Supported Datasets

Here is a list of the supported datasets.

Dataset

Format

Hours

Songs

Genre

Melody

Chords

Multitrack

Lakh MIDI Dataset

MIDI

>5000

174,533

misc

*

*

*

MAESTRO Dataset

MIDI

201.21

1,282

classical

Wikifonia Lead Sheet Dataset

MusicXML

198.40

6,405

misc

O

O

Essen Folk Song Dataset

ABC

56.62

9,034

folk

O

O

NES Music Database

MIDI

46.11

5,278

game

O

O

MusicNet Dataset

MIDI

30.36

323

classical

*

Hymnal Tune Dataset

MIDI

18.74

1,756

hymn

O

Hymnal Dataset

MIDI

17.50

1,723

hymn

music21’s Corpus

misc

16.86

613

misc

*

*

EMOPIA Dataset

MIDI

10.98

387

pop

Nottingham Database

ABC

10.54

1,036

folk

O

O

music21’s JSBach Corpus

MusicXML

3.46

410

classical

O

JSBach Chorale Dataset

MIDI

3.21

382

classical

O

Haydn Op.20 Dataset

Humdrum

1.26

24

classical

O

(Asterisk marks indicate partial support.)