Posts

Posted on

Adagio for E370 in Dmin7

It had been a couple of weeks since I’ve been able to patch, and I was itching. Having just been on a spring break high school baseball trip to the beautiful Emerald Coast, I was inspired by the sea. It was the first trip in over a year I hadn’t taken some music making device, and even if I didn’t really have the time to patch during this trip, I missed not having something.1 I thought all week about what I would be making if I had my synth, and Swell Physics was first and foremost in those musings.

When I first set up this patch it was a mess. A cacophony of sound where each part seemed to work well on its own, but as soon as it mixed with other parts the whole thing turned to shit. As a non-professional musician, I try not to fall into that trap of being too hard on myself when my results are less than what I’d hoped for. Failure, after all, is the key to progress and improvement, and if a recording sucks it’ll just go into the folder of dozens of other recordings that suck. My livelihood is not at stake, and I don’t yet have a reputation to ruin. When I turned on my synth today, I listened to that latest patch still plugged in, and I was unhappy. It was a mess, and only reminded me that the recording I had heard of that patch a dozen or more times that week just wasn’t it. I was resigned to tearing the patch down and starting from scratch, but since I wanted to use the same control scheme I instead chose first to tweak a few settings and see if I couldn’t salvage the patch.

I don’t really like tweaking already-built patches. I mean, I like tweaking controls, but not patch points. It’s generally not been an exercise that has produced meaningful fruit, and I usually find it easier to simply start over rather than navigate a nest of patch cables. Today’s first change, however, wasn’t a physical patch change, but a setting change to the E370. Rather than using one of the User wavetables I’ve taken fancy to, a very cool wavetable leftover from the previous owner, I switched all four channels over to ROM A, and the difference was immediate. In fact, it was pretty much the sound I envisioned before i even started patching. So much for tearing down the patch.

The patch starts, as many of my patches often do these days, with the Addac System Addac508 Swell Physics. Since first receiving Swell Physics many of my favorite recordings are based around it. Whether I’m using the wave outputs directly to control levels, or in conjunction with modules like the Nonlinearcircuits Numberwang and Let’s Splosh to control various facets of a patch, I’m constantly finding new ways to integrate it into my work. The four outputs were sent to an array of Intellijel Amps’ CV inputs to control the levels of four modulated outputs from the always lovely Synthesis Technology E370 quad wavetable oscillators and blue noise to simulate the spray of the ocean. Each oscillator was fed pitch information from the NOH-Modular Pianist, one note each of a D minor seventh chord (though I’ve no idea what the oscillator was tuned to). From the Amps, the mixed noise and audio outputs went to the ST Modular Sum Mix & Pan to place in the stereo field of the mix.

Modulation of the E370 wasn’t particularly heavy in this patch, but each oscillator’s X axis was moved around by a complex mix of CV from the Nonlinearcircuits Frisson mixed together in the Atomosynth Transmon CV controlled matrix mixer. Four randomly chosen outputs from Frisson were sent to the four inputs of the Transmon, with the other four Frisson outputs modulating four separate nodes in the mixer itself. This modulation to the E370’s X axis was at a medium slow speed and moved waves around beautifully, slowly shifting the sine wavetables.

The Atomosynth Transmon is a curious module. Nominally it’s a matrix mixer, made by a very niche maker based in Lima, Peru. But this matrix mixer has several tricks up its sleeve. The first row can be used for the attenuation or attenuversion of signals, a switch controlling which mode is active. The second row can do both, and each node has an individual output. The last two rows each have a discreet VCA input per node, also with individual outputs. If one of those individual outputs is used, it’s removed from the final mix of the row. Each node can also be used as a 12V offset voltage using jumpers on the back. The Transmon is a very powerful and flexible mixer that can be used in a myriad of ways. Most often I use it for simpler mixes, or even as individual VCAs, but this patch called for going whole-hog, mixing four chaos signals, while using four of Transmon’s individual VCAs to modulate my modulation. These new modulated modulation CV mixes create all of the tonal shifts in the E370 voice as they move the sine waves through a series of notes along the X axis of the wavetable. The pitch CV being sent was constant through the recording. I could have manually tuned each oscillator to those notes and the result would have been no different.

Although I was much more pleased with the result after changing wavetables, it wasn’t enough change. The original delay I had chosen for the E370, the Holocene Electronics Non-Linear Memory Machine, was no longer the best tool for the job. Although its inherent aliasing worked great with the original wavetable, I didn’t like it at all with the much cleaner waves I was using at the time. It created too choppy a repeat, even with some diffusion and without pitch shifting. So I switched over to the very clean Olivia Artz Modular Time Machine, which worked a treat. With a medium length delay time, notes were lengthened and it gave the overall sound a bit of thickness, even if I didn’t pay much attenuation to how I set up each tap.

With the primary voice more-or-less set, it was time to fill in the cracks. The flowing waves of the E370-plus-noise mix with its delay and some reverb were beautiful, but lonely. It was heavy on high(er) pitches, and I wanted to add a bit of body to it. In comes Panharmonium. Though it’s highly capable, I very rarely use Panharmonium as a main voice. But I use it all the time as a supplement to the primary voice as a means of reinforcement, most often shifting the pitch in one direction or another so that various frequency ranges are represented. In this patch I initially went with a two octave downward shift. It was deep, but not overpowering, in a beautiful way, though there was simply too much that wasn’t being heard at all because the frequencies were just too low, which too often caused a flutter that sounded like clipping. No good. So I compromised by going down only one octave, (introduced at 1:15), which transmogrified the audio into a warm bath. I don’t normally modulate Panharmonium, I haven’t found it needed, but for this patch I decided to use offset and attenuated versions of the Swell Physics Average output to modulate both the Center Frequency and Bandwidth, which created a slight swirl in the output which added texture.

But I didn’t stop there. The Panharmonium voice was beautiful, but it needed delay for a bit of thickness of its own, so I sent its output to a pair of very slightly driven Echofix EF-X2s, with its speed lightly modulated by a triangle wave from a Frap Tools Falistri, giving it a hint of vibrato in the delay tails. The offset delays helped create a beautiful stereo image, and added a slight bit of focus.

Now that we had a beautifully flowing soundscape of waves, it was time for some ornamentation. Recently I’ve been reaching primarily for my pair of ever wonderful Pladask Elektrisk Dradd(s) when I’m looking for granular synthesis, but I knew I wanted at least one thing that Dradd cannot do: a two octave shift to create sparkles. It can do a one octave shift in Tape Mode, but not two octaves. So I turned towards the Mutable Instruments Beads (introduced at 2:30). I’ve had Beads a long time. It was one of my first modules, bought on a whim when I saw they were in stock one day. Beads can be many things from delicate to bombastic, and with this patch I wanted subtle, and Beads excels at being subtle. Like all granular processors, finding a nice sweet spot (or range) can be a challenge, but the controls are intuitive and once you find that range, Beads can produce wonders to behold. With slight modulation from Tides to Size, along with use of the attenurandomizer circuit to Time, and I had a beautiful spray of shimmering grains to follow in the wake of the E370 and Panharmonium waves. But even with this beautiful flow of waves I wanted something more. Something to use as juxtaposition. Something with more edge.

One module that I’ve criminally underused is the Frap Tools Brenso. I’ve certainly used it before, but it’s daunting, and I haven’t really put much brain power towards learning it, so I’ve tended to use its standard shaped outputs, while ignoring the gobs of waveshaping and wavefolding on the right side of the module. It’s a shame, really, but I thought that this day would be a good day to start exploring the extensive tonal options on the panel. I decided to start simple, using Sapel’s two smooth fluctuating random signals to slowly and lightly modulate both the waveshaper and wavefolder. This created an almost crackling electricity sound at times. It sputtered. It growled. It was an uneasy signal, fraught with fear while trying to exert its independence. I ran the Final output from Brenso through CUNSA, mixed with noise like the E370 waves, for some light filtering to attenuate some of the most egregious frequencies. The last time I made a patch like this I used a sequencer and pure sine waves for the epic sub bass line. It was way too loud, even if it was intentional, and repetitive. It was also pitched far too low. This time around I decided to use the same module to create the bass line, but rather than use a sequence, I used the Doboz T12 in Touch Keyboard mode and manually played the part. I wanted to time the note changes by feel rather than a clock. It’s still too loud, and still probably too low in pitch (some lessons are harder to learn than others), especially as I go up in register, but placing this very uneasy and slightly dirty wave in contrast to the smooth waves of the E370 and Co was a nice touch. Were I to record this patch again, I would definitely do a better job of controlling the level of this voice. I think the recording still sounds great, but I was certainly a bit overzealous with the volume of this voice.

For reverb I used my trusty Walrus Audio Slöer in Light mode for a choral sounding octave up shimmer. This reverb really has become my primary reverb. I only move away from it once it proves itself to not be the best tool for the job.

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

Modules Used:
Addac System Addac508 Swell Physics
Addac System Addac814 6×6 Stereo Matrix Mixer
Synthesis Technology E370
Nonlinearcircuits Divide & Conquer
Nonlinearcircuits Frisson
Atomosynth Transmon
Intellijel Amps
Intellijel Quad VCA
Olivia Artz Modular Time Machine
Frap Tools Sapel
Frap Tools Brenso
Frap Tools CUNSA
Frap Tools 333
Frap Tools Falistri
Doboz T12
Rossum Electro-Music Panharmonium
Vostok Instruments Asset
Mutable Instruments Beads
Mutable Instruments Tides v2
ST Modular SVCA
Knob Farm Ferry

Outboard Gear Used:
Walrus Audio Slöer
Noisy Fruits Lab Lemon
Echofix EF-X2 (no idea why this is no longer on their website)

  1. Technically I had my iPad, which, from my Jamuary experience, is more than enough music making machine, but I simply wasn’t inspired by anything on it at the time. ↩︎
Posted on

Bonus Track – Colored Strings Rev 1

I didn’t set out to re-record my latest patch the other day. I was set on using some other sample set in some other way. But there were a few things about my last recording that bothered me a little. I was getting too many pitches that were too high; well outside the zone of what “Low Strings” should be. There were points when way too many colors were being shouted at listeners. The delay was a bit too much. And I wanted to streamline the patch a bit.

I won’t go too in depth on the patch, you can read about the details here, but I did make a few changes that improved the recording substantially….

The most major change I made was to eliminate the Qu-Bit Electronix Nautilus, and use the pair of Echofix EF-X2 tape delays on the string samples. This helped to both fill out the space, while simultaneously cutting away excess in the overall soundstage. Rather than simply repeating notes and phrases, the tape delays lengthened and widened them.

