A recent forum thread asked “what modules wouldn’t you buy?” One of my answers was “a noise source,” because the DAW can produce many different flavors of noise or I can use field recordings etc. Or I could just use the white noise output on my Minibrute 2S, or weird crawling chaotic insectoid noise from Strega’s delay clocked too slowly, or an AM radio. Or I can generate noise in the modular using all kinds of different methods, without using a module specifically for it.
That got me wondering: how many of my modules can be used to produce white noise, or something like it, without assistance from other modules? This is by no means an exhaustive catalog of the possible flavors of noise, but it should demonstrate that there are many, many possibilities.
Starting from the top row of my rig:
Interstellar Radio: This thing is full of noise, none of it white, all of it at least a little bit pitched. This example uses Type 3 with the Error output patched into Carrier CV.
Akemie’s Castle: Strong enough FM modulation (including feedback) approaches white noise — a technique I’ll come back to a lot. With AC, depending on one’s choice of waveforms and ratios one can get pretty close to white noise. In this case I used the 4-op serial algorithm, and patched Osc B’s output into Op 2’s multiplier CV input just to break it up a little bit more thoroughly.
Drezno: I don’t know about the new mk2 version, but the original version of this module is pretty noisy. Here I present two flavors: one with the input unpatched (amplified in the DAW a bit), and then feeding bit 0 back into the analog input.
Odessa: Self-patching FM is the obvious route to take, so instead I patched the Even output to Tension while monitoring the Odd output. This example starts with the Tension knob at noon, turns it counterclockwise a bit to get to white noise, then further to get a nice detuned pitched sound. (Warning, this begins with a really high-pitched squeal that made me wince and one of our dogs bark…)
On the top row, Tallin, Katowice and Koszalin couldn’t be induced into noise on their own. So that’s 4/7 so far.
On the second row:
Harmonic Shift Oscillator: Second output patched to V/OCT for self-FM. Harmonics level at max, and after a couple of seconds of noise, sweeping the Stride downward to quickly traverse other flavors of sound.
Marbles: not designed for audio rate noise generation, but you can still make noise with it. This was the t2 output, with Rate turned way up and the Y output patched to Rate CV.
Stages: this is the regular mode of the final official firmware — no Easter Egg, no third-party additions. The first segment is an LFO at max rate, which is used as the audio source as well as clocking the other 5 segments as a step sequencer which modulates the LFO rate up into audio territory. There’s nothing random going on here but it’s certainly “noisy” in a sense.
Blades: both filters at max resonance, the second used to FM the first (which we’re hearing).
Beads: I’m offering two flavors of noise from Beads. This one uses its internal wavetable oscillator with literally all the knobs just turned up to max. It’s obviously not white noise, but a zero-effort dark ambient soundscape!
And this flavor uses the mode where the R output emits pulses for each seed, and self-patches that to the L input, with the density and pitch randomized.
Rings: The initial “spark” from plugging the a patch cable from one of the outputs to V/OCT (in this example) or Strum is enough to trigger an initial sound, and the feedback keeps it going.
On this row, Zero2, Blinds and Shades were the only non-participants (I’m not counting Beads twice). That puts our score at 10/16 so far.
Row 3:
Peradam: it doesn’t really do anything close to white noise as such, but it’s full of honking growls, bird chirps and other oddities due to it its feedback. Here’s the example, but it doesn’t score a point.
Ana: patching the Box output to In2 and monitoring some of the other outputs (in this case, “VCA”) while carefully turning knobs can produce some noise, but it needs a bit of a boost. We’ll count it.
Alan (and Engima and Morcom expanders): literally has a white noise output, but where’s the challenge in just using that? So for the example I’ll bend the rules a little: feed the noise output through a comparator (Zero2), and clock Alan from that, while monitoring the CV output. (Noise through a comparator is how the “chance” circuit works internally anyway, you just don’t get direct access to the comparator.) But you could also clock it steadily rather than randomly at audio rate, and can also use the outputs of Morcom for other noise textures too.
Mimeophon: There were lots of complaints about noise with the Mimeophon, until the latest firmware update. Personally I didn’t have an issue, but I wouldn’t call it pristine either. With high enough Repeats and Color the noise will build up. Here I’ve also got the right output multed to the Color CV. This recording starts with Repeats at minimum and quickly cranks it up, then lets the sound evolve a while. It’d be good for cymbals I think, maybe with some granular processing…
Spectraphon: once again, strong FM does the trick. In this case I have side A in SAO mode, am feeding the Even output to side B’s V/OCT and turned up the FM index on A. The CV output of B is modulating the FM amount (no attempt to keep it subtle!)
For this row I’ll count Alan and its two expanders Enigma and Morcom as one. That leaves the Mazzatron Mult+Passthru, Univer Inter, A-150, Jones O’Tool+ and Function as the silent partners (Function is totally periodic and won’t generate noise without external help). The score is now 14/26.
Fourth row:
OptX: technically I’m almost always using this module, as it’s the path to my audio interface. It does have a little bit of low-level noise but you have to amplify it a ridiculous amount to hear it. If my math is right, this recording was boosted by at least 72dB, and then normalized for an additional boost.
Ruina Versio: feedback patching (right output to left input, monitor left output) gives a variety of noise possibilities, some of them pretty dramatic! There’s a bit of knob turning here to explore a few.
Lacrima Versio: again, feedback patching is key — along with the chorus option, high Q and high Mod but low Sat.
Ataraxic Iteritas: no feedback or FM needed, there is a Noise knob here. Also linear feedback shift register noise is what the oscillator design was based on in the first place.
Roucha Legio: why not both feedback AND self-FM? Right output is patched to left input and filter frequency. High resonance, moderately high folding.
Kermit mk2: oscillator A has an area of its wavetable that produces obviously digital noise, while oscillator B instead inserts some kind of messed-up version of A in some kind of bitmask operation. But a bit of cross-FM gives us pretty good stereo white noise.
Shapeshifter: Two flavors again. This one is monitoring the Pulse output, set to “RND 2” and sweeping both frequency and ratio (to give it an extra wide range).
The second uses cross-modulation: Osc 1 PMs Osc 2, which FMs Osc 1. Depths start at 0 and are turned up during the recording. Out1 is set to XOR, and both oscs are using “noise” wavetables (which are periodic, so they don’t really generate noise per se without this kind of modulation). We’re monitoring both outputs in stereo.
This leaves Clep Diaz, Natural Gate, Planar² and Gliss as non-noisemakers. (You can get some “scratchiness” from the latter two but it’s not great as a noise source.) Now we’re at 21/37.
In my Pod60 I have:
Just Friends: here outputs 1, 5 and 6 are self-patched to Intone, Ramp and Curve respectively, and we’re monitoring the Mix and 4 outputs in mid/side.
Teletype: again, a module not designed to process or generate audio at all, but you can get the Metronome script running at approximately 2ms. Here I’ve got a script just outputting random voltages on outputs 1 and 2 every 2 or 3 ms (also randomized). It’s closer to brown noise, but we’ll count it!
The other module in the Pod is Sweet Sixteen which won’t produce noise, so our total score for self-noise-producing modules is 23 out of 40.