information. The stock wavetables are a gold standard. The software UI is easier to navigate using a more standard encoder. The physical UI is also far better arranged. With the 4VOX, the screen is in the middle of the module, knob locations are not symmetrical, and are more difficult to wiggle once everything is patched up. The E370 has everything laid out very neatly. The screen is on the far left, I/O on the far right, with knobs in the middle, leaving more than enough room to wiggle. It’s really a premium user experience. The only advantages the 4VOX has are its price, size, and complete polyphonic midi capabilities. The 4VOX has always been less expensive than the E370, and that remains true on the secondary market. However, the price differential on the used market is much closer than their respective MSRPs, as the E370 can be purchased for well under 50% of the original retail cost. The price difference on my units, both purchased used within a week of one another, was $100. The size, however, cannot change, and in that regard the 4VOX has the E370 soundly beat. At 29hp the 4VOX is still large (and odd hp 😕), but it’s dwarfed by the massive 54hp E370. It’s the massive size, however, that makes the E370 such a pleasure to wiggle.
Once mixed to mono in the Atomosynth Transmon, the 4VOX chords went through the venerable Industrial Music Electronics Malgorithm MkII, a powerhouse FSU-type module with bit crushing, sample reduction, and various types of waveshaping available to have anything from subtly crunchy through completely mangled audio at the output. Using Malgorithm was an absolute treat. Most of the lo-fi effects I tend towards are of the vintage variety, tape sounds, record pops, etc, vs just slightly old sounding digital artifacting, so it was a different sort of experience. On any other day I likely would have chosen distortion in this role, but the day I started this patch I precipitously chose to go with a different kind of dirt. And it was perfect. I was still able to get some nasty distortion via the “Axis” waveshaper (whatever that does), with the bit crushing and sample reduction playing a slowly increasing role. It’s starts clean, then moves to understated digital artifacting, and finally waves of full blown destruction, ending clean once again. One aspect of Malgorithm I enjoyed was the interaction between input level and the waveshaping. It responds similarly to tube distortion circuits, where the harder you drive the input, the more distortion there will be ranging from just barely there to outright obliteration. Each of these waveshaping circuits has three different levels, red, orange, and green, and all of them have their own character. These waveshapers can even interact with each other for nuking your audio from orbit if that’s what you want. I rode faders on the very awesome Michigan Synth Works XVI to control both the input level as well as the wet/dry mix in order to provide a performative aspect to this patch. Both the bit crushing and Nyquist parameters were modulated by the Addac506 Stochastic Function Generator, with a fairly wide range of both rise and fall times between medium and long. Each of the parameters were set to moderate crunchiness with the knobs, with their modulation moving towards a full-resolution signal. This created an absolutely amazing effect from the sound of dying batteries to the fabric of the universe being unzipped and sewn back together. I would highly recommend Malgorithm to anyone, but you’d have to find one first.
Once through Malgorithm and into the stereo matrix mixer, these now buzzy chords went to the Holocene Electronics Non-Linear Memory Machine, with a very light perfect fifth shimmer in the feedback loop. I initially went with a full octave shimmer, but decided against it as it was too prominent and spiraled too far out of control too quickly. This created a very subtle sheen on the chords that isn’t noticeable much of the time, but is a nice effect nonetheless. Feedback and Spread were both modulated by attenuated versions of the Average output from Swell Physics.4
The color tones of each chord were all sent to the mighty Frap Tools CUNSA, a quad filter extraordinaire, and pinged in a pair of Rabid Elephant Natural Gates. Though I was tempted to use the simple sine waves from each LPF output, I later decided to use the HPF output as a means of each oscillator frequency modulating itself in order to add some harmonics, which worked a treat. In retrospect, I could have simplified the patch significantly had I pinged CUNSA itself instead of running the output to Natural Gate, but I chose the Natural Gate route because Natural Gate.
