Application of Genetic Programming to Creative Sound Synthesis in Web Audio

In modern music, sound synthesisers are everywhere. A sound synthesiser (henceforth ‘synthesiser’) is a physical or software device which generates a sound based on some set of input parameters. These devices have an extremely steep learning curve, and often require significant financial investment, so have very low accessibility. In this paper, we present a tool which allows users to creatively generate their own sounds (henceforth ‘patches’) with no prior knowledge required, and with no physical or digital dependencies other than a modern web browser. This is accomplished by applying genetic algorithms to automatically explore the parameter space of a simulated modular synthesiser, with the user selecting their favourite sound on a regular basis. Full control over the parameters is also provided, as well as the option to export each patch in the Faust DSP format, for more experienced users. The intended outcome is to produce a novel, easy-to-use, browser-based synthesiser generation tool that improves upon previous work on this field and allows users of all categories to take part in sound design.