This patch is relatively simple: it’s just two, 3-OP FM voices (mod1 into mod2 into carrier). Each connection is modulated by a sine wave LFO with its rate modulated randomly. The carriers go into low-pass filters, then into a high-pass filter to cut out the gunk in the low-end.
The mod2s in each chain are fixed and tuned by ear for a dissonant/distorted sound. Mod1s are randomly changed each time the sine wave LFO controlling their VCA resets.
The operators all track an akebono scale, which I am not familiar with but seems to be working well for this purpose!
The reason the patch is super simple is that this is, I think, only the third time I’ve used a ghostverb, and I just said, well, there goes the CPU. I’m not the biggest fan of the ghostverb, but it works well in this context.
My albums of generative patches: https://christopherhmjacques.bandcamp.com/releases
Also, I restarted my Patreon, since people have asked a lot about that, if you want to support the stuff I do (overly long livestreams, videos of trees, etc.):