Using a patch technique I’ve used often, the gates that pinged the Natural Gates were created by running the four waves from Swell Physics into the Nonlinearcircuits Numberwang. But rather than simply choosing four gate outputs, I ran several Stackcables so that each strike input on the Natural Gates were each derived from three Numberwang outputs. This filled in space much better. The notes are still sparse, but they’re triggered at a much better pace using three gates each rather than just one. These notes fill out chords in interesting ways. They’re very short, but combined with delay and reverb, those colors hang around long enough to create intrigue in the overall sound without being intrusive.
These notes were sent to what is becoming one of my favorite delays, the Chase Bliss Audio Reverse Mode C, a re-imagining of one of the modes on the legendary Empress Effects Superdelay. Although it certainly does standard stereo delay stuff, it excels at being a quirky sort of delay, able to output normal delays, reverse delays, and octave up reverse delays, by themselves, or in a mix. Mixing delays creates a beautiful sound space of echoes bouncing all around the stereo field, at different speeds and octaves, which is an incredible aural treat. I haven’t yet learned to properly modulate the Reverse Mode C, but that’s a function of not having a firm grasp on midi. As I figure that out, things ought to get very interesting, with different sorts of delays fading in and out in very creative ways.
The last synthesized voice in this patch is the Good and Evil Dradds as an effects send, sending both the chords and ornamental color notes for some granular action. The Dradd(s) outputs went to separate EF-X2 tape echoes with different settings. Ever since getting a second Dradd, I’ve been infatuated by what I can do with them, and this patch may be the best result yet. Both were set to Tape mode with similar P2, but different P1 knob positions, with the P1 parameter on both being modulated by an attenuated version of the Average output on Swell Physics. The Dradds, in some ways, steal the show. They create all sorts of movement in the stereo field and fill the space between chords and color notes in ways that keep the piece from becoming still. They’re the wake left after a large swell passes by. The bio-luminescence after a crashing wave.
The spoken voice is a set of three samples that were triggered in Koala on the iPad. Triggers emanated from the gate outputs on Swell Physics combined in the new Nonlinearcircuits Gator, sent to the Joranalogue Step 8 and then the Befaco CV Thing and converted to midi notes that were sent to trigger Koala samples on the iPad. It took me a while to figure this one out, though it worked exactly how I envisioned. Gates from Swell Physics were combined in Gator, which triggered Step 8. Each of the first three steps sent its individual gate output to a different CV Thing input. This ensured that the three samples were always triggered in the correct order. The samples themselves were then sent to a new collaborative delay plugin, Moon Echo, by AudioThing and Hainbach.. Moon Echo is a modeled simulation of bouncing sound off the moon, and has a very distinct character. The delay was set to fully wet, and has a delay of about 2.5sec, though that changes depending on the day. The moon is not at a fixed distance from the earth, and the plugin reflects that. By “pinging” the moon upon startup, you will get the current distance to the moon, and a new delay time down to five decimal points (1/100,000 of a second). Fucking cool.
One thing I did differently with this patch paid off high dividends, and will absolutely become a staple in my recordings. I’ve been patching for a few years, but am still an absolute novice at standard studio stuff. Mixing, EQ, compression, and everything else in that sphere evades me. I’ve used some very basic EQ in the past, but really only on the final output, which, as I’ve discovered has several drawbacks. This patch was the first I’ve ever recorded using EQ, the highly regarded Toneboosters TB Equalizer 4, on individual channels as they were being recorded. The chords, ornamentals, and reverb send received EQ that greatly improved the sound quality, even if it could still be better. I did, however, neglect to put EQ on the Dradds, which proved to be a mistake, as there is a very occasional pitch that pierces through in what can’t be far from dog whistle frequencies. It’s not eardrum busting, but I can hear it, and it annoys me. I didn’t catch that behavior when recording, and never EQ’d it out. That said, it was also the first time I’ve recorded a modular patch in separate multi-tracks, including the chords, ornamentals, Dradds, spoken voice, reverb return, and the mixed stereo signal (presented here). I can go back and make changes or additions should that be something I want to do, or send the parts to someone else for mixing and mastering should I ever choose to release it.
