English

FMA: A Dataset For Music Analysis

Sound 2017-09-07 v3 Information Retrieval

Abstract

We introduce the Free Music Archive (FMA), an open and easily accessible dataset suitable for evaluating several tasks in MIR, a field concerned with browsing, searching, and organizing large music collections. The community's growing interest in feature and end-to-end learning is however restrained by the limited availability of large audio datasets. The FMA aims to overcome this hurdle by providing 917 GiB and 343 days of Creative Commons-licensed audio from 106,574 tracks from 16,341 artists and 14,854 albums, arranged in a hierarchical taxonomy of 161 genres. It provides full-length and high-quality audio, pre-computed features, together with track- and user-level metadata, tags, and free-form text such as biographies. We here describe the dataset and how it was created, propose a train/validation/test split and three subsets, discuss some suitable MIR tasks, and evaluate some baselines for genre recognition. Code, data, and usage examples are available at https://github.com/mdeff/fma

Keywords

Cite

@article{arxiv.1612.01840,
  title  = {FMA: A Dataset For Music Analysis},
  author = {Michaël Defferrard and Kirell Benzi and Pierre Vandergheynst and Xavier Bresson},
  journal= {arXiv preprint arXiv:1612.01840},
  year   = {2017}
}

Comments

ISMIR 2017 camera-ready

R2 v1 2026-06-22T17:14:52.813Z