Generative Study 1

Published on Nov 10, 2016 voltlife

"An experiment in structured generative patching, bringing some larger-scale structure to a completely self-running patch. There are several distinct "movements" that have subtle differences between them, as well as the melodic and timbral variations that come from the interplay of random and sequenced elements.

The rhythmic and structural core of the track comes from an A143-4 and a Pittsburgh Sequencer. One A143-4 LFO acts as a clock, and another much slower one speeds the clock up and down, from low audio rates down to one beat every couple of seconds. The square out from that second LFO clocks the Sequencer, so that it gets a different value each time the main clock comes down from audio rates. The slow LFO outputs are multed to several other modules, allowing other parameters to depend on the speed of the clock.

The audio core of the patch is a Rubicon/Dixie II/uFold combo, acting as a classic Buchla-esque complex VCO. The Dixie sine FMs the Rubicon, the output of which goes through a VCA. One Maths envelope opens the VCA, with the fall time lengthening as the clock slows. Another exponential Maths envelope drives both the TZFM index and the Symmetry of the Rubicon: this allows the attack to make the most of the richer TZFM while sacrificing tuning, while the tail comes back into tune. The fall time of this envelope is modulated by yet another A143-4 LFO, and depending upon this and the pitch, the Rubicon's sounds vary from bells and chirps to deep drums and woody bass notes.

Dixie's triangle output goes through uFold into Optomix. A division of the main clock goes through Branches into the Optomix strike input, and Branches' probability is modulated by the main clock speed, so that in the slower passages we hear occasional plucks from the wavefolded Dixie. A ViLFO goes through a VCA into the Optomix Control input, and a Pitts Sequencer opens that VCA. This adds slower swells during some "movements": exactly the same audio source as the plucks, but these sound a bit like solo strings.

The melodic content comes from a Moskwa sequence through a VCA, which is CVed by a Verbos Multistage clocked at /8 of the main clock. This varies the range of the melodic CV. The CV is then quantised to a minor scale by an A156, then is distributed by an A185-2 to the Rubicon and Dixie to keep them in tune. Another clock division goes through Branches to Moskwa's direction input, bringing further melodic variation.

Finally, there's a graunchy sci-fi noise that come in during the slowest parts of some movements. This is a Pittsburgh GenXpander's OE output through a Pittsburgh Filter in Bandpass mode into a VCA and then a Wow & Flutter. The VCA is only opened when the main pitch is mostly static, thus giving a textural counterpoint when the main sequence gets too boring. This works by feeding the main pitch CV into an Elby Slope Detector, whose output goes into Function with a slow rise time and sharp fall. The Function output then goes into a Z4000 envelope. Thus, the envelope is only triggered when the pitch has been static for long enough. This envelope then opens the VCA as well as driving the cutoff frequency to create a noisy sweep. Finally, the Pitts Sequencer also controls an A152 which sites between the Slope Detector and Function: when the Sequencer value is beyond a certain point, it cuts off the CV to Function. This means that the noise sweeps can only occur when the string-like Dixie swells are absent or quite, bringing further variation between the movements."