mdeff/fma
A 917 GiB open dataset of 106,574 Creative Commons-licensed music tracks with metadata, pre-computed audio features, and genre taxonomy for MIR and deep learning research.
Not currently ranked — collecting fresh signals.
star history
The Free Music Archive (FMA) dataset provides full-length high-quality audio, pre-computed features, track metadata, tags, and user-level information arranged in a hierarchical taxonomy of 161 genres. It includes code and Jupyter notebooks for data loading and exploration. The dataset was created to overcome the limited availability of large audio datasets for the growing interest in feature-based and end-to-end learning approaches in music information retrieval.
Frequently asked
- What is mdeff/fma?
- A 917 GiB open dataset of 106,574 Creative Commons-licensed music tracks with metadata, pre-computed audio features, and genre taxonomy for MIR and deep learning research.
- Is fma open source?
- Yes — mdeff/fma is open source, released under the MIT license.
- What language is fma written in?
- mdeff/fma is primarily written in Jupyter Notebook.
- How popular is fma?
- mdeff/fma has 2.6k stars on GitHub.
- Where can I find fma?
- mdeff/fma is on GitHub at https://github.com/mdeff/fma.