hey ben - david from dolores labs here<p>I've wanted to build something kind of similar to this - evented simulation of "soundscapes" (coffeeshops, train stations, etc) programmable using a similar graph formalism to yours, except with a bunch of concurrent loops running that probabalistically trigger one or more downstream audio samples. Goal would be to create an audible environment that "sounds like" many of the places I sit, but with enough variance that it doesn't sound like a loop. I estimate it would take a few hundred samples to do this.<p>Let me know if you want to sit around and hack someday, maybe I'll open a PR.