Datasets:
metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 22050
- name: genre
dtype: string
- name: label
dtype:
class_label:
names:
'0': blues
'1': classical
'2': country
'3': disco
'4': hiphop
'5': jazz
'6': metal
'7': pop
'8': reggae
'9': rock
splits:
- name: train
num_bytes: 586664927
num_examples: 443
- name: validation
num_bytes: 260793810
num_examples: 197
- name: test
num_bytes: 383984112
num_examples: 290
download_size: 1230811404
dataset_size: 1231442849
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
task_categories:
- audio-classification
tags:
- audio
- multiclass
- music
GTZAN Music Genre Classification
GTZAN consists of 100 30-second recording excerpts in each of 10 categories, and is the most-used public dataset in music information retrieval (MIR) research. Following Kereliuk et al. (2015), we use the "fault-filtered" partitioning version of GTZAN, which is constructed by hand to include 443/197/290 excerpts. This version of database could be found and downloaded from here.
Citations
@article{kereliuk2015deep,
title={Deep learning and music adversaries},
author={Kereliuk, Corey and Sturm, Bob L and Larsen, Jan},
journal={IEEE Transactions on Multimedia},
volume={17},
number={11},
pages={2059--2071},
year={2015},
publisher={IEEE}
}
@article{sturm2014state,
title={The state of the art ten years after a state of the art: Future research in music information retrieval},
author={Sturm, Bob L},
journal={Journal of new music research},
volume={43},
number={2},
pages={147--172},
year={2014},
publisher={Taylor \& Francis}
}
@article{tzanetakis2002musical,
title={Musical genre classification of audio signals},
author={Tzanetakis, George and Cook, Perry},
journal={IEEE Transactions on speech and audio processing},
volume={10},
number={5},
pages={293--302},
year={2002},
publisher={IEEE}
}