Using a Human-in-the-Loop Evolutionary Algorithm to Create Data-Driven Music

Lindenmayer-systems (L-systems), a formal grammar of symbols and replacement rules, have been used to model plants and create fractal images and musical output depending on the L-system interpreter used. An L-system variant called a Circular L-system (CL-system) adds the capability of receiving numerical data input and providing a musical response in real time. The project described in this paper uses a CL-system to produce a population of musical outputs (compositions) that are optimized by a human-in-the-loop evolutionary algorithm to create an improved composition. The fitness function for this process is the musical taste of the user guided by several musical principles.