SMUG: Scientific Music Generator

Music is based on the real world. Composers use their day-to-day lives as inspiration to create rhythm and lyrics. Procedural music generators are capable of creating good quality pieces, and while some already use the world as inspiration, there is still much to be explored in this. We describe a system to generate lyrics and melodies from real-world data, in particular from academic papers. Through this we want to create a playful experience and establish a novel way of generating content (textual and musical) that could be applied to other domains, in particular to games. For melody generation, we present an approach to Markov chains evolution and briefly discuss the advantages and disadvantages of this approach.