Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 260 - Statistics and AI in Music
Type: Topic Contributed
Date/Time: Tuesday, August 4, 2020 : 1:00 PM to 2:50 PM
Sponsor: Royal Statistical Society
Abstract #311009
Title: Statistics and AI in Music
Author(s): Ahmed Elgammal and Mark Gotham*
Companies: Artrendex / Rutgers University and Universität des Saarlandes / Cornell
Keywords: AI music; Beethoven; AI Generation
Abstract:

2020 marks the 250th anniversary of Beethoven’s birth in 1770. Among the many celebrations taking place, one that captured the imagination of the international public and press was a project funded by Deutsche Telekom to create a realization of Beethoven’s fragmentary sketches for a 10th symphony using machine learning. This project brought together a team of music and computer science experts (including the two present authors) to make all the necessary musicological and computational decisions about how to take on this intriguing task.

In this talk, we highlight some key decisions in that process, including: making sense of Beethoven’s scant and fragmentary plans, converting those ideas into a machine-readable format, identifying suitable music generation tasks (extending a sketched melody, adding an accompaniment, … ), identifying and sourcing suitable training materials for each task (depending on the nature of the musical materials), and the curatorial decision-making process over which generations to use. We cast this as a possible prototype for future human-machine interactions, harnessing the processing powers now available to us, while giving human author the final say.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program