Overall I’m very pleased with this patch. It was originally composed in a different key and completely different chord progression, and for a special group of online friends. The chord progression I used in this recording wasn’t composed, as such. At least not by me. I asked ChatGPT for a “sad progression, yet with a sense of hope.”5 I asked for it to be more sad, and it changed key from Amin to Dmin, and ended in a non-diatonic chord (DMaj), which I found a wonderful “choice.” Then, as a means to test the Pianist, I asked for several chord extensions and inversions, and ChatGPT complied, giving us what we have in the recording.
Improvised and recorded in 1 take on iPad in AUM via the Expert Sleepers ES-9.
I studied music performance in college, and have a decent grasp on music theory. The last 30 years, however, have pared that knowledge down to basics. I’m certainly no expert, but I can read chord charts and identify chord notes, even if I have to think for a second. ↩︎
The Humble Audio Quad Operator I purchased did not have the latest firmware update, and the internal VCAs all bled badly. I was unable to install the latest firmware with a modern Mac. I was fortunate to have an older one available to me that I was able to use. ↩︎
There are no fewer than seven modulation points in the patch that are all modulated by an attenuated version of the Average output from Swell Physics. ↩︎
This was literally the first time I’ve ever considered purposefully using AI for anything. ↩︎
I’ve used all kinds of slow modulation sources in my patches. Wave simulators, chaos, and slow, free-running LFOs are all staples in my modular practice, and I’ve used all of them as the engine that drives an entire patch. But one thing I’ve never tried in a lead role in a patch is good, old fashioned random modulation. Sure, I’ve used sample and hold or smooth random generators like the Frap Tools Sapél, Mutable Instruments Marbles, or Make Noise Wogglebug for specific tasks within a patch, but outside of Marbles > Rings-type patches, I’ve never based a patch on staochastic movement before.
Random is, well, random, and it’s hard to be intentional when you can’t really expect what’s going to happen next. In some circumstances, that unpredictability is perfect. In other cases, it just makes a really big mess. There are certainly ways to corral in a random signal to fit within particular parameters to allow for more predictability. That’s the basis of stochastics. Sapél has a range probability knob which directs it to choose output values predominantly from a particular range. Marbles has a switch for voltage ranges. Wogglebug similarly has a mechanism for more or less drastic changes. You can even hone random voltage in on your own with a little offset and attenuation. But it’s still random, and even if completely random might work for certain facets of a patch, using it as the main driver becomes difficult while still remaining musical.1 But I was determined to make it work, and I had just the module to help.
The Addac506 Stochastic Function Generator is a powerhouse modulation source made up of four function generators, and most of the bells a whistles one could want. Need cycling envelopes? Check. Slew limiter? Check. One shot envelopes? Check. Audio rate to very slow? Check. EOR and EOC trigger outputs. Check. Offset and attenuation to get your generated functions in the exact range you need it? Check? Some comparator action for related modulation? Check? And this checklist of features goes on and on. But the killer app of the Stochastic Function Generator is its ability to set a very precise range for both Rise and Fall times. This ability allows for some very compelling modulation that changes every cycle, but, particularly when using very slow modulation, does so organically in a way that seems transparent. It can be set to no random generation (standard envelopes), a very wide range with wildly changing Rise and Fall times, or a very narrow range where changes are subtle, and each of these can be done in three modes, slow, medium, or fast (per generator). The Addac506 is a very powerful module with a compelling feature set that can drive entire patches.
I had set out to make another E370 patch using slow modulation to fade the four wavetable oscillators in and out to create a cloud of a chord that is constantly shifting, yet still always the same. I’d been using slow moving bipolar signals for this purpose on several patches of late, but I wanted to try something a little different this time around. I’ve used chaotic systems, and I’d used both free running and synced LFOs, but I’d had problems using cycling unipolar functions in the past because no voice is ever truly out for very long. It’s a constant chord where each note changes volume, but is almost always audible. Mystery, drama, and tension are minimized; each note like a yo-yo rather than a graceful flow in and out and back in again. But the Addac506 is a little different. Unlike most function generators, with the Stochastic Function Generator you can tailor the outputs to any range you want. Those functions needn’t be unipolar, nor with wide ranging levels, and the ability to offset and attenuate signals allows one to shape your function to suit your destination without issue, and it’s this ability to perfectly condition CV before ever leaving the module that enabled me to use it in this slow fading movement I was looking for.
