Basics of Making Sound - Week 3 Fall 2022 MUS 499C - Intro to SuperCollider 

Good question. Yes, Limiter can be used as a wrapper, assuming the iteration block returns the signal to be limited ('collect' returns the results of its function, 'do' merely returns its receiver). Reducing the number of iterations is an option, but if you don't want your iteration count to change, you can instead divide the summed signal by the iteration count. In most cases, this is a good starting point for producing a nominal-level signal, but some tweaking may be necessary, depending on what's inside your iteration function. If I'm not working in a live context, my preference is to "do the math," i.e., divide the signal by the iteration count: ( { var sig, iter = 10; sig = iter.collect{ SinOsc.ar( freq: {ExpRand(200, 2000)} ! 2, phase: {Rand(0, 2pi)} ! 2 ); }.sum; sig = sig / iter; // compensate }.play; ) Note that some UGens, like Splay, have amplitude level compensation built into them. The Limiter option would look like this: ( { var sig, iter = 10; sig = iter.collect{ SinOsc.ar( freq: {ExpRand(200, 2000)} ! 2, phase: {Rand(0, 2pi)} ! 2 ); }.sum; sig = Limiter.ar(sig, -1.dbamp); }.play; ) There are also a few different quarks that automate this process. I've used StageLimiter from time to time: run Quarks.gui, check the box for BatLib, recompile the class library, and then run StageLimiter.activate; after the server is booted. StageLimiter and similar quarks are a great option if your iteration block generates multiple _Synths_ instead of multiple UGens summed together in a single Synth, in which case, a single Limiter won't necessarily get the job done.

I don't have strong feelings in any particular direction. I guess I'm a little curious about what happens next. The SC community has already been talking about this, to some extent: scsynth.org/t/how-will-chatgpt-affect-this-online-community/6999 I imagine generative AI will gradually get folded into our everyday tools, to be utilized by artists/creators in new and unexpected ways. Personally, I enjoy the pursuit of making sound art through creative coding - I wouldn't want an AI to do it for me, because then it wouldn't really be "mine." There are other moments in history where new technology threatened to upend the foundations of an artistic discipline, like the advent of magnetic tape recording. Suddenly, we had the ability to cut, splice, and "frankenstein" recordings from multiple takes. To some people, this was a little horrifying, because recordings would no longer be equivalent to the "authentic" experience of being at a live performance. But, time passed, it became the industry standard, fostered some weird and excited experimental practices, and now tape is sort of an older relic. So maybe something similar will happen with AI, or maybe not. I'm not that surprised ChatGPT was able to make a _basic_ distortion synth (there are plenty of examples out there). Did it make a _good_ one?

@@elifieldsteel I didn't test it out I was just playing around with it, I was mainly surprised by how quick it was able to generate the code with a basic prompt. I agree with your second point, as a writer I wouldn't feel 'spiritually connected' to my work if I just had some AI system write things out for me especially if the stories are personal. When it comes to the art community, I hope people continue to pursue their craft for the joy of the pursuit and not let new technologies discourage them. In times like this, having a healthy foundational philosophy is as important as ever.