Fractal of Periodic Musical Elements
a re-taxonomy of music for machine learning
Abstract
The study presents a taxonomy of periodic musical progressions inspired by the properties of Shepard tones, designed for training machine learning models. It is a system, implemented in new computer software, for the synthesis and analysis of music harmony, rhythm, and dynamics in an algorithmic context. In the introduction, motivation is explained for the creation of a new music taxonomy designed for future use with machine learning. In the main part of the text, a new music notation is introduced that enables the representation of any progression of harmony, rhythm, and dynamics in a form of multidimensional arrays. Furthermore, a new taxonomy of Shepard-tone-inspired "Periodic Musical Elements" is introduced and a new method for analysis of musical material is presented together with examples. The paper discusses the benefits of using this method in the perspective of further research connected to algorithmic music composition with machine learning.
