Distributing Generative Music With Alternator

Abstract

Computers are a powerful technology for music playback: as general-purpose computing machines with capabilities beyond the fixed-recording playback devices of the past, they can play generative music with multiple outcomes or computational compositions that are not fully determined until they are played. However, there is no suitable platform for distributing generative music while preserving the spaces of possible outputs. This absence hinders composers’ and listeners’ access to the possibilities of computational playback. In this paper, the authors address the problem of distributing generative music. They present a) a dynamic format for bundling computational compositions with static assets in self-contained packages and b) a music player for finding, fetching, and playing/executing these compositions. These tools are built for generality to support a variety of approaches to making music with code and remain language-agnostic. The authors take advantage of WebAssembly and related tools to enable the use of general-purpose languages such as C, Rust, JavaScript, and Python and audio languages such as Pure Data, RTcmix, Csound, and ChucK. They use AudioWorklets and Web Workers to enable scalable distribution via client-side playback. And they present the user with a music player interface that aims to be familiar while exposing the possibilities of generative music.