The Panharmonium (introduced at ~2:40) was pitched down an octave, which helped lend some depth. More voices were added, and it was set to analyze a wider spectrum. The Panharmonium, along with the Walrus Audio Slöer (which was also set to pitch an octave down), really added weight to the entire recording. I removed a delay from the Panharmonium’s signal path (previously the tape delays), and allowed it to sing only through the reverb, which was a substantial improvement.

No changes were made to the Dradd(s).

No changes were made to Plaits-does-Robot-Speak except sending it far fewer gates.

Modules Used:
Nonlinearcircuits The Hypster
Nonlinearcircuits Lets Get Fenestrated
Nonlinearcircuits Stochaos
Nonlinearcircuits Triple Sloth
Vostok Instruments Asset
Expert Sleepers Disting NT
Addac System Addac814 6×6 Stereo Matrix Mixer
Rossum Electro-Music Panharmonium
Pladask Elektrisk Dradd(s)
Mutable Instruments Marbles
Mutabke Instruments Plaits
Mutable Instruments Beads
Knob Farm Ferry
ST Modular SVCA
Intellijel Quad VCA

Outboard Gear Used:
Echofix EF-X2
Walrus Audio Slöer
Noisy Fruits Lab Lemon

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

Posted on

Made Noise – Sketch 7 (A Multimod Experiment)

I’ve had a full Make Noise synth for a while. In fact, I had far more Make Noise modules than I had space for, and with the recent release of Multimod, their new modulation tool, I decided to spring for a second 4-Zone CV Bus Case to avoid having to constantly rearrange the case from patch to patch. Multimod is an interesting tool. Billed as an eight output buffered multiple with several tricks up its sleeve, it presents a new method of spreading modulation signals around a case. It uses phasing, time, and speed manipulation in order to conjure this newfound magic. Fugue-like sequences can be created, as can wild modulation schemes, and many other things besides, all created from a single input signal. Being a Make Noise fan, I ordered one knowing only that it was a modulation source. If Make Noise has a gap in their lineup, it’s modulation sources. Once I saw the initial demo videos by Sarah Belle Reid and Red Means Recording, I ordered a second.

Although upon receiving them I used the new Multimods briefly to get an idea of how they worked, I hadn’t yet delved into what they can really do. I roughly recreated a couple of patches found in the videos, such as sending a sequence to several oscillators, and even trying it as a multitap, pitch shifting delay, but these were surface level primers, not any sort of real exploration. Being that Multimod presents a very new way of sending modulation around a synth, I was unsure how I might want to use them. My initial thought in ordering two was to use one for CV manipulation, and the second as an audio processor. I’ll certainly do more of that in the future, but I wanted to see if I couldn’t tackle an issue I noticed right away.

One of the highlights of the video demonstrations were the creation of fugue-like sequences, where many oscillators would receive the same sequence, but at different speeds or phased in time in interesting ways. But I also noticed immediately that although the sequence pitch data was spread around, there wasn’t a good way of spreading around an accompanying gate sequence for articulation. Some attempts at remedying this conundrum that I’ve seen involve using a separate gate sequencer, or even just a clock. And while deriving pitch and gate sequences independently is a particular strength of eurorack, that method seemed haphazard and insufficient for this application. I wanted more. I wanted the entire sequence at all of the outputs regardless of the speed or phase. So I thought about it, and suddenly a potential answer came to mind. Use one Multimod for processing the pitch sequence, and the other for processing the gate sequence. But there was still the problem of movement. I needed to move the knobs the exact same amount at the exact same time in order for the outputs on both modules to follow. How would I modulate them so that they were always doing the same thing at the same time and the gate sequences were following their pitch CV counterparts?

The CV Bus.

Although a Maths channel two or three could be used as an offset in the same way, I decided to use the CV outputs of the Noisy Fruits Lab Lemon, a very excellent standalone CV/MIDI controller, to manually control Multimod’s Time and Spread controls. My goals here were simple: to use two Multimod’s to control a single René CV and gate sequence, with the outputs on both being in sync with one another. If Multimod were an all analog module, I think this experiment would likely be dead in the water. I’m not convinced that parts and control tolerances would be close enough such that they would remain identical through modulation. But since Multimod is all digital, I surmised that so long as all of the Init settings were the same on both, they should (very theoretically) react the same to identical control voltage received at the same time.

To start, I copied a x2 clock from Tempi to both Multimod Tempo inputs, with a second /16 clock to both Reset inputs. This ensured that the clocks of both units would always be in sync. I then patched René’s X CV channel to the first Multimod input, and the X Gate output to the second Multimod input. The Init setting on both Multimods were simple: Phase at full CCW, Spread at Noon, with the Spread CV attenuverter at full CW, and Time at full CCW. With these settings, things worked just as they should: a single CV and gate sequence from all eight outputs happening at precisely the same time. So far, so good. I then patched the Lemon fader CV outputs to the CV Bus, and then to both the Time and Spread CV inputs on both Multimods.1 A single CV source should (again, theoretically) modulate both units identically such that as the CV sequence speeds up (or slows down), the gate sequence simultaneously does the same, and corresponding outputs on each Multimod should remain in sync.

For the oscillators I chose the Spectraphon.2 The first Multimod output one was patched to Oscillator A, with Output four patched to Oscillator B. The mixed outputs of each side of Spectraphon were then patched to QMMG inputs one and two. Outputs one and four of the second Multimod were patched to QMMG’s CV inputs, with both QMMG channels in LPG mode (though it sounded cool in LPF mode too). With identical modulation going to both the Time and Spread CV inputs of both Multimods at the same time, the CV and Gate sequence outputs should (theoretically) also be in sync with changes. The CV sequence from output one on the first Multimod should match the gate sequence from output one on the second Multimod. Ditto with the other outputs.

To be honest, I’m not sure if the patch worked like it should have in theory. What I mean to say is that the result is ultra-cool whether theory was borne out in practice or not. Things sound correct with the resulting sequence from Multimods output four, the output that is affected least by changes to Time and Spread. But I can’t really tell if the pitch and gate sequences from Multimods output one, the output affected most by changes to Time and Spread, is synced in the same way. The sequence is moving pretty fast. Hearing individual triggers can be tough at super high speeds, especially when slamming triggers into vactrols, like those in QMMG. Not only is there inherent bleed, but the fall response is such that you can’t always hear individually triggered notes if they’re in quick succession. DXG would likely have been a better candidate for testing the precision of the altered sequence because each trigger is annunciated clearly in a way that isn’t possible with vactrols. That said, I’m a huge fan of vactrol bleed. It can be a beautiful effect when used with intention, and although I wasn’t looking for any particular sound with this patch, vactrol bleed certainly showed off its character in this test recording.

This diagram is missing the clock and reset to the Multimods.

The QMMG outputs were sent to X-Pan, with the Channel four sequence mixed in the middle, while the Channel one sequence was slowly panned left to right and back again through the stereo field by a cycling Maths triangle function. The stereo output of X-Pan was sent to Mimeophon for some finishing with beautiful repeats and Halo.

Modules Used:
Tempi
René Mk2
Multimod
Maths
Spectraphon
QMMG
X-Pan
Mimeophon
Wogglebug

Outboard Gear Used:
Noisy Fruits Lab Lemon

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

  1. As mentioned previously, the same thing can be accomplished with any offset generator like Maths or ChnSvr. ↩︎
  2. Seriously, Make Noise: Spectraphon is awesome, and I’ve never even explored it beyond SAO mode with the default Spectra. ↩︎
Posted on

Colored Strings

Ever since I started using the Polyphonic Multisample algorithm in the Disting Ex, and now the Disting NT, I wanted to do a string patch. For some unexplainable and inexplicable reason I never did. I got a taste of using string samples during a handful of Jamuary patches, particularly Jamuary 2507 and Jamuary 2510, when I experimented with using the very excellent Alexandernaut Fugue Machine and Decent Sampler, but I really wanted to do a string sample patch modular-style. At first I wanted to use the same string samples I used during Jamuary, the DK Solo Cello Spurs from Pianobook, but that did not work at all.1 After quickly pivoting to the LABS Low Strings Long samples included with the Disting NT, I was off.

Much of this patch is a simple altered duplicate of my last patch, so I won’t rehash the entire patch here. Many knob settings are different, and of course I triggered string and not piano samples, but the control and audio paths are largely identical. Because these string samples are inherently longer than piano notes, I had to drastically slow down The Hypster, the initial source for the master clock, in order to compensate and have Stochaos produce fewer gates. This slow down was coupled with using outputs on Stochaos that change much less often. I also tuned in the delay used on the string samples, the Qu-Bit Electronix Nautilus, to what I thought sounded best, thought I can’t recall the settings other than a long delay time with moderately high feedback. The Panharmonium settings were also changed drastically, cutting the number of voices, zeroing in on the desired frequency range, and really tuning the sampling time. Panharmonium is not always instant bliss, but given some gentle massaging it can become transformative. The Dradd(s), however, weren’t changed at all.

But even after being able to hone in on what I had hoped for the strings and the existing sound chain, there was something missing. It needed some kind of ornamentation. Something to juxtapose against the somber sounding string sequence plodding along in C minor. I initially thought of high pitched bells or sparkles of some kind, but then remembered a patch I did last summer as a test shortly before bringing a travel synth on a trip to Alaska. In that patch I used Plaits in the Vowel and Speech Synthesis algorithm (the last Green algorithm) to repeat four colors, Red, Orange, Yellow, Green. In this patch I took a slightly different approach, by using a random output from a Mutable Instruments Marbles to select the spoken color. I was quite happy with the result in the moment, but was carried away a bit during the performance with too many triggers. It should have been an occasional color spoken in a sad-ish voice to reflect the somberness of the strings, though it sometimes ended up being a robot talking over himself. I used a CV output from the Noisy Fruits Lab Lemon to control Marbles’ clock, which had a high amount of Jitter so as to not be regular. When I wanted more from Plaits, I pushed the fader to create more gates with Marbles. However I clearly pushed it too far during a couple of points. Initially I had Plaits running straight to the mixer, but later in the recording used Beads in full wet delay mode, which altered the tone slightly, and added a low number of soft repeats that really only served to add to the confusion. Plaits is simply triggered to often.

The proverbial icing on the cake was the Walrus Audio Slöer in “Light” mode, which is an octave pitch shift. But rather than a standard pitch shift, Slöer adds more of a choir or symphonic strings sound, which is absolutely beautiful. This patch used basic settings. Pitch shift volume at max, and clock speed at the slowest setting. There was a long decay, and a moderate amount of modulation.

Overall I’m quite happy with how this patch turned out. It’s beautiful and not deterministic. That said, the piece could use for some composed string sequences, even if only occasionally, in order to maximize tension and relief. I’m definitely going to try other adaptations, particularly in the timing of gates and pitch. I’m thinking the Addac508 Swell Physics might be a good place to start.

