Downloads

Client Downloads

With the discontinuation of AcousticBrainz, we no longer provide client tools to submit data.

If you are interested in computing acoustic features on your own music, you can still download the command-line essentia extractor and run it yourself:

SHA1 sums

Newer versions of the essentia extractor are also available on the essentia website

Data Downloads

2022-07-06: We provide downloadable archives of all submissions made to AcousticBrainz (29,460,584 submissions)

Low-level and High-level json dumps

High-level downloads
Low-level downloads

Dumps are split into 30 archives, each with 1 million data files. Archives are compressed with zstandard compression. Filenames inside the archives are structured such that they will all uncompress into the same location.

Files in each archive are named according to the following structure:

type/mb/i/mbid-n.json
Where type is one of lowlevel or highlevel, mbid is a uuid of a MusicBrainz Recording Identifier, m, b, i and d are the first, second, third and fourth characters of the MusicBrainz Identifier, and n indicates the ordinal submission offset of duplicate data files present for the same MusicBrainz Identifier. There will always be a file with submission offset 0.

The format of the json files in each archive are described in the data page.

Sample json dumps

Sample downloads

The same as the above full dumps, but only containing 100,000 items for small-scale testing.

Low-level feature dumps

Feature downloads

Smaller CSV files containing some basic features that may be useful for some tasks. Split into three different files based on feature type. Each file contains 29,460,584 rows of data.

  • lowlevel: average_loudness, dynamic_complexity, mfcc_zero_mean
  • rhythm: bpm, bpm_histogram_first_peak_bpm_mean, bpm_histogram_first_peak_bpm_median, bpm_histogram_second_peak_bpm_mean, bpm_histogram_second_peak_bpm_median, danceability, onset_rate
  • tonal: key_key, key_scale, tuning_frequency, tuning_equal_tempered_deviation
See the essentia documentation for streaming_extractor_music for a description of what each of these features are.

Pending: Data files for acoustic similarity

2022-07-06: We will provide a downloadable archive of the data files used in the recording similarity API.

Pending: Low-level and High-level dump of deduplicated items

2022-07-06: We will provide new json and feature dumps of the database after de-duplicating to only one instance of each recording MBID (approximately 7 million items)

Pending dumps will be announced here and on the AcousticBrainz forum in the coming weeks.