Author: Reed Nessler, Second Year, Chemistry and Mathematics
Amount Requested: $2754
For all that music has been systematized and formalized, its allure remains obscure. Melody in particular tenaciously resists efforts to demystify it, to describe in objective terms the difference between melodic and amelodic note sequences or good and bad melodies. This obscurity surrounding melody likely explains the high acclaim given to composers of beautiful melodies, but it also represents a tempting area for rigorous study, and any demystification accomplished need not demean the artistry of composers. A deeper understanding of their art can only lead to a deeper appreciation.
A radical way of approaching the problem is proposed. A music-generating algorithm will be programmed in light of results from modern psychoacoustic research and then refined according to the empirical sonic properties of its output. The process of creating this algorithm will be a source of knowledge about the nature of melody. Additionally, an algorithm capable of producing perceptually musical note sequences will itself be a work of artistic merit.
The existence of a large amount of scholarship on traditional music theory demonstrates the reasonableness of treating music as a subject of intellectual study. And while an empirical and algorithmic approach deviates from tradition, the goal of cultural neutrality requires precisely this deviation.
As a further means of combating parochiality, a trip to India is planned. This trip will be an efficient and direct way to absorb an exotic musical idiom—namely, Carnatic classical music—and will be a rich source of ideas for algorithm design.
The direct fruits of this study will be an algorithm and some technical knowledge about its tunings. But more generally, the study will contribute to the broader projects of investigating the essence of melody and formulating a mathematical basis for musicality. These benefits, together with the artistic value of the perfected algorithm, are the motivation for performing this work.