Modules Used:
Nonlinearcircuits The Hypster
Nonlinearcircuits Lets Get Fenestrated
Nonlinearcircuits Stochaos
Nonlinearcircuits Triple Sloth
Vostok Instruments Asset
Expert Sleepers Disting NT
Addac System Addac814 6×6 Stereo Matrix Mixer
Qu-Bit Electronix Nautilus
Rossum Electro-Music Panharmonium
Pladask Elektrisk Dradd(s)
Mutable Instruments Marbles
Mutabke Instruments Plaits
Mutable Instruments Beads
Knob Farm Ferry
ST Modular SVCA
Intellijel Quad VCA

Outboard Gear Used:
Echofix EF-X2
Walrus Audio Slöer
Noisy Fruits Lab Lemon

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

  1. I was getting all sorts of sputtering and general ugliness using these samples. I’m not sure if I was overwhelming the algorithm with too many gates, or whether the sample rate or bit depth of the samples was too high, or perhaps something else, but I quickly abandoned these samples in favor of the stock LABS string samples. ↩︎
Posted on

A Piano Dream

As I was un-patching my Fall patch, I got a hankering. I’ve used Multisample Piano in several patches over the last couple of years, but I wanted to make another one with the piano as the focus of the patch, rather than an accompanying ornament. These sorts of patches aren’t terribly hard, but they are fun, and I love relaxing to them as they just play.

My first thought was to get a random distribution of triggers and let it roll. I began by using the same sub-patch that created the Fall emulation; a series of random envelopes cycling within a defined range, with the End Of Cycle trigger striking one of four gate inputs programmed on the Disting NT. That worked okay, but there was something not quite right. I was never able to pin down exactly what that was, but I decided early on to abandon that patch and try a combination of patches that I’ve used before to some really nice effect. Once I decided to switch things up, I knew exactly what I wanted to do.

Let’s Get Fenestrated, by Nonlinearcircuits, is one of Andrew’s newest designs. It’s a triple comparator with the sole job of spitting out gates once the inputs reach certain voltage levels. According to an email exchange I had shortly before its release, Andrew’s idea when designing this module was to create wonky clocks with chaotic sources. When I inquired, I was looking for a couple of tools, some type of comparator being one of them.1 Upon his announcement of its release I had Scopic Modular, the guy I use for all of my NLC builds and all around nice guy, order and build it for me. But despite having had it for a while I hadn’t used it much. In the time between seeking some form of comparator and receiving Fenestrated, I had worked up several patches to get chaotic gates. Numberwang was my primary tool, but also others. But as I started to think about how I would come up with a different clock algorithm for this patch, I immediately thought about using Fenestrated.

The patch started with a chaos signal from The Hypster. I initially went with the U output, because it has the biggest range of the four outputs, but despite modulation happening via patch-programming, the output was just too regular. No matter what I did with the comparator setting on Fenestrated, I got more or less a steady-ish beat. Not on a grid, but just a little too close for the style of gate generation I was going for. Switching to The Hypster’s Y output, and adjusting the window on the first comparator of Fenestrated, fixed that in short order. The new clock output from Fenestrated was patched to Stochaos, which uses chaos (or random, or both) to generate gate patterns. One advantage to using Stochaos is that, unlike Numberwang or using the End Of Cycle outputs on various free running function generators, it generates multiple gates at the same time, meaning I’d have both dyads and chords, along with singularly generated notes, which is not possible using those other methods. Numberwang spits out exactly one gate at a time, and the chances of two random, free running cycles of a function generator finishing at the exact same time is exceedingly low. Having found a good cadence of notes, I moved on to giving them a pitch.

I’m a fan of using a very small number of modules as what I like to call an engine. The thing that makes the patch go. It’s quite often that I’ll use only one or two modules to control an entire patch. Having used The Hypster to control gate generation, I initially decided to use its other outputs as a pitch generator, patching the X, Z, U, and -Y outputs to the Disting NT CV inputs, via the Vostok Instruments Asset so that I might massage the notes for each input into a good range focused on the lower-middle to middle parts of the keyboard. Notes that don’t require one to be a dog to hear, nor ones that often only contribute to a muddy soundstage if used too often, especially in a reverb-rich environment. But I wasn’t completely satisfied with the result, so decided then to use the four CV outputs on Stochaos, which worked wonderfully, even if I can’t explain why it was better.

At first I wasn’t sure how I wanted to ornament the piano, not that a piano and some reverb aren’t enough to be beautiful. I wasn’t set on a sound, so I began to experiment with delays. Normally I would go to the Venus Instruments Veno-Echo, but decided instead to use the Qu-Bit Electronix Nautilus. I wanted intermittent reverse delay, which the Veno-Echo can do, but I wanted to CV control the reverse parameter, and not simply gate it on and off. Although I don’t always appreciate prescriptive controls, if those controls are lightly modulated the result need not feel prescripted. With most delays, reverse delay is reverse delay. It’s on or off. But due to it having up to eight delay lines, the Nautilus takes a different approach. Rather than an on of off dichotomy, it prescriptively assigns reverse repeats as you turn the knob. At full counter clockwise there are no reverse repeats. But as you turn the knob clockwise, you get reverse repeats in patterns. From the manual:

I set the knob at just above full CCW. I wanted reverse repeats, but I didn’t want them to overwhelm regular repeats. Using an attenuated version of one of the Triple Sloths outputs (a medium length cycle) I lightly modulated the reverse knob, which ended in a wonderful mix of mostly forward repeats, augmented by the always beautiful zips of reverse delay. But it wasn’t quite enough. I wanted to make it a bit dusty, so chose to put a very light amount of sample reduction as the Chroma.2 This matched perfectly with the slow clock speed I had running on my reverb, the Walrus Audio Slöer. I next decided on how long of a delay I wanted, and ultimately went with a fairly long delay time, and used an internal cross-feedback pattern for the repeats.

Wanting to fill in some of the space, I decided to go with a combination of Panharmonium and the Dradd(s). I initially had an idea that I would pitch the accompaniment in opposite directions, Panharmonium down an octave and the Dradd(s) up an octave, but that created all kinds of sonic havoc, especially with the already pitched up reverb. It just was too much going on in too many frequency ranges to be coherent. One issue with this patch is that it’s a bit difficult to pick out the Panharmonium. It’s pitched downwards an octave to give the piece some depth, but it seems to get lost a bit. It’s noticeable when Panharmonium is not present, especially as I added some saturation via the Echofix EF-X2 pair towards the end, but it’s hard to pick out as a separate voice in this recording. I’m not sure whether that’s good or bad, though I tend towards wanting every voice to stand on its own. I’m not exactly sure why this voice is so buried in the mix, and I’m not sure if it’s a bad thing in the end, but it is a little frustrating.

The Dradd(s), however, came out exactly like I’d hoped. They were set to Grain Mode, and time stretched at a slow crawl, one channel in reverse, the other forward. I love granular synthesis. The textures it can create are wonderful, and this patch is no exception. Each piano note, and its successive repeats from the Nautilus, stretched to the furthest extremes, filled out space in a particularly interesting way that I found compelling; the Piano notes seemingly stuttered as they were dragged out as long as the Dradd(s) could manage. Not only did the Dradd(s) serve to fill in space, but they added a wonderful lo-fi texture to otherwise smooth piano notes. Beautiful.

One thing I’ve long wanted to experiment with is using multiple reverbs. Not simply stacking reverbs or using two (or more) in parallel, but by trying to use them as instruments unto themselves. I’m not exactly sure when I first heard this technique, though it was surely in the context of ambient guitar, but it wasn’t until I heard Music Major by A Last Picture From Voyager that I saw its full potential. I recently made a recording during which I featured the freeze effect from the Dreadbox Darkness, and it was great, even if it all started with an accident. I was mesmerized by the beautiful reverb tail hanging as if it were a mist. But despite initially patching in the Darkness, I was simply unable to find the same kind of magic present in that first recording. While trying to fiddle around with Darkness, I discovered that, with shifting soundscapes, timing is everything. Hitting the freeze switch a smidge too early or too late and the capture isn’t what one hoped it might be. Whether too soft or too loud, slightly dissonant or too plain, hitting freeze at just the right moment proved to be more difficult than I originally imagined. So I decided to try a new reverb that I got around the start of the year but hadn’t yet used, the Old Blood Noise Endeavors Dark Star Stereo, to see if I might be able to get better results.

Buying the Dark Star Stereo was a long time coming. OBNE has been making highly compelling effects pedals for a long time. Alpha Haunt, a flexible, nasty fuzz,, was my first foray with them. And even if I sold that pedal for something much more basic, I knew it sounded awesome and that I was totally into the OBNE ethos. Dweller (Dweller!) was my next OBNE purchase, and that thing is ace. A delay circuit inside of a phaser circuit that sounds both unique and beautiful. Next was the Rêver and its sibling, Minim, which are both absolutely brilliant reverse delay/reverb. But as I grew, and especially with modular, I was pretty adamant that my reverbs all be stereo, and despite having wanted to use OBNE reverbs for their unique tones, none of them were stereo. A little while back they released the Dark Light (now discontinued), which is a “stereo” mashup of the Light and Dark Star pedals. But I always felt the implementation was odd, and some sounds were disjointed. It was seemingly more a dual mono reverb than a stereo one, and it just didn’t sound right so much of the time. So I waited. Then a few months backs, OBNE finally released a true stereo reverb, this time a fully featured version of the Dark Star, their most popular reverb pedal (and my favorite of those I’ve heard), and I jumped on one almost immediately. When I bought it I knew I didn’t have room in my pedal rack. But after selling my Oto trio and getting the proper cables to patch it into my synth, it quickly made its way to the synth FX rack, even if I all but ignored it during Jamuary when I didn’t touch it once.

The Dark Star Stereo is a lo-fi reverb, complete with pitch shifting (up or down), filtering, saturation, and sample reduction, along with mix, volume, and stereo spread. It’s designed primarily for soundscape and pad generation, but works great on any source. The default sound without any pitch shifting, filtering, sample deduction or overdrive is soft and gentle. But once you begin to shape the sound it begins to texturize in haunting ways. A bit of crunch to add some dustiness here, some high pass filtering there, and you have a beautiful ambient reverb that can last for days. In this patch I used a smidge of sample reduction and high pass filtering, along with pitch shifting up an octave. Compared to other implementations, the pitch shifting feature is…different. Rather than a cheesy sheen or beautiful choir-like effect, Dark Star Stereo produces more of a granular sounding pitch shift that can be a bit jagged sounding (in good ways), though I do wish that one could control the amount of pitch shifting in the output so that I can more easily get less of it. In this patch the Dark Star Stereo is introduced briefly at about 3:25, a second time around 4:35, then I used it very subtly as a parallel reverb from about 5:18 through the end. The result here was “okay.” It wasn’t exactly what I’d hoped for, but it was a good first foray into using reverbs in this manner., and a step in the right direction for future exploration.

