# mdeff/fma

FMA: A Dataset For Music Analysis

Repository: https://github.com/mdeff/fma
Canonical: https://ross.abutalabs.com/products/fma
Homepage: https://arxiv.org/abs/1612.01840
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: dataset, music-analysis, music-information-retrieval, deep-learning, open-data, open-science, reproducible-research
Last push: 2023-01-05T15:15:38+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3561, "days_push": 1336, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2653, forks 458 (observed 2026-08-28T04:07:07.149833+00:00)

## What it is
FMA is an open dataset of 106,574 Creative Commons-licensed music tracks (917 GiB, 343 days of audio) with metadata, pre-computed features, and a 161-genre hierarchy, plus smaller subsets for benchmarking. It ships with Jupyter notebooks and code for reproducible music information retrieval research, notably genre recognition baselines.

## Use cases
- train a music genre classification model
- benchmark deep learning models on audio
- find a large open music dataset for research
- compare against GTZAN-style genre recognition baselines
- analyze music metadata and tags with pandas
- evaluate music information retrieval algorithms

## When to choose
- you need large-scale, openly licensed full-length audio for MIR or deep learning research
- you want pre-computed features and metadata alongside raw audio
- you need a reproducible benchmark with train/validation/test splits

## When to avoid
- you need commercial or top-40 music
- you only need small labeled clips and GTZAN suffices
- you need streaming-quality audio without large downloads

## Facets
- artifact type: dataset
- maturity: stable
- function: machine-learning, data-science, audio-processing, nlp
- domain: machine-learning, data-science, deep-learning
- platform: python, cross-platform
- tags: music-information-retrieval, audio-dataset, genre-classification, creative-commons, open-data, benchmark, audio

## Member repositories
- mdeff/fma (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:07.149833+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:18:40.003584+00:00, confidence not recorded.
  - readme: https://github.com/mdeff/fma (fetched 2026-08-28T04:07:07.149833+00:00, sha 3d8646626c52)
  - homepage: https://arxiv.org/abs/1612.01840 (fetched 2026-08-29T10:01:33.721649+00:00, sha 4a6a8ec2965a)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T10:01:33.730693+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T10:01:33.734512+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T10:01:33.736319+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T10:01:33.732712+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.