Below you can read the papers of the past Conferences
New Music Concepts 2021
ISBN:
978-88-944350-2-3
Author(s):
Jason Palamara, Scott Deal
Abstract:
This project illuminates the use of and approach to behavior modes in the AVATAR musical improvisation software which employ machine-learning data in different ways, approximating human performance. While the list of possibilities for ways to play a grouping of notes on a given instrument is practically infinite, a much smaller number of behaviors is regularly used by any one performer. With the goal of the AVATAR project being an interconnected system of softwares, which maps to a real-life human model, these behavior modes have been modeled after the improvisation style of the project's primary model, composer-percussionist Scott Deal.
A behavior mode is here defined as "observed tendencies which a given performer returns to with regularity". Once observed and catalogued, it is our hope that these tendencies may be used proscriptively to generate new musical material which adheres closer to the original musical dataset than that generated by a standard Markov model from the same dataset.
Keywords:
Behavior Modes, Machine-Learning, AVATAR, Markov-Chain Generated Music