The end-of-chain reverb in this patch is the ever-beautiful Walrus Audio Slöer, with a smidge of its choir-like pitch shifting that is exceptionally beautiful. The Slöer has been my go-to reverb since receiving it, and I simply can’t envision not having one.

Modules Used:
Nonlinearcircuits The Hypster
Nonlinearcircuits Lets Get Fenestrated
Nonlinearcircuits Stochaos
Nonlinearcircuits Triple Sloth
Vostok Instruments Asset
Expert Sleepers Disting NT
Addac System Addac814 6×6 Stereo Matrix Mixer
Qu-Bit Electronix Nautilus
Rossum Electro-Music Panharmonium
Pladask Elektrisk Dradd(s)
Knob Farm Ferry
ST Modular SVCA
Intellijel Quad VCA

Outboard Gear Used:
Echofix EF-X2
Old Blood Noise Endeavors Dark Star Stereo
Walrus Audio Slöer

  1. I also inquired about a gate combiner, to which he responded that would be a good idea, and wrote back a couple of days later with the design for Gator. ↩︎
  2. Chroma is an effect that’s applied to the delay feedback path. Other effects are a LPF, HPF, saturation, wavefolding, and heavy distortion. ↩︎
Posted on

A Spin On Norns Fall – The Evolution Of A Patch

Like all good little electronic ambient musicians, I’ve finally picked up a Monome Norns. Any quick look around the web, and you’ll see Norns is a common tool for creating ambient music, used by musicians from around the world. Norns is a small, open source, community driven music computer that runs scripts written in a language called Lua. There are hundreds, if not over a thousand, Norns scripts that perform any number of musical tasks from sequencing, being a sound source, or processing audio. Overall, it’s a very charming piece of kit that is elegant, minimalist, and can sound beautiful. To me, no script represents Norns better than Fall, a beautiful generative sequencer designed to be a soundscape of falling leaves, and synthesizer made of pure relax-ium, written by Ambalek. It’s simple, elegant (especially when paired with a Grid), and creates absolutely beautiful sounds that will lull you to sleep, or send you on a journey in a bed of peace.

The Init settings on Fall’s sound engine are seemingly simple. A chaotic (or random) process guides virtual leaves falling to the ground. As they hit the ground, sound emerges. A 13-bit filtered square wave is brought in with a gentle rise, and fades out with an even slower decay. This note is sent through two separate stereo delays, the first with a one second delay time, and the second delay with a ten second delay time, all followed up by long reverb. Although I can send midi from Norns to the synth, I wanted to try my hand at patching my own Fall algorithm from scratch. ”This should be pretty simple”, I thought as I started to patch. A few square waves into a filter, followed up by two delays and reverb. “That should be easy enough.”

My first consideration was how I might create a similar cadence to Fall using the tools I have in the case. This all important process determines when notes are played, and I was hopeful to get a reasonable approximation of a representative sample of how often notes are fired in Fall. Fall assigns a note value to each leaf as it’s generated, which can change if the key or scale is changed before a leaf hits the ground. That’s not possible in Eurorack. There are no predetermined leaves here. The only part of that simulation we have in this patch is once these proverbial leaves hit the ground and are actuated by the gate triggering an envelope. Since it’s not clear to me what sort of process generates the falling leaves in Fall, I decided to use chaos. A leaf falling is a chaotic phenomenon created by any number of nonlinear conditions from wind speed and direction, to humidity and leaf density. In that light, I decided to use Swell Physics, a CV generator based on ocean waves, another chaotic process, and ran the four wave outputs through the Nonlinearcircuits Numberwang in order to create nonlinear, sporadic gates. Increasing the simulation speed of Swell Physics would render “leaves” hitting the ground more often.

A second copy of the four Swell Physics outputs were sent to the Vostok Instruments Asset, a very handy six channel attenuator and offset generator, in order to tune them into a range that was good for pitch changes as to not have pitches too high or too low. Though I was able to accomplish a reasonable range, I wasn’t entirely happy. There are too many notes in a very low register, and they repeat too often. I’ll need to really sit down and further tune this set of CV outputs to get it right.

Though I was pretty happy with the cadence of triggers fairly quickly (I’ve done this sub-patch many times), something wasn’t right. Whether from a bad case of mis-remembering or simply being wrong, I struggled with the envelope that shapes each note, which created all sorts of downstream issues. The first delay was utterly lost, which also lengthened the note in a bad way, which also affected the second delay. It also created way too much dissonance with more newly created notes in this generative sequence. I couldn’t figure out the problem until after several listens. When I went to Norns to check the Init settings, I discovered I was wildly off on the envelope times. I somehow remembered a three second rise with a nine second fall, or thereabouts, while in reality the rise is around one second, with an approximately three second fall. About three times shorter than the settings I used. I’ve no doubt that shortening the envelope will prove to be fruitful in many ways. I used Falistri for all four envelopes, which served fine, but I’ll need to shorten those envelopes in order to get closer to Fall. Falistri has become my “go-to” function generator. Whether for modulation or actuating sounds, I use Falistri in almost every patch. I use Falistri so much that I’m in the process of expanding my Frap Tools case to include a total of four Falistris. The module is fantastic and flexible, and if I could keep only one function generator it would be Falistri.

I used the Humble Audio Quad Operator bank as my four square wave oscillators. Operators one and three were tuned in unison, with operators two and four tuned one octave higher. There’s no good reason why I chose to stagger high and low pitched oscillators, but it doesn’t make much difference in the long run. I tuned each oscillator to each other, but none are tuned to a specific pitch, a mistake I’ll remedy when I tweak the patch. The audio from each channel went to the Frap Tools CUNSA for some low pass filtering, and was enveloped by Falistri in CUNSA’s VCAs.

The delays used were the Venus Instruments Veno-Echo for the short delay and the Vongon Polyphrase for the long delay. I’ve spilled much proverbial ink writing about Veno-Echo. It’s a seriously wonderful delay with a feature set that’s tough to beat. I’ve used it in a goodly chunk of my patches since getting it. But because the envelope of the note was so long it’s barely audible despite being turned up much higher in the mix than the Init setting calls for. The long delay is also “weird” because it too followed too closely to the note played. It needed more space to be effective in the way it is in Fall. I chose the Polyphrase because of its very long delay times. Many delays can do one to three second delays, but finding one that can do full ten second delays without serious degradation is tough. The Polyphrase records up to 22 seconds in its continuous buffer, so very long delays with high fidelity are possible. Degradation in repeats can be lovely, as I was reminded of multiple times during Jamuary while using the Bizarre Jezabel Quarté Mk2, but Fall calls for a more clear set of repeats, not disintegrating fragments of what was once there.

Although I wouldn’t describe Fall as inherently lo-fi sounding, it can definitely get in that territory via its control over bit depth. We can go from crisp 32 bit resolution down to a cascade of distorted, bit-reduced blocks, with the Init settings at a fairly low 13 bits. It’s not chiptune sounding by any means, but there is a certain dustiness to Fall that is charming. Wanting to get my clear filtered square waves closer to the OG Fall sound, I decided to patch in Malgorithm for some light bit and sample reduction. With Fall, the bit depth can be modulated by an internal LFO, for which I substituted a slow, cycling triangle function from the Joranalogue Contour 1 to bring the square wave from lightly bit and sample reduced to clean and back again.

But I didn’t want to just replicate Fall and call it a day. As instructive as replicating a patch or sound can be, I wanted more. I decided on using the Disting NT Poly Multisample with LABS Music Box samples, triggered randomly by the two lowest outputs of the CuteLab Missed Opportunities, using the Average output from Swell Physics as a pitch source via the Disting quantizer. I wasn’t unhappy with the results, not even a little, but I was unhappy when I discovered that things were a bit out of order because the oscillators weren’t tuned to a pitch, meaning the Fall notes and short plucks of the music box were playing in different keys. The registers are different enough that it doesn’t clash too strongly, but it has dissonance where there should be none.

I fed the output of the Disting NT to the Holocene Electronics Non-Linear Memory Machine, via the Addac 814 6×6 Stereo Matrix Mixer, with a fairly slow delay time. Like in a couple of Jamuary patches, I occasionally froze and scanned the buffer for a granular-like effect which I find absolutely charming. The wet and dry signals were mixed in a Knob Farm Hyrlo at about an 45/55 mix and sent to the output mixer for some reverb.

The last voice in this patch was Mutable Instruments Beads, playing grains in reverse at two octaves up. You know; for some sparkle on top. I initially only sent the “Fall” audio to Beads, then as the patch progressed introduced the music box samples to give the patch even more sparkle at the top end. Beads provided a nice flourish to the patch.

This patch interests me enough to want to improve it. Fall is a fascinating script, and I’d love to be able to get closer. I’ll definitely be making the adjustments I’ve identified and giving the patch another shot.

Rev 1

After recording this patch and listening back several times, I had ideas. I wrote furiously in my Notability notebook where I keep all of my synth and patch notes trying to suss out what changes I wanted to make, and the best way to go about making those changes. After a couple of days of writing I identified no fewer than eight different parts of the patch that I wanted to tweak, or at least think about tweaking. No part of the patch was left unexamined, and after writing and pondering I was ready to try again.

After tuning the oscillators to C2 and C3, as opposed to merely tuning them to each other at a random pitch value (it was ~Eb), the first change was to the envelope length. In my first iteration, whether through faulty memory or bad information, the envelope for each note was way too long. Something like three times too long, which caused other problems downstream in the audio chain. After fumbling around with a high level of uncertainty, coupled with the desire to be as close as possible to the Fall settings, I decided to run the envelopes to the Mordax Data in order to get their rise and fall times as close as possible to the Init settings. With this simple change the patch was instantaneously improved, and, as hoped, it did fix several other issues like pitches changing mid-note.1 It’s amazing how such a simple change can make such a huge difference. Although the Init settings in Fall imply that its envelope is a simple AD function, it’s not clear to me whether there is an appreciable curve in one direction or another. I decided to stick with a linear function unless I get some sort of clue or confirmation.

Things weren’t perfect, however, and it wasn’t something I’d noticed until I was deep into the recording: with shorter notes, more of them can be heard more clearly, and it dawned on me that although the cadence of gates hadn’t really changed, there were too many notes. When I had way-too-long envelopes controlling note generation during the original recording, many notes were camouflaged by others. By mid-recording on this revision (or thereabouts) I adjusted the Simulation Speed on Swell Physics to slow it down, and although that had the desired effect of creating fewer notes, it was still too many notes happening too quickly. It wasn’t slow enough, and their tendency to pile up, particularly with more discernible repeats, was a bit distracting, and caused a few problems in the process. Lots of notes with various levels of repeats created dissonance, and even if it wasn’t overwhelming most of the time, it was noticeable too much of the time. At least to me. The cadence wasn’t overwhelmingly fast, but, between the notes and their various echoes, the cadence was just quick enough to still have too many notes floating about simultaneously.

