metadata

license: cc-by-nc-nd-4.0
task_categories:
  - audio-classification
  - image-classification
language:
  - zh
  - en
tags:
  - music
  - art
pretty_name: Music Genre Dataset
size_categories:
  - 10K<n<100K
viewer: false

Dataset Card for Music Genre

The raw dataset comprises approximately 1,700 musical pieces in .mp3 format, sourced from the NetEase music. The lengths of these pieces range from 270 to 300 seconds. All are sampled at the rate of 22,050 Hz. As the website providing the audio music includes style labels for the downloaded music, there are no specific annotators involved. Validation is achieved concurrently with the downloading process. They are categorized into a total of 16 genres.

Viewer

https://www.modelscope.cn/datasets/ccmusic-database/music_genre/dataPeview

Dataset Structure

Raw Subset

audio(.wav, 22050Hz)	mel(spectrogram, .jpg, 22050Hz)	fst_level_label(2-class)	sec_level_label(9-class)	thr_level_label(16-class)
		1_Classic / 2_Non_classic	3_Symphony / 4_Opera / 5_Solo / 6_Chamber / 7_Pop / 8_Dance_and_house / 9_Indie / 10_Soul_or_r_and_b / 11_Rock	3_Symphony / 4_Opera / 5_Solo / 6_Chamber / 12_Pop_vocal_ballad / 13_Adult_contemporary / 14_Teen_pop / 15_Contemporary_dance_pop / 16_Dance_pop / 17_Classic_indie_pop / 18_Chamber_cabaret_and_art_pop / 10_Soul_or_r_and_b / 19_Adult_alternative_rock / 20_Uplifting_anthemic_rock / 21_Soft_rock / 22_Acoustic_pop
...	...	...	...	...

Eval Subset

mel(.jpg, 11.4s, 48000Hz)	cqt(.jpg, 11.4s, 48000Hz)	chroma(.jpg, 11.4s, 48000Hz)	fst_level_label(2-class)	sec_level_label(9-class)	thr_level_label(16-class)
			1_Classic / 2_Non_classic	3_Symphony / 4_Opera / 5_Solo / 6_Chamber / 7_Pop / 8_Dance_and_house / 9_Indie / 10_Soul_or_r_and_b / 11_Rock	3_Symphony / 4_Opera / 5_Solo / 6_Chamber / 12_Pop_vocal_ballad / 13_Adult_contemporary / 14_Teen_pop / 15_Contemporary_dance_pop / 16_Dance_pop / 17_Classic_indie_pop / 18_Chamber_cabaret_and_art_pop / 10_Soul_or_r_and_b / 19_Adult_alternative_rock / 20_Uplifting_anthemic_rock / 21_Soft_rock / 22_Acoustic_pop
...	...	...	...	...	...

Data Instances

.zip(.jpg)

Data Fields

1_Classic
    3_Symphony
    4_Opera
    5_Solo
    6_Chamber

2_Non_classic
    7_Pop
        12_Pop_vocal_ballad
        13_Adult_contemporary
        14_Teen_pop

    8_Dance_and_house
        15_Contemporary_dance_pop
        16_Dance_pop

    9_Indie
        17_Classic_indie_pop
        18_Chamber_cabaret_and_art_pop

    10_Soul_or_RnB

    11_Rock
        19_Adult_alternative_rock
        20_Uplifting_anthemic_rock
        21_Soft_rock
        22_Acoustic_pop

Data Splits

Split	Raw	Eval
total	1713	36375
train(80%)	1370	29100
validation(10%)	171	3637
test(10%)	172	3638

Dataset Description

Homepage: https://ccmusic-database.github.io
Repository: https://huggingface.co/datasets/ccmusic-database/music_genre
Paper: https://doi.org/10.5281/zenodo.5676893
Leaderboard: https://www.modelscope.cn/datasets/ccmusic-database/music_genre
Point of Contact: https://huggingface.co/ccmusic-database/music_genre

Dataset Summary

This database contains about 1700 musical pieces (.mp3 format) with lengths of 270-300s that are divided into 17 genres in total.

Supported Tasks and Leaderboards

Audio classification

Languages

Multilingual

Maintenance

GIT_LFS_SKIP_SMUDGE=1 git clone git@hf.co:datasets/ccmusic-database/music_genre
cd music_genre

Usage

Raw Subset

from datasets import load_dataset

dataset = load_dataset("ccmusic-database/music_genre", name="default")
for item in ds["train"]:
    print(item)

for item in ds["validation"]:
    print(item)

for item in ds["test"]:
    print(item)

Eval Subset

from datasets import load_dataset

dataset = load_dataset("ccmusic-database/music_genre", name="eval")
for item in ds["train"]:
    print(item)

for item in ds["validation"]:
    print(item)

for item in ds["test"]:
    print(item)

Dataset Creation

Curation Rationale

Promoting the development of AI in the music industry

Source Data

Initial Data Collection and Normalization

Zhaorui Liu, Monan Zhou

Who are the source language producers?

Composers of the songs in the dataset

Annotations

Annotation process

Students collected about 1700 musical pieces (.mp3 format) with lengths of 270-300s divided into 17 genres in total.

Who are the annotators?

Students from CCMUSIC

Personal and Sensitive Information

Due to copyright issues with the original music, only spectrograms are provided in the dataset.

Considerations for Using the Data

Social Impact of Dataset

Promoting the development of AI in the music industry

Discussion of Biases

Most are English songs

Other Known Limitations

Samples are not balanced enough

Additional Information

Dataset Curators

Zijin Li

Evaluation

https://huggingface.co/ccmusic-database/music_genre

Citation Information

@dataset{zhaorui_liu_2021_5676893,
  author       = {Monan Zhou, Shenyang Xu, Zhaorui Liu, Zhaowen Wang, Feng Yu, Wei Li and Baoqiang Han},
  title        = {CCMusic: an Open and Diverse Database for Chinese and General Music Information Retrieval Research},
  month        = {mar},
  year         = {2024},
  publisher    = {HuggingFace},
  version      = {1.2},
  url          = {https://huggingface.co/ccmusic-database}
}

Contributions

Provide a dataset for music genre classification