Skip to content

Dataset · ccmusic-database

300 pop songs with musical structure annotations

ccmusic-database/song_structure

A multimodal dataset of 300 tracks with song structure annotations by musicologists.

What's inside

The dataset includes 300 pop compositions in .mp3 format from NetEase Music and text annotations of their structure (intro and other sections) in .txt format. The annotations were created by a professional musician and professor at the China Conservatory of Music. The sample features 208 Chinese, 87 English, 3 Korean, and 2 Japanese songs.

What you can do with it

  • Training music segmentation and track structure recognition algorithms
  • Multimodal analysis of audio and text metadata
  • Benchmarking automated music composition analysis systems
Tasks
Time series forecasting
Languages
English
Size
up to 1K rows
Downloads this month
145
Likes
36
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("ccmusic-database/song_structure")