Learning to Recognize Musical Genre from Audio

TL;DR

This paper discusses a challenge on musical genre recognition from audio data, highlighting the task, challenge design, submission statistics, and results to advance research in automatic music genre classification.

Contribution

It introduces a new open-data challenge for musical genre recognition, providing insights into the task setup, data, and evaluation results.

Findings

Challenge attracted diverse submissions

Results demonstrate varying model performances

Insights into effective features for genre recognition

Abstract

We here summarize our experience running a challenge with open data for musical genre recognition. Those notes motivate the task and the challenge design, show some statistics about the submissions, and present the results.