Setting the stochastic functions on the Addac506 is simple: set minimum and maximum Rise and Fall times, and call it a day. At the instantiation of every cycle both Rise and Fall will receive a random value between those set minimum and maximum times, and that stochastically determined envelope will come out. Flick the cycling switch and every cycle is something different. And the Addac506 gives us that, times four. It should be noted that this behavior can be replicated in any function generator that has both End Of Rise and End Of Cycle gate/trigger outputs, plus CV inputs to control the rise and fall times independently. You’ll need to add in a sample and hold or random voltage generator, plus a module for offset and attenuation to define the range of random. All this times four is a lot of patching, and a lot of modules. The Addac506 does it all seamlessly under the panel.
At first I was a little confused. Although I hadn’t initially set any offset, the output was still silent until the voltage as somewhere between 1-2 volts.2 I had anticipated needing to use negative offset to create some space between the fading out of one wave and when it will become audible again, but instead found myself using positive offset to get the flow I was after. Fortunately the flexibility of the Addac506 allowed for quick and easy adjustments to put each wave in the right zone. I added a fair amount of positive offset to the bass note of the chord so that it was always audible, while still having level changes to keep it moving. The other three oscillators used a very slight positive offset. Combining their need to get to somewhere between one and two volts before becoming audible, with the very slow nature of these ever-evolving functions determined the use of positive offset. Up to 40% of their positive range was already inaudible. Giving the functions a bit of a voltage floor rise was in order, lest far too much silence ensue.
Having already decided on my four oscillators for this chord soup, the quad wavetable oscillators of the Synthesis Technology E370, and having had the level control roughly framed out, it was time to look at modulating the wavetables to create movement within each note. The pitch of each oscillator would remain static, and to accompany the change in level, a change in timbre is natural (especially with a morphing wavetable oscillator). Having already used up the outputs of the Addac506, I looked to a familiar module, and one right next to the Stochastic Function Generator, the Addac508 Swell Physics.
Although Swell Physics is definitionally a chaos-based system, and not a random one, the uneven flow up and down of each output was perfect for the job. The bass note oscillator received modulation only on its Detune CV input, while the other three oscillators received both Wave and Detune modulation, the former from Swell Physics and the latter three free running sine wave LFOs from Batumi II. The triangle wave LFOs from Batumi II were used to pan these same three signals slowly through the stereo field with the ST Modular Sum Mix & Pan, while the bass note was planted firmly in the center. Once mixed down to stereo, the four oscillator cloud went to the AI Synthesis 018 Stereo Matrix Mixer.
Once in the matrix mixer it was time to smooth out some of the rough edges with the Holocene Electronics Non-Linear Memory Machine, a favorite of mine since it went in the rack. As a non-clockable delay I found it a wonderful tool for long, drawn out ambient delays, especially with how it can smear repeats into an almost reverb-like sound. Although I tend to pretty heavily modulate the NLMM, I chose to go with no modulation in this patch, with smearing at about 12 o’clock on the knob, and feedback at around 11 o’clock. The NLMM was mixed with the dry signal in the matrix mixer at about a 50:50 ratio, and perhaps even favoring a bit more delay than dry signal.
This mixed signal was sent to both the output mixer and The Dradds for some broken tape machine treatment. Both sides, left and right, or light and dark as it exists in my case, were set to Tape mode at an octave up, with each side behaving slightly differently. The left side tends to randomly switch between forward and backwards playback at double-speed, while the right side randomly switches both direction and speed, although both sides were lightly modulated and sometimes change temporarily. Although I’ve always been enamored by the Dradd(s), it wasn’t until I really started to learn it have my results been what I was always hoping for. Since I sat down to really understand the Dradd(s), my patching has had an opportunity to be more intentional, with my results more satisfying and effective. Plus, having dual Dradds for a stereo field really maximizes the movement and variation it creates.
