Enrico Serotti

Show Notes

My interest for self generative music sparkled in 1979, when I read "The Alchemical Marriage of Alistair Crompton" by Robert Sheckley. In the book there's a very concise description of a sort of home appliance, a "self expressing musical machine" that composes and plays - on the fly - highly empathic music, based on the user's mood. That was a sci-fi novel, but I took it very seriously. At that time I was a teenager, I played guitar in a no-wave band called Confusional Quartet. Electronic music was still at its analog stage, but in a few years, with the advent of personal computers and midi, everything changed. In 1985, in my graduation thesis titled "La Musica Artificiale" (Artificial Music), I put together and described the various efforts made in the field of self generative music, from the 17th century's musical automata to Xenakis. Being mainly a musician, my poor programming skills didn't allow me to get involved in the production of such a marvelous musical machine. Then, in 2001, Reaktor came out. It's a software used mainly for building synthesizers and effects, but under its hood there's a real programming language that allows the coding of complex logics by connecting modules inside a graphical interface. So I started my own research. The (few) commercially available self generative softwares use big databases of precompiled phrases and rhythms. So the compositional process is a sort of copy and paste, managed at very sophisticated levels, but still copy and paste. This means that you find a lot of recursion in the music generated by these programs. After 30-40 songs, you start noticing repetitions... and get bored. I decided to go for a different approach. Basically in my machines there are data tables that are filled with random numbers each time you generate a song. Then these tables are processed by a set of rules, a grammar that also varies randomly for every new song. The sounds are produced by soft synth modules, samplers and fx, and all their parameters vary randomly, too. The main issue was (and still is) entropy, the balance between order and chaos. With too many rules you have too much recursiveness, with few rules you have high levels of chaos. I really didn't know where all this work was leading me, but at a certain point the machines started to produce something interesting, a strange mixture of easy listening, minimal and ambient music... Something I didn't expect, and I was surprised. In this process I mainly compose by listening. It may sound strange, but it's what I do. I listen critically to the flux of self-generated pieces, trying to find the weak spots, then change some rules, listen again... and when I hear something interesting I record the output, to keep that particular moment. Even if I haven't really composed that bit of music, I feel it mine in some ways. Hopefully, in a future stage one of these buggy machines will become a better entertainer... in the meantime, I have selected a series of self-generated songs, the documentation of this work in progress, humbly dedicated to Robert Sheckley. Enrico Serotti, 2008-2015