Although I didn’t make any appreciable changes to the delays themselves, the delay times stayed the same, I did add slightly more feedback, and adjusted their levels in the matrix mixer to be closer to that of Fall. In the first recording I had the delays both set at about 3 o’clock on the dial. That’s roughly 75%. In Fall’s Init settings, the delay “Gain” is listed as “.400” and “.600” for the short and long delays respectively. Although it’s not exactly clear just what that means, I’ve interpreted those settings as roughly 40 and 60% of the dry level. Even if I was wildly off in my hypothesis, the results bore positive results. The delays no longer overwhelmed the dry signal, and it added a more somber tenor to the patch, with repeats drifting in the distance on different time scales. Just beautiful.

I briefly considered consolidating this part of the patch down to a mostly Frap Tools affair by switching oscillators from the Quad Operator to using the EOR gate outputs on Falistri as square/pulse wave generators, but it would have required a fairly major overhaul to the patch, including installing a new case which I do not like doing while there’s cables hanging everywhere, and so quickly decided against it. Besides patch consolidation, I’m not convinced that there would have been much difference.

Another problem faced in this iteration is that there were too many Music Box notes too. There was just a tad too much CV going to the Density input on the CuteLab Missed Opportunities, which resulted in too many random triggers. This too wasn’t something I’d noticed until well into the recording, and even when I tried to alter it, I didn’t do enough to save the recording. I also worked on the wet/dry mix between the Music Box samples and Non-Linear Memory Machine, allowing the echoes to be a bit more prominent, emphasizing when the buffer was frozen and being scanned. Even just this simple level adjustment between dry and wet was a drastic improvement to this voice, even if I’d still add yet more in the next revision.

Rev 2

Still not yet content with the recordings I’ve captured, I once again started to identify tweaks to make. Too many “Fall” notes. Too many Music Box notes. Not enough feedback or level from the long delay. Wanting to change something about Beads, but not sure what.

My first challenge was getting the Fall cadence down to something manageable. Enough notes to remain relevant in the patch, while not becoming a distraction. My first idea was to simply lower the Simulation Speed on Swell Physics, and while that had an effect, the effect was too great. I was at a point with far too few notes, oftentimes hearing the long echo repeat two or even three times without another note being produced, leaving the space too sparse. I knew that speeding the simulation up again would lead back to too many notes, so I tried a different solution: using Stackcables to combine gates in Numberwang. The switching in Numberwang’s outputs wouldn’t happen any faster, the input signals controlling Numberwang were still running at the same speed, but more gate outputs were being used to trigger the same number of notes (six as opposed to four). These extra gates worked a treat, allowing notes to be triggered more often, but without being overwhelming.

But after several listens of this revision I’ve come to the conclusion that more notes are paramount. There are too many spots with way too much empty space. I’ll have to decide on whether raising the speed of the simulation, or a couple more Stackcables, or indeed a combination of both, is the best route to make that happen. But I’d like two to three notes every ten seconds or so, not one note every ten to 20 seconds as sometimes happened while recording this revision. At first I had too many notes, now too few. Finding that medium ground is my key concern with the next revision.

Having the desire to somehow vary Beads in this patch, I changed how it would fade in and out of the mix by using the Swell Physics Average output to control the output levels via the excellent Intellijel Amps. Rather than being a constant volume once introduced into the mix, Beads faded in and out along with the flow of the wave simulation that controls most of the patch.

In addition to this change in Beads, I also used the same gate that allows Swell Physics to scan NLMM’s buffer to change the repeats to an octave up for a very nice effect. I very much enjoy scanning NLMM’s buffer. It’s an incredibly playful way of presenting a different side of the audio source. The overt aliasing when pitched (especially up) adds a character that is lively and expressive; bits of audio that crackle like sparks from a fire.

Rev 3

Immediately after hitting Stop in AUM, I thought I had captured the recording I was after. But as so often happens, the memory of a moment can be misleading. While recording, it felt right, but after even the slightest bit of examination, I knew I needed more notes. The recording was simply too sparse: the noise floor being the only thing audible in too many spots, which is a weird thing with multiple overlapping delays. This isn’t to say that every second of every recording needs to be filled with something. John Cage showed us that a recording need not be filled with anything. Silence is a fantastic juxtaposition to sound, but too much silence can leave a listener bored rather than enraptured with anticipation, and that’s the feeling I had after listening to revision 2.

Getting more notes in the Fall pattern was my only real priority in this recording. I’ve been after that perfect cadence the entire time, and I hadn’t yet found it. While studiously examining how the gates fire from Numberwang and its relationship to Swell Physics, I tried everything I could think of. Increasing the wave simulation speed, adding another Stackcable, cursing the gods, but it wasn’t until after a while that I discovered that gate generation started to slow and then stall for a bit when the Swell Physics waves were collectively in a trough, all of them well below 0v. Once the waves started to move upwards again, the gates would fire more or less how I wanted them to fire, but I knew that the collective signals from Swell Physics needed some positive offset to keep them below 0v for much less time.

There’s still a bit too much silence in a couple of spots of Revision 3 while Swell Physics was in a lull, particularly from about 2:17-2:42 of the recording, but after several days of trying to proverbially shove a square peg into a round hole, I finally came to accept that Swell Physics was simply the wrong tool for replicating the falling of leaves. Both are chaotic processes, but they are still fundamentally different from one another in a lot of ways. Falling leaves don’t oscillate, they only go downwards, while ocean waves move both ways. If I want to replicate the falling of leaves in any meaningful way, I need a different algorithm in my patch to get me there, and I think I have an idea, though that will be for another day.

Rev 4 – A Rework

When I was done with Revision 3, I was content to tear the patch down and maybe come back to it another day. But there was a gnawing in the back of my mind. As I was writing what was going to be the conclusion to this post, an idea sprang to mind about how to better cadence my interpretation of Fall.

I thought about what sorts of movement I could create that was in the direction of falling leaves, and immediately thought of a saw wave, or falling ramp. The start of each cycle would be the moment a leaf breaks away from the tree, and the fall time that leaf’s journey to the ground. But I also knew that I couldn’t just have cycling saw waves. That would create an unwanted pattern, so I chose the Addac506 Stochastic Function Generator as the new engine of the Fall portion of this patch. Every time an envelope cycled the Addac506 randomly assigned a new fall time to the function within a pre-determined range, and the End Of Cycle outputs triggered the quantizer and envelope for the Fall note. Each leaf fell at a different rate, just as leaves falling from a tree do. The functions themselves were patched to the CV input of the quantizer, via the Vostok Asset for some attenuation and offset. After un-patching Swell Physics and re-patching the Addac506, I realized I made an absolute rookie mistake.

Normally when I’m creating a polyphonic patch, I keep everything fairly orderly. The first trigger goes to the first envelope generator which goes to the first filter, etc. Having quad modules like Quad Operator, CUNSA, the Addac506 and 508, dual Natural Gates, Sum Mix & Pan, etc. generally make polyphonic patches a much easier proposition. But when an End of Cycle trigger is a patched to a quantizer with that channel’s function as CV input, you will always trigger at 0v (unless there is an offset on the function). Not cool. Every trigger is the same note! To remedy this mistake, I simply switched the functions around in the Vostok Asset, with the Ch 4 and Ch 1 envelope outputs, and the Change 2 and Ch 3 outputs. Once I made the switch, each end of cycle triggered a different envelope, and not its own, resulting once again in random pitches.

With this new engine in place the Addac506 drove the patch at a much better and more evenly distributed cadence. I could change it easily by adjusting the fall time range, and it was much more consistent than when I used Swell Physics. Lulls in note creation disappeared, and overall the patch became much easier to control.

All of the delays and other effects remained the same with the exception of how I froze and scanned the buffer on the Non-Linear Memory Machine. The method was the same, but rather than using Swell Physics gate outputs I used one of the End Of Rise gates on the Addac506.

Conclusion

Though I’m sure there could be many changes made to get a better recording, I’m happy enough with the result of this fifth version to finally move on. My re-imagining of Fall, though not perfect, is a credible stand-in and I’ve found a new way to facilitate patches like it. The Addac506 has turned into something resembling a super-modulator. A very flexible module offering several methods to create different timing systems and control entire patches.

Overall, this was a very fun patch to make. Trying to mimic a sound or patch can be very instructive, and this patch was no exception. The process of revising a patch can be a tedious one. Small changes and decisions often have an impactful outcome.

Modules Used:
Addac System Addac508 Swell Physics
Addac System Addac506 Stochastic Function Generator
Addac System Addac814 6×6 Stereo Matrix Mixer
Nonlinearcircuits Numberwang
Nonlinearcircuits Divide & Conquer
Humble Audio Quad Operator
Frap Tools Falistri
Frap Tools CUNSA
Vostok Instruments Asset
Calsynth uO_C (Quantermain)
Expert Sleepers Disting NT
CuteLab Missed Opportunities
ST Modular Sum Mix & Pan
ST Modular SVCA
Intellijel Amps
Intellijel Quad VCA
Knob Farm Ferry

Outboard Gear Used:
Vongon Polyphrase
Vongon Ultrasheer

  1. These changes didn’t completely alleviate rogue pitch changes, you can hear one in quite literally the first note of the recording, but it greatly reduced them. I’m pretty sure that first untimely pitch change is the only one in the recording. ↩︎
Posted on

A Polyphonic Experiment

During one Jamuary patch, I had the pleasure of using the Alexandernaut Fugue Machine Midi sequencer extraordinaire. I enjoyed it so much that I sought a MIDI > CV converter so that I could bring that particular brand of magic into my synth. Sequencing is easily the most challenging part of eurorack for me. I’m quickly learning that, at least in the immediate term, “battleship” sequencers and me don’t work well together. The options are oftentimes overwhelming and programming them can be a constant exercise in frustration. When you’re trying to play a polyphonic patch, these frustrations compound as the patch gets more and more complex. Of the large sequencers I’ve used René v2 is easily my favorite. I find it to be the most intuitive sequencer I’ve used. The sequencer in the Doboz T12 is also easy to catch on, and the Verbos Voltage Multistage is ultra-simple. But I’ve done nothing but get my teeth kicked in by some of my larger sequencers. Frap Tools USTA, Oxi One, and 4ms Catalyst Sequencer in particular. Perhaps it’s the lack of effort with learning them, or that I simply quit on them too quickly, but nothing kills a moment like manual digging in the midst of a patch. Fugue Machine, even if it’s walled in its own very small box, is but one of many tools that make sequencing much less of a chore, and can help spark the creative drive to explore more advanced sequencing. Fortunately, there are good ways of leveraging MIDI tools in Eurorack.

