|
FMA: A Dataset For Music Analysis Data Set
Download: Data Folder, Data Set Description
Abstract: FMA features 106,574 tracks and includes song title, album, artist, genres; play counts, favorites, comments; description, biography, tags; together with audio (343 days, 917 GiB) and features.
|
|
Data Set Characteristics: |
Multivariate, Time-Series |
Number of Instances: |
106574 |
Area: |
Computer |
Attribute Characteristics: |
Real |
Number of Attributes: |
518 |
Date Donated |
2017-05-24 |
Associated Tasks: |
Classification, Clustering |
Missing Values? |
N/A |
Number of Web Hits: |
117455 |
Source:
Michaël Defferrard, Kirell Benzi, Pierre Vandergheynst, Xavier Bresson, EPFL LTS2.
Data Set Information:
* Audio track (encoded as mp3) of each of the 106,574 tracks. It is on average 10 millions samples per track.
* Nine audio features (consisting of 518 attributes) for each of the 106,574 tracks.
* Given the metadata, multiple problems can be explored: recommendation, genre recognition, artist identification, year prediction, music annotation, unsupervized categorization.
* The dataset is split into four sizes: small, medium, large, full.
* Please see the paper and the GitHub repository for more information ([Web Link])
Attribute Information:
Nine audio features computed across time and summarized with seven statistics (mean, standard deviation, skew, kurtosis, median, minimum, maximum):
1. Chroma, 84 attributes
2. Tonnetz, 42 attributes
3. Mel Frequency Cepstral Coefficient (MFCC), 140 attributes
4. Spectral centroid, 7 attributes
5. Spectral bandwidth, 7 attributes
6. Spectral contrast, 49 attributes
7. Spectral rolloff, 7 attributes
8. Root Mean Square energy, 7 attributes
9. Zero-crossing rate, 7 attributes
Relevant Papers:
N/A
Citation Request:
Michaël Defferrard, Kirell Benzi, Pierre Vandergheynst, Xavier Bresson. FMA: A Dataset For Music Analysis. [Web Link], 2017.
|
