metadata
license: cc-by-nc-sa-4.0
This is a multi-label music emotion recognition dataset that contains more than 16k songs.
The dataset was obtained through web scraping from NetEase Cloud Music, with specific details referenced from the paper ().
The NPZ file contains the MFCC features extracted by librosa from the middle 30 seconds of all songs. The playlist IDs in the JSON file correspond to the song IDs in the CSV file.
For full access to the original mp3 files, please contact geluoge@gmail.com