I searched for a couple of weeks for my ideal MIDI > CV converter. My main consideration was for sequencing polyphonic patches, so it needed to have at least four channels of pitch CV, gates, and velocity. There are several options. One of the more compelling options is the Der Mann Mitt Der Maschine DROID. It’s a CV generating and processing powerhouse that can do almost anything that can be done with CV, including MIDI > CV conversion. I even have a DROID, and it has a killer feature set. If you can program it. It turns out that I can program DROID. It’s not terribly hard, even if I still do it by hand rather than using the GUI tool to create patches. But DROID would take more space, and would only be useful in this capacity, needing all eight CV outputs for pitch and velocity CV. So I kept searching, and finally settled on the relatively new Befaco MIDI Thing V2. I already have and use the Befaco CV Thing CV > MIDI converter and find it good at its job, even if the screen is not fit for middle-aged eyes. It’s only 6hp, and can sit right next to the CV Thing.

One of the MIDI Thing’s features is that it has 12 outputs. Exactly enough for four voices worth of outputs, and what’s more is that the exact configuration for my initial intentions with MIDI > CV conversion is already saved as a preset, or Pre-Def in Befaco-speak, for quick and easy input and output configuration. I simply set the MIDI Thing to “Predef 2: Multi Timbric” in the Global Menu and it automatically set the incoming MIDI channels to 1-4 and preconfigured the outputs. Pretty slick. Since I was already in for a lot of patching, I opted to forego using the velocity outputs. That wasn’t a step I was terribly interested in today. I wanted pitch CV and gates to trigger envelopes, and by golly that’s what I got after spending less than one minute in the MIDI Thing configuration screen.1

Once I had my MIDI routed in the AUM MIDI Matrix, and properly set MIDI ”Predef-ed” in the MIDI Thing, I fired up Fugue Machine. Since I was testing the functionality of the MIDI Thing and how that would work with a modular system, I wasn’t overly worried with an elaborate, or even original, sequence, and just used one of the included presets. Although I had played with the sequence length and transposition while running through the patch before recording it (which was all supremely cool), I simply forgot when I improvised the recording. I was more preoccupied with timing each channel, the levels of each oscillator, and crossfading the ending. As a result, this recording is a repetitive sequence. It doesn’t repeat exactly because of the modulation, but there’s no variation in anything other than oscillator timbre.

For oscillators I chose the always excellent Synthesis Technology E370, with a User-Loaded Wavetable titled NOV that was left from a previous owner. Thanks, dude. It’s pretty outstanding. Tuning each oscillator to unison (in Morph X/Y mode), I ran each of the pitch outputs from the MIDI Thing to the v/oct inputs. I followed that up by using the eight outputs of the Nonlinearcircuits Frisson to modulate both the X and Y parameters of each wavetable for all four oscillators. This constant timbre changing caused by the modulation also causes dramatic volume changes as waves morph in and out of more and less prominent waveforms in the table. I initially wanted something glitchy, and turned Glitch on HIGH with Interpolation Off for each of the four channels, and while it was cool, it didn’t fit at all in with the overall tenor of the sequence itself. The tonality is too “positive”, being in the key of A Lydian, and the sequence too upbeat. I suspect it would work fantastically on slower, more drone-ish material.

After I routed the pitch CV and modulation, I ran the gates to two Frap Tools Falistris for some enveloping. In addition to being enveloped at the Control inputs, three of the four gate outputs were also multed to the Hit input on one of two Rabid Elephant Natural Gates to provide some beautiful pinging of these ever-changing tones coming from the E370. All four channels were processed through Natural Gate, but only three were pinged. The bass note was only enveloped. In this process I made a grievous oversight. While Falistri is a perfectly good tool for this job, particularly the pluckier notes, more defined shaping with a ADSR would have really served the slower voices well. It so happens that I have perhaps the most advanced ADSR generator in Eurorack, but I didn’t even think to use it, and when I did, I was way too deep in the patch to re-patch and reconfigure the envelopes, so I let it go.

To be honest, I was expecting hurdles to cross, but I was surprised when it all just worked. CV did what it’s supposed to do. Gates did what they were supposed to do. Everything was perfectly in tune and on time.

Once the notes were created in their respective Natural Gate, all four outputs went to the ST Modular Sum Mix & Pan. While three of the four channels were panned to mono, the fastest moving and highest pitched voice was being slowly panned in the stereo field. The stereo output was then routed to the Addac814 6×6 Stereo Matrix Mixer, and sent to the Venus Instruments Veno-Echo. To ensure a solid clock for the delay, I used the now-defunct (or at least unavailable in the United States according to the App Store – which would seem weird) CoVariant Clock AUv3 plugin, which converted the MIDI clock in AUM to CV and sent it out an ES-9 output as an analog clock. I’ve never really sought to use a MIDI clock as the master clock outside of the iPad. I’ve certainly never used it as the master clock in a Eurorack patch. But this clock was flawless, likely to due CV being generated directly at the source, minimizing switches and pass-thru cabling or USB MIDI jitter. Hopefully CoVariant remains a working plugin in iOS for a while to come as there are currently no other direct MIDI > CV clock converters on the iPad. Veno-Echo was set at x2 on both sides with similar feedback just shy of noon. I also added a smidge of drive in order to enhance the sample reduction I put in the feedback loop. Veno-Echo, with its cross-feedback and width parameter, really can create an enormous stereo field

And so can the Dradd(s). It’s no secret I’m absolutely smitten by dual Dradd(s). Despite this infatuation, my first instinct was to patch in Beads, but Beads just didn’t really have what I was looking for today. At least I couldn’t find it. But the Dradd(s) did. In fact, I had to decide between two modes which both had something very cool to offer. I ultimately chose the Tape mode because the octave up was too much to resist. I slowly started to fade out the oscillators once the Dradd(s) were at full volume, and allowed its magic to guide the rest of the recording, fading out in a glorious wash of the Rain algorithm on the Walrus Audio Slöer.

Modules Used:
Befaco MIDI Thing V2
Synthesis Technology E370
Frap Tools Falistri
Frap Tools Sapel
Rabid Elephant Natural Gate
Nonlinearcircuits Frisson
ST Modular Sum Mix & Pan
ST Modular SVCA
Addac Systems Addac814 6×6 Stereo Matrix Mixer
Venus Instruments Veno-Echo
Pladask Elektrisk Dradd(s)
Knob Farm Ferry

Outboard Gear Used:
Walrus Audio Slöer

Plugins Used:
Alexandernaut Fugue Machine
CoVariant by Alan Clifton

  1. It should be noted that the MIDI Thing also produced velocity CV, I just didn’t use it. ↩︎
Posted on

Made Noise – Sketch 6

When I turned on my Make Noise synth tonight, I had anticipated some more experimentation with filter wobble. I had both QMMG and QPAS right in front of me, and so far in my brief experimentation, these two filters had been the best at it. And I did do some of that. It was the first thing I did for about 30 minutes. But, as it so often does, the patch sent me in a different direction. It’s still a delicate patch, but not in that kind of way.

This patch won’t get the full patch breakdown, but I’ll lay out the basic framework.

René X Channel > Spectraphon B (stereo outputs) > QPAS (Smile Pass 😁) > X-Pan
Spectraphon B (Sub) > QMMG Ch 2 (LPG) > QMMG Ch 1 (VCA) > X-Pan
The recording starts out with only the QPAS outputs, before bringing in the LPD’d version. Both are crossfaded once the LPG’d version is brought in. The resonance, particularly on QPAS, got a smidge out of hand, but nothing ear screaming or offensive.

René Y channel > Spectraphon A (stereo outputs) > QMMG Ch 3 & 4 (LPF) > X-Pan (Aux)
This is the slower, lower voice that was brought in last. Spectraphon’s A side Focus and Slide were modulated by Wogglebug. Once bright in, this voice was only level controlled by the cutoff of the filter in QMMG. I slowly brought up the envelope amount in Maths to open it further and further.

X-Pan (stereo outputs) > Mimeophon > Output

Maths is doing a lot of work, modulating the filter cutoff of both Both QPAS and QMMG, Radiate on QPAS, controlling the envelope level to the QMMG in LPG mode, as well as the crossfading of the continuous playing sequence and the pinged one. I need to find a better way to activate the crossfader. Since I was using a copy of a Maths Unity output in both of X-Pan’s Crossfade CV inputs, I couldn’t just attenuate it to 0v then introduce the crossfading by turning a knob. So I inserted the cables when I was ready to introduce the voice. You can hear that little fumble as that voice is brought in.

I used a reverb send from my mixer to the Maneco Labs Otterley Reverb for some reverb (duh) as well as a touch of granular treatment.

Modules Used:
Tempi
René Mk2
Spectraphon
QPAS
QMMG
X-Pan
Maths
Function
Wogglebug
Mimeophon
Maneco Labs Otterley Reverb

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

Posted on

A Resonance Wobble Experiment

One of the beautiful things about eurorack is the many happy accidents that we all run into on occasion. Those times when some combination of conditions present at just the right moment seems to produce something magical. You don’t necessarily know what got you there, but nonetheless, here it is and it’s glorious. Though we may not always know exactly what leads to these enigmatic moments of splendor, there are things we can consider when seeking to be able to use those sounds as part of your artistic arsenal. It’s one thing to hap into something beautiful, however you might describe that term, but it’s another thing altogether to reproduce whatever it is you heard to make that magic an intentional part of your sound. To play it, rather than have it fall in your lap.

A couple of days after Jamuary concluded I made a patch on the Make Noise synth that made me stand up and stare. There was a whisper. An oscillator speaking softly into a filter’s ear, quivering as it tried to muster enough courage to get sound out. It wasn’t unlike bowing a string as lightly as possible, or trying to play a wind instrument as quietly as one can. There was a vulnerability in the voice, seemingly lacking the confidence to speak, or like trying to speak when you’re crying and your lips quiver. There was a wobble that was absolutely intoxicating, and I was set on trying to reproduce that wobble.

When I first set out to try and recreate this sound, I first isolated the conditions of the patch I wanted to emulate. Of course oscillators can’t speak softly. They only know one output level, generally speaking. I documented every patch connection and knob setting from this sub-patch. I verified modulation sources and any peculiarities. I thought about this patch a lot, writing extensively in my Notability patch book, and narrowed it down to three factors. At least theoretically.