A heaping portion of distortion, courtesy of the Bizarre Jezabel Mimosa, was also used towards the end of the patch, which created occasional blips and stutters, and a sense of heavy drama. It was only used on the chord cloud and the Dradds, and even then I chose to keep a fair bit of the original dry signal in the mix. I really like the resulting sound, but something is happening that I don’t quite understand. As distortion was introduced into the mix, I did not lower the dry signal. I would have expected the volume would rise, yet it didn’t. The volume got lower, even once the dry and wet distortion signals were nearing full volume in the matrix mixer. It wasn’t until I started to lower the dry signal (the chord, its delayed signal from the NLMM, and the Dradds) that the overall volume started to become louder. My guess is there is some form of phase cancellation happening, although I don’t really know what phenomenon might explain it. You can hear it starting at about the 10 minute mark, with me starting to lower the dry level at about the 11:22 mark, which corresponded in a total output level rise. Curious, that.
Although I was quite happy with my shifting chord after a whole lot of tinkering with envelopes and CV input attenuators, I knew something was missing. I had blips of granular synthesis to break things up, but it needed more. I was hesitant to use SD Multisample as I’ve done a lot of that, but wanted something in a similar vein. Reaching back to one of my very first patches, an attempt at creating a wind chime-like sound and pattern, I decided to add in a randomly created, toy piano sound (or baby R2-D2 sounds as my oldest relayed ). I really enjoyed making that patch when I did it the first time around. It was the first patching technique I’d ever figured out on my own, and a sense of nostalgia pushed me over the edge, even if only to see if I could remember how to replicate it.
This part of the patch started at the Stochastic Function Generator’s “Average” output. This output was patched to the input of the Joranalogue Compare 2. When the Average out from the Addac506 was inside the comparator window, it spat out a gate to the cycling input on a Frap Tools Falistri. While the function was cycling, the End Of Cycle gate triggered Sapel and Quantermain. Sapel sent a value from the N2 output to both Quantermain, in order to send quantized pitch CV, and the Decay CV input of Falistri that would change the envelope length with each note. The VCO changed pitch, and the cycling envelope opened the VCA, each note with a different pitch and length. The sound went from the first VCA to the output mixer, as well as a second VCA which panned the signal to two separate delays, which were then mixed and sent to the output mixer to be mixed with the dry signal and the rest of the patch.
Although the method I ultimately used in this patch is very different than the first time around, the results are similar. The sounds are much higher in pitch, much faster, and much shorter than my original version, but the theory behind creating the sounds was the same, even if I took a different route to get there.3 I wanted short, uneven bursts, and I got them. It’s quite often these days that I realize there are many routes to the same end. Overall, this part of the patch played a very minor role, but an important one. After several listens, I’d probably integrate this part differently were I to do it again. I’d definitely have it more forward for most of the mix rather than hanging around the background. It seems to get lost sometimes, when it should be more prominent. I would also not mess up a patch connection with my second delay, the Qu-Bit Electronix Nautilus, where I plugged into the Right input, and not the left, which, with the feedback pattern chosen, put virtually all of the delay signal from Nautilus on only the Left side. Some might not even notice, but it bothers the shit out of me.
The last part of this patch is the periodic “bubble burst” with accompanying echoes (courtesy of the Xaoc Devices Sarajewo). I stole this idea straight from one of my favorite patches on the Make Noise channel. It’s a simple sound that is both unobtrusive, and only very periodic. The sound was created by pinging (ringing?) CUNSA, which also had its frequency modulated by the highly attenuated HP output of the same filter (another trick I stole from Make Noise – Thanks, Walker!). It was triggered by using a Stackcable to combine three separate gate outputs from the Stochastic Function Generator. These long functions output only very periodic gates at a random interval, so there’s no overcrowding. It also helped provide a sense of scale and depth of the musical space.
Everything ended with a round trip through the Walrus Audio Slöer, which is quickly becoming my favorite reverb. Although I’ve tended towards one of the pitch shifting algorithms, I ultimately decided on using the Dream algorithm, though the Rain algorithm also sounded nice.
