FURI | Fall 2022

VivyNet: A New Approach to Creating AI-Generated Music

Data icon, disabled. Four grey bars arranged like a vertical bar chart.

This FURI semester, this project focuses on experimenting with the concatenation of pre-trained encoder and decoder checkpoints to quickly perform task-specific tasks. Our research has uncovered possible ways that can translate text into poems. This research provides an entryway into further research that can solve task-specific problems that involve mapping data from one type to another. Future work about this research should look into other applications other than text-to-audio tasks to validate the idea of the concatenation of pre-trained checkpoints.

Student researcher

Benjamin Herrera

Benjamin Joseph L. Herrera

Computer science

Hometown: Phoenix, Arizona, United States

Graduation date: Spring 2025