  1. As low a level going into the filter as possible. On my original patch, I used QPAS as my first filter, and controlled levels with the input VCA knob. I initially did this out of necessity because the other voice in the patch was very quiet by its nature, and when any real level was given to the oscillator going through QPAS it was too loud. I discovered later that night that the lower the level, the better the conditions for interesting wobbles. If your filter does not have input level control, you can use an attenuator or VCA before going to the filter input.
  2. A filter with some fairly aggressive resonance. It need not scream like a Polivoks, but the resonance needs to be pronounced before it goes into self oscillation. If the resonance is non-linear, it’s even better. I’ve also surmised, perhaps errantly, that a vactrol-based filter would be better suited to this job because of the inherent drag and voltage drift of vactrols. The filter creating pronounced wobble in my Make Noise patch was QMMG, a vactrol-based filter, processing a signal that had already gone through QPAS. I could be wrong, but my experiments trying to reproduce the wobble seem to bear out this conclusion. I don’t have a vactrol-based filter in my main synth, but none of the filters I used in my experiments had the same sort of wobble as that produced by the QMMG.
  3. A slow moving modulation signal that moves the cutoff frequency through the fundamental frequency of the note. I’ve found the slower the better, but there are diminishing returns to that proposition. When you add resonance to a filter, you’re creating a small hump in the EQ curve at the cutoff frequency by feeding it back into the filter’s input. When that cutoff frequency intersects and passes through the fundamental frequency of whatever signal you’re passing through it, you get a small wobble. The resonance itself and the modulator’s frequency and shape can alter that wobble some, and can change its character, as do any curves you might put on your modulating signal.

Of course all of this was theoretical, and much of it still is, even if I’ve received some form of verification via Google AI, and tangible signs that I’m on the right track through my subsequent experimentation.1

When I set out to make a patch yesterday, my goal was to first experiment with recreating filter wobble through my main synth where I have a plethora of filters of all sorts. My first thought was to experiment wholly within the Frap Tools Cunsa. I can create sine waves with the first filter and still have three filters with which to experiment, all with normalized patching to make things simple. I was pretty quickly able to create some wobble using my three guideposts listed above, but it was very consistent, and exciting as it was to know I was on the right path, I felt that perhaps the Cunsa was simply too polite a filter to get the best results. Abandoning Cunsa, I next went to the Joranalogue Generate 3 feeding Filter 8, but I never felt like I could get anything close to what I wanted. The cutoff was always too high, and I couldn’t tame the harmonics in a way I wanted. So I switched to a single sine wave from Filter 8 feeding the Bizarre Jezabel Seju Stereo, which was okay, but not special, so I went to the Pkhia, which didn’t work very well. I moved on to the Pkhi Mk3, and had a promising start, but it didn’t progress much. Finally I went to the Blossom, a multi-output filter inspired by the legendary Mannequins Three Sisters, and I heard…something interesting. The wobble was there, and had a bit more character than the simple hump like the rest of the filters. I had found the subject for the rest of the day’s experiments. I spent well over two hours exploring different filters, and of those I tried, a simple sine wave into Blossom was definitely the most compelling. I have other filters that I think are good candidates, namely the Verbos Amp & Tone and Instruo I-ō47, but neither of those cases were in the synth when I turned it on. I’m definitely interested in trying those filters, as both have just the right kind of resonance, I think, to be compelling options.

After I’d finalized a base sound I wanted to use for the rest of the patch, a single sine wave into a resonant low pass filter, I worked up a sequence in C Lydian on the very excellent Doboz T12, and went to work. Like the voice used in my Make Noise patch, this voice would also be completely un-gated, sauntering along, only being level modulated in the filter by the slowly moving function of a cycling Contour 1. The cutoff point is set lower than the lowest fundamental frequency so that there would be times when no notes of the sequence come through. Because Blossom doesn’t have level input control, I ran the output of Filter 8 through an attenuator to initially make the sound as quiet as possible while still being (mostly) audible.

I decided to use a staggered clock. One that is gated by a clock divider, so as to never have continuous repetition. I multed a single x1 clock output from the Sitka Gravity to the Nonlinearcircuits Divide & Conquer in order to create my gate. Because the Gravity is in its infancy, there are several basic things it can’t do. As of now, Gravity’s clock (and sequencer) only outputs triggers, and not gates, so I couldn’t use the duty cycle of a gate output (like those on Pamela’s Pro (and New, and OG) Workout) to gate the x1 trigger that would ultimately go to the clock input on the sequencer. I wanted the clock to start and stop every five beats, and Divide & Conquer was able to provide a gate that enabled that staggered clock for my sequence to follow. The sequence itself is simple. It’s a couple of scale lines going up, with a very low probability (11%) of getting a quantized random pitch within seven semitones (a fifth) of any given step of the sequence. But because the cutoff frequency of the filter goes below the lowest fundamental pitch, the sequence flows in and out and isn’t steady. Notes hold in beautiful ways, and the sequence doesn’t repeat despite being only 16 steps long.

Once through the now occasionally wobbling filter, the audio went straight to the Bizarre Jezabel Mimosa. Mimosa is what I consider to be the most beautiful distortion I’ve heard in any format, short of very high dollar guitar amps. Of course the word beautiful is doing a lot of heavy lifting here, but what I mean is that it can heavily distort something, while still allowing the source to shine. It adds life, even when the dials are pinned. No matter what, you can always let some of the original dry signal through which helps keep shape in the audio regardless of the amount of distortion is applied. From gentle piano notes and sine waves (such as those in today’s patch) to ripping saw clouds Mimosa just does the right things whether using it for some gentle saturation or full on sonic destruction.

In this patch Mimosa started gently, with both the output volume and distortion amount both turned low. As the piece progressed, gain was adjusted upwards at multiple points. The first place was the original audio signal on the way into the Blossom. This allowed its resonance to growl a bit more rather than wobble. Higher input levels into the filter also mean higher output levels from the filter, and Mimosa is very sensitive to input level. Even at the same knob settings, input level is a crucial determinant of the final sound. Lower levels at the input might just have a bit of coloration or slight crunch, whereas loud sounds will rip or soar. It’s part of Mimosa’s magic. After I adjusted the initial input as loud as I dared, I started to slowly raise the output volume and distortion amount on Mimosa, as well as the amount of wet signal. From gently whispering and quivering to finally finding one’s voice to sing, all from nothing but subtle gain changes along the way. From Mimosa, the audio went to a new addition in the synth; the Addac Systems Addac814 6×6 Stereo Matrix Mixer to be distrusted to several effects.

It should be reiterated here that every sound is this patch arises from one single sine wave fed by one simple sequence, with but a single parameter being modulated by a lone triangle function (the filter cutoff). Of course that isn’t the only sound in total. That lone sine wave is repeated all over the place with overlapping delays, looped with four simultaneous digital tape heads and a delay of its own, and reverberated. The first delay, a Bizarre Jezabel Quarté Mk2, mostly added some lovely texture. A bit of a wash of decays in the wake of the melody, its gritty tail disintegrating into nothingness, which added depth and color. Repeats were set to moderately long, with a slow(er) delay time. With the PT2399 delay chips, the longer the delay time, the noisier it will be. The second delay was an Olivia Artz Modular Time Machine, with 4 active taps. It’s clear digital voice echoing the distorted sines near perfectly. Both delays are set to different times which really served to fill out space and maximize this one simple voice with the most basic of sound waves.

As beautiful as this very simple sequence was, I wanted to see if I couldn’t perform some complimentary embellishments, and decided to once again delve into the Cutlasses Instruments Gloop. I’ve only used Gloop a couple of times, but it’s already captured my attention. Some modules take some time to gel with. Despite some sloppy transitioning in my first couple of uses, I immediately took to Gloop. Its interface is (mostly) intuitive, and it’s a capable looper with some very cool tricks. It’s pretty easy to create compelling loops with Gloop. That said, it does have some drawbacks, at least in its current iteration. Though it’s packed with some clever effects that can be eminently useful with a looper, delay, reverb, and a host of tape-related effects like tape degradation, wobble, noise, and saturation, these effects can only be used on one channel or the other, and not both. Though Gloop has two outputs, it’s not really stereo, but dual mono. Each of the four heads can be panned in a stereo fashion and be used in one or both outputs as if it were a stereo signal, but for reasons I don’t understand the effects can only be used in one output at a time. This imbalance can definitely be a problem when trying to create a consistent stereo field. I was hoping to use the degradation effect, wherein the audio degrades as it would on a tape machine with each successive loop, fading out to nothing after a time. In loopers this is generally simulated by constantly low passing the signal at progressively lower cutoff frequencies in order to gradually roll off the highs. It’s a crucial component of Frippertronics, for instance. Allowing a loop to fade to nothing is also a beautiful way to end a track, and unfortunately I can’t do that with Gloop while using both output channels. In lieu of using Gloop to add tape hiss, I was able to add noise to both channels in the mixer via the very excellent DAW Cassette by Klevgrand, but that was an improvised half-measure. I think I can patch a workaround, but it definitely won’t function in quite the same way. A slow moving negative function into a wide open filter cutoff should get me at least part of the way there. Timing would be an issue. How long should this envelope be? What happens if I get to the end of the function, and I haven’t pressed stop on the recorder? Will the cutoff reset to fully open? That would be bad. But those are problems for another day.

I recorded a length of the sequence to Gloop, then while the sequence continued to play configured the four play heads and slowly started to raise the level on the looper, while lowering the level of the continuously sauntering sequence. Though this transition isn’t perfect, it’s much smoother than in tries past. I would use a crossfader like the WMD AXYS to more smoothly move between the two parts, but because the individual voices were being multitracked separately, I crossfaded in the mixer by hand using the Michigan Synth Works XVI Faderbank CV and Midi controller. The first and fourth heads were hard panned left and right at 2x forward and 4x in reverse respectively, while heads two and three were panned in the middle at 1x forward and .5x forward. I manually played the loop size and location within the loop of all four heads until it I manually faded out the hard panned parts before fading out the base melody and its half speed sibling. But not even Gloop was without its own dedicated delay, the ever-excellent Venus Instruments Veno-Echo. I used a x4 output from Gravity, with a /3 clock division set in Veno-Echo, which gave me a dotted eighth note delay, an always interesting pattern.

Both voices were mixed together in AUM and sent to the Walrus Audio Slöer using the Rain algorithm with almost no diffusion, and the clock speed at its slowest, adding to an already textured outcome. A medium long decay and high modulation finish off the track.