I really enjoyed this patch, even if there are a few things I would definitely change. It was both challenging and highly rewarding. It offers answers to some questions, but also to more questions to explore in a future patch.
Although what is or isn’t “musical” is certainly subject to wide interpretation, I think we can still make useful generalizations, while drawing lines between something most would call musical, with yet other examples most would call sound(s) or noise. ↩︎
I’m not sure, but I suspect that the VCA/mixer I was using to process the audio has a logarithmic response, delaying its response with slow(er) signals. The manual doesn’t illuminate the VCA topography, unfortunately. ↩︎
After looking at a patch diagram of the original patch, the methods used between these two patches was quite different, even if it started similarly. It’s nice to know that there is generally more than one way to accomplish some patching goal. ↩︎
There’s a patch that caught my ear in the days after a recent shoulder surgery while I was stuck in the recliner hopped up on pain killers, and looking for every distraction I could find in the synth-based slice of the internet to keep my time occupied. Unfortunately for my wallet during this time, I happed upon Tom Churchill’s excellent tutorial demos on the Xaoc Devices Odessa and Sofia (which led to an entire new Xaotic Dreams Subsystem). Stazma’s (AKA The Junglechrist) demo of the Verbos Harmonic Oscillator (particularly how it pairs so beautifully with the Multi-Delay Processor) led me on similar path with Verbos.
But it was one patch in particular that I kept coming back to which had me captivated. Over and over I’d listen to it, allowing its peace to wash over my ailing mind and soothe my temporarily excruciating existence. As I watched I would try to make out through an oxycodone fog what was happening, hoping that as soon as I was able I could create something like it. I took notes as best I could, and I traded my BitBox Micro so that I could acquire a Disting Ex. Although BitBox Micro is capable of playing polyphonic multisamples, and is a fantastic module in its own right, it can only do so with MIDI, and I needed to use CV and gates.
My first foray with this patch was a mess. The basic framework was more or less built out, but lots of things were off. My first problem was a horribly calibrated Disting Ex. Its pitch was wildly off, and nothing was in tune despite receiving carefully calibrated pitch CV. Once that was fixed things were much better, but I still had far too many dissonances that weren’t at all what I was looking for, and I wasn’t sure how to fix it. I couldn’t get the speed to something I felt was just right. I couldn’t get a satisfying note distribution. I wasn’t at all satisfied with it. I recorded it, but every time I’d listen back, I’d hear mere flashes of my goal disappear into a dark cloud of dissonance. Each listen prompted me to revisit the idea.
I’ve made over a dozen other modular recordings since that first try. Jamuary was productive. Not a single one of them attempts to use polyphonic multisampling. Not a single one of them uses Marbles or Disting Ex (in any capacity). I’d stepped away from it. During that time, when I would take a few moments to think about the patch, I gained some perspective, as well as some newfound patching knowledge and experience with the various patches I was making at the time. I was now better prepared to revisit this piano patch with a fresh take.
I knew that one of my initial problems to solve was that the original patch had too many notes available for CV to address. My first attempt at this patch was in C Major, and I left all 7 notes of the scale available between C1 and C5 to all 3 CV inputs. Marbles spits out randomly generated pitch CV and gates from its X and T outputs, which goes to the 3 CV and gate inputs on the Disting Ex, via Quantermain in the Micro Ornament and Crime. Despite having quantized pitch CV, using the maximum number of notes meant that any number of dissonances can happen at any point. Nobody wants to hear E juxtaposed with F, or any other dissonant intervals, with any regularity. Particularly not when there can be several dissonances occurring simultaneously. It’s harsh, when I’m looking for smooth.
Along with a key change (C Major to C Minor) I also made adjustments such that each pitch CV input would only generate particular notes of the scale, and not all of them. This change made dissonances few and far between, which meant those dissonances were now artistic tension that would quickly resolve to something more pleasing, and not a stream of clashing notes that barrel into one another, overtaking the piece. I made some minor adjustments to the Rate and Jitter controls on the Mutable Instruments Marbles to get a more satisfying pace and amount of sloppiness in note instantiation. I also tweaked the note distribution to get something resembling a bell curve in pitch generation. More notes in the middle of the range than at either extreme. Super low and piercingly high notes are good for effect, but not as a matter of regular course throughout a piece. These notes are generated randomly, but the process still needs well set boundaries in order to remain interesting. Unfettered random is every bit as boring as a fully repetitive pattern.
