Procedural Music Generation with Grammars

Lukas Eibensteiner

Supervisor(s): Martin Ilčík

TU Wien

Abstract: In this work we present an implementation of a generative grammar for procedural composition of music. Music is represented as a sequence of symbols with user-defined parameters, such as duration, pitch, harmony, volume, or active instruments. These sequences are developed by stochastic application of production rules that mutate, clone, or split their input. The generated sentences are converted to an output format (MIDI) using mapping functions. We demonstrate the system's capabilities through the generation of various types of musical structures, including phrase patterns, chords, melodies, and metrical hierarchies.
Keywords: Multimedia
Full text:
Year: 2018