Modules Used:
Sitka Gravity
Nonlinearcircuits Divide & Conquer
Doboz T12
Joranalogue Filter 8
Joranalogue Contour 1
Bizarre Jezabel Blossom
Bizarre Jezabel Mimosa
Bizarre Jezabel Quarté Mk2
Addac Systems Addac814 6×6 Stereo Matrix Mixer
Olivia Artz Modular Time Machine
Cutlasses Instruments Gloop
Venus Instruments Veno-Echo
Intellijel Amps
ST Modular SVCA
Knob Farm Ferry

Outboard Gear Used:
Walrus Audio Slöer

Software Used:
Klevgrand DAW Cassette
Toneboosters TB Equalizer 4

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

  1. I put very little stock in the accuracy of AI at this stage in its development. However, it stated the same three conditions I had independently surmised, and so choose to engage in a bit of sweet, sweet confirmation bias. ↩︎
Posted on

Made Noise – Sketch 5

When I finished Jamuary I was resigned to taking time off before patching again. Jamuary was exhausting, and I’m tired. I wasn’t sure how long it might be before I patched again. It turns out the answer was “immediately.” By 10pm on February 1st I was bored, and so decided to put something together on the iPad for fun. I had no intention to record it. I was just playing around with a few things to see how they work. The same thing happened on the 2nd. I started to watch a YouTube video, and decided that I’d rather just do a patch on the synth that was sitting not eight feet from me than watch someone else do one. And so I did.

I knew from the start I wanted it to be in a minimalist vein. Maybe not the dictionary definition of Minimalism as a musical discipline, but still something with not much going on. You know; minimalist.

The patch started with Tempi. The base tempo was ~60bpm:

  • Channel 1 (x1) > Mimeophon
  • Channel 2 (x1) > René X Clock
  • Channel 3 (x2) > René Y Clock
  • Channel 5 (/5) > René X Mod
  • Channel 6 (/7) > René Y Mod

René received the clocks and Mod gates from Tempi on both the X and Y channels. Mod on both channels was set to Start/Stop on the Fun page. When the gate is low at their respective Mod inputs, the sequence moves forward, when it’s high, it stops. Since all four clocks are at different times, there is no continuous repeating pattern, each channel starting and stopping every bar and change. Gates and Access were adjusted for both channels throughout the performance to guard against becoming stagnant. René controlled all four voices in the patch, using just two oscillators, Spectraphon and STO. As far as I know, the scale is in C Lydian, but it sounds like I may have neglected to make the key change between C Major and C Lydian in one (or more 😬) of the channels. It doesn’t happen often, but we get a hint of dissonance on occasion.

The X Channel sent pitch CV to Spectraphon’s A side, and the trigger from the X Gate output to DXG’s Channel 1 Strike input. The Y channel was routed similarly to Spectraphon’s B side and DXG’s Channel 2 respectively. These two oscillators form the first voice, comprised of fairly sparse pings in the DXG. The mixed outputs from Spectraphon A and B were sent to DXG in a way to remain discrete left and right with their separate gate patterns. When you plug something into the Left input on any DXG channel, it normalizes to the Right channel and becomes a mono signal at the output. In order for the Left (mono) input to remain in the left channel only, a dummy cable should be plugged into the Right input. This dummy cable breaks the normalization, and will send audio at the Left channel input to only the Left channel at the output. I plugged Spectraphon’s A side Mix output into DXG’s Left input on Channel 1, along with a dummy cable in the Right input. Spectraphon’s B side Mix output went to DXG’s Right input on Channel 2. This kept the A side pings on the left side of the stereo field, and the B side pings on the right for a delightfully stereo experience of pings and echoes.

Both sides of Spectraphon were tuned to C one octave apart. Spectraphon was modulated in three places, but only moderately. The A Side Focus and Slide were modulated by a Maths envelope and Wogglebug’s Stepped outputs respectively. The B Side Slide was modulated by the Maths OR output. The slight modulation helped to have subtle timbre changes on the pinged notes, some brighter and others darker. Both sides also FM’d each other slightly. The FM Bus Index for both channels were around 8:30 on the knobs. There’s some FM, but not very much at all. Just enough to give notes a kind of bounciness once struck in the LPG. One really nice feature of the Spectraphon’s FM capability is that its sine waves always stay pure in order to avoid the problems associated with cross-modulating oscillators. No matter how much one might FM Side A, its sine wave can still modulate the B side with a clean sine wave rather than one that is FM’d. Most oscillators, once they become carriers, are useless as modulators. Not so with Spectraphon where both oscillators can be both modulator and carrier oscillators at the same time. Very nifty.

René’s Cartesian Channel performed an identical role with STO and QMMG as the X and Y channels with Spectraphon and DXG. I wanted something well above the predominant audio register in the patch. High pitched tings and drips, in the same manner as the Spectraphon pings, only even more sparse. These were designed to be ornamental notes, not the star. The Cartesian Channel CV output sent pitch CV to STO’s v/oct input, and its trigger output went to Channel 2 of QMMG for similar pinging with the Cartesian trigger. The STO’s sine output was used to keep the notes with as soft a texture as I could with pings. One interesting difference between using the QMMG and DXG as a LPG is that QMMG’s decay, at least my QMMG’s decay,1 is noticeably longer on higher pitched notes than the DXG when pinged with a trigger. In the DXG, higher pitched notes are sometimes just barely blips. This sort of behavior is generally expected with just about every LPG because of how they filter the upper harmonics. But through the QMMG, those high notes are seemingly longer. It’s certainly a result of the those juicy QMMG vactrols, and a good argument for keeping vactrol LPGs around, cadmium eaters be damned. These pings in the QMMG were mixed in with the Spectraphon A and B side pings via the Aux input in the DXG.

After I worked up the pinging I was after, I knew I wanted something more, but it had to be complementary and juxtapose itself against the very delicate pings. I was in a stream of consciousness-like trance when building this patch, and so even though I’ve documented all of the final patch connections for the entire patch, I’m not exactly sure what thought process led me to how I was going to fill in the space in a graceful way. A bit of experimentation, some clever routing, and tinkering seemed to be the answer.

I first decided I wanted to use QPAS. I’m not sure how I decided on it being heavy filtering, but I knew I needed the voice to be subtle so as not to overtake light pinging happening in the stereo field. I sent Spectraphon’s A side Sine wave to the L input on QPAS, and the B side Sub output to QPAS’s Right input. The trick was to have both oscillators filtered by QPAS, yet remain separate in the outputs. QPAS essentially became a dual mono filter with shared controls.2 The frequency knob on QPAS, for most of the performance, was moderately low, around 10 o’clock on the knob, though it was being modulated by an unsync’d Wogglebug’s Smooth output, while both Radiate knobs were being modulated by the Woggle output.

QPAS’ Left and Right outputs would become completely separate voices, unfettered by any gating or enveloping, being tamed and shaped only by the filter cutoff(s) and resonance(s) before going straight to the output. Because these outputs weren’t being gated or enveloped, they were always present, moving along with their respective pitch sequences from René, Spectraphon’s A side following the X Channel, and the B side following the Y channel. I remember really liking the sound of the voices and the feel they added, but struggled to find a solution to these sequences droning along overtaking the pings. The answer was simple: only send as much volume to QPAS’ inputs as is absolutely necessary, and allow the resonance to do some of the lifting. It’s a delicate balance between not being audible and drowning out everything else; the output needed to be always present, but delicate enough to not use all of the space in the sonic field.

The Right Low Pass and Smile Pass outputs went straight to X-Pan Channel 2 A and B inputs where they were crossfaded and slowly panned across the stereo field (by the same cycling envelope). This melody carried a mostly present sequence from the Y channel, though quite muffled by the filter and constantly swirling with the crossfading, and smoothed out with resonance, then copious amounts of Mimeophon with Halo. From the time it was introduced, this Y melody is omnipresent, filtered to various degrees, and allowed to drift through the stereo field.

The Left outputs is where the routing became a bit of voodoo. I know what connections led to this sound, but I’m not sure I understand the mechanisms that led to the result. It was a sine wave playing a sequence heavily filtered by QPAS,3 and then very heavily filtered again by QMMG. The Left Low Pass and Smile Pass outputs were routed first to QMMG channels 3 and 4. I tried all four modes, but only the Low Pass mode gave me the specter of a ghost lightly singing in the background, occasionally wavering and trembling as the pitch of the input, slowly moving filter cutoff, and resonance interacted with one another. When the voice was introduced, the filter on QMMG was completely closed, only being modulated by a cycling Maths envelope. Resonance started at about 8:30 on each knob. I slowly added more resonance, then more again, before also slowly raising the cutoff. By the time I hit stop on the recording, both the cutoff and resonance for both channels were at about 1 o’clock on the knobs. It sounds as if it’s a feedback patch, though, outside of the copious amounts of resonance in the QPAS and QMMG signal paths, tamed by the controlled input into QPAS, there is no feedback patching.4 These two outputs from QMMG were crossfaded in X-Pan, so that the sound constantly drifted and resonated in interesting ways. This led to wavering cries that occasionally had a smidge of growl enough to resonate through the Mimeophon in an incredibly beautiful way.

This voice, although the most subtle and delicate, as well as the least present, is by far my favorite part of this patch. It brings the patch to life. It’s one of the coolest sounds I’ve gotten from any patch. When I first heard that sound I stood tall and stared straight at the QMMG as if to ask it to teach me its wizardry. It was the first time I’ve looked at the QMMG as an instrument; as something more than a set VCAs, LPGs, and filters, with a mix output.

All four voices were mixed in the X-Pan, and sent to the Mimeophon for some delay-ification and Halo, the soft noise of the Mimeophon cushioning the edges in subtle ways.

Like so many patches I made during Jamuary, this patch is an open testament to a cohesive Make Noise system being a fluid instrument. It’s an absolute pleasure to play.

One small issue I had, which reinforces my desire for a couple of VCAs than can boost levels before going to the ES-9, was that the recording was ultra-quiet. The pings are very quiet, which necessitated a low volume for everything else in the patch. I needed to add a full 12.5dB in post in order to bring my peaks up to ~ -12dB in AUM.

Modules Used:
Tempi
René Mk2
Spectraphon
STO
QPAS
QMMG
DXG
X-Pan
Maths
Function
Wogglebug
Mimeophon

Improvised and recorded in one take on iPad in AUM via the Expert Sleepers ES-9.

  1. Each QMMG will have a different response because of the natural variability of vactrols. Make Noise does a great job of matching vactrols in an individual unit as closely as possible, but there are (sometimes noticeable) differences in the decay length between different units. This is the same for all vactrol-based LPGs. ↩︎
  2. Definitely a first time using QPAS in this way. ↩︎
  3. Filtering a sine wave is about as pedestrian a job as a filter can do. It’s normally unremarkable as filters thrive on upper harmonics. ↩︎
  4. If the input level was much louder, I’m confident the resonance, particularly in QMMG, would have been screaming. Both filters have a very pronounced resonance that can run away quickly. ↩︎
0:00
0:00