A new approach to onset detection: towards an empirical grounding of theoretical and speculative ideologies of musical performance

This article assesses aspects of the current state of a project which aims, with the help of computers and computer software, to segment soundfiles of vocal melodies into their component notes, identifying precisely when the onset of each note occurs, and then tracking the pitch trajectory of each note, especially in melodies employing a variety of non-standard temperaments, in which musical intervals smaller than 100 cents are ubiquitous. From there, we may proceed further, to describe many other “micro-features” of each of the notes, but for now our focus is on the onset times and pitch trajectories