Can MusicGen Create Training Data for MIR Tasks?
Abstract
We are investigating the broader concept of using AI-based generative music systems to generate training data for Music Information Retrieval (MIR) tasks. To kick off this line of work, we ran an initial experiment in which we trained a genre classifier on a fully artificial music dataset created with MusicGen. We constructed over 50 000 genre- conditioned textual descriptions and generated a collection of music excerpts that covers five musical genres. Our preliminary results show that the proposed model can learn genre-specific characteristics from artificial music tracks that generalise well to real-world music recordings.
Cite
@article{arxiv.2311.09094,
title = {Can MusicGen Create Training Data for MIR Tasks?},
author = {Nadine Kroher and Helena Cuesta and Aggelos Pikrakis},
journal= {arXiv preprint arXiv:2311.09094},
year = {2023}
}
Comments
This is an extended abstract presented at the Late-Breaking / Demo Session of the International Society for Music Information Retrieval Conference (ISMIR) 2023 (Milan, Italy)