Now that I had a stream of piano notes I was content with, it was time to address the FX, which is what would make this piece interesting. The piano sets the path. The FX create the atmosphere around that path. During my first attempt at this patch, I tried to keep FX to a minimum. Some reverb courtesy of the Mutable Instruments Beads, followed later by its granular synthesis engine. The FX were sparse, but not really in a good way. The piece was empty and cluttered at the same time. Octave repeats spiraling off almost uncontrollably, awkwardly filling in empty space and becoming unruly when note generation temporarily sped up. It was a mess. A mess so bad, in fact, that I simply stopped recording out of frustration, and pulled the patch apart.
This time I started with the reverb. Although the FX is what would be most interesting about this piece, I knew I wanted to have the reverb set based on how it sounded with just the piano notes. I wanted a massive open space. Something between a huge cathedral and space. There needed to be bounds, but they needed to be pretty far out there. I had initially chosen to use the Vongon Ultrasheer for the reverb, but in a last minute decision decided to use the Oto Bam with its Ambient algorithm. Some adjustments to the size and decay, along with a bit of modulated chorus adjusted in the Bam, and we were set. In retrospect, I wish I had added even more chorus to lofi it up, or that I had used the Ultrasheer with some randomly modulated vibrato, but I’m still quite pleased with the result.
But what next? I had bad memories of out of control granular repeats, so decided to forego using it in favor of delay. I knew I wanted some reverse delay (because if it doesn’t have reverse delay is it even ambient?), but all reverse delay all the time quickly became taxing, its zips easily taking over the soundscape. So I decided to work on panning technique which would gradually move the source piano sounds between 2 different delays using 2 different sorts of repeats. The technique isn’t hard, even when adapting it to a full stereo signal. You can find the patch diagram here. These 2 delays are mixed, creating seamless transitions between 1 delay and the other. Very cool. I got movement and gradually changing variation. As one delay fades out, the other begins to fade in. Pretty much exactly what I wanted.
But even with gradually shifting between 2 wholly separate delays, the reverse delay was still a bit over-prominent, which meant that I needed a method to turn the reverse function on and off. Chaos to the rescue. After some tweaking to various facets of my clock, the delays were sorted. But there was still something missing.
I’ve had Panharmonium for a long time. It was one of my first large Eurorack purchases, and I definitely didn’t have the modular chops to use it effectively. In fact, I’d had a very hard time getting anything nice sounding from it. I could never seem to find the right analysis interval, and a couple of other settings really mystified me, resulting in what I can only describe as sonic sludge. A glob of abruptly shifting dissonances that sounded terrible. It was nothing like what the many YouTube demos promised were all there to be unlocked.
After watching a video titled “How I Use My Rossum Panharmonium (Part 1)” by Baséput I noticed something. The source he used is fairly sparse. I had been using full melody lines as a source and it hadn’t worked out at all. He was using the granular-like output from Morphagene as a source, which is fairly sparse, and had a beautiful result with just a couple of knob twists. Depending on your settings, just a second of a Morphagene output could hold an entire piece’s worth of sonic info. With that info in mind, I decided to use the mixed delays as my input into Panharmonium. A few more tweaks to the delays to give me shorter repeats, plus an octave up on one of them, and I felt like I had good sonic ingredients for Panharmonium to shine.
And it worked. Despite never having anything special from Panharmonium before, I was greeted almost instantly with exactly the sorts of sounds I’d always imagined making with it. In just 1 patch the Panharmonium turned from a module I had listed for sale into one I doubt I’d ever get rid of. The results are an almost choir-like accompaniment, which gives the entire piece an ethereal feel, which was pretty much what I was going for. More on this to come.
