Data Science Research Collaborative
University of Illinois at Urbana-Champaign

Computer Generated Bells (Altgeld Chimes)

By: Natalia Ozymko, Stephen Taylor, Wade Fagen-Ulmschneider, Karle Flanagan, and Ashley Grudzinski
Released On:

Background

This work is the first released composition of Natalia Ozymko’s senior thesis work on the AI-generation of study music. The work’s goal is to create an algorithm that can generate a new, hour-long piece of music each and every day that could be listened to while studying as background music.

This composition, “Computer Generated Bells”, is the result of training an LSTM model with existing study music broken up into 65,573 sub-sequences of 100 notes. After 200 epochs, the model was used to create a final sequence of notes that was manually cleaned to ensure all notes could be reasonably played on the Altgeld Chimes.

View the full composition on musescore >

Natalia Ozymko’s thesis is advised by a collaboration of professors: Wade Fagen-Ulmscheinder (Computer Science), Stephen Taylor (Music), and Karle Flanagan (Statistics). Ashley Grudzinski works alongside the team as she works with LSTMs on her own senior thesis.