image
imagewidth (px)
512
512
genre
stringclasses
5 values
song_number
int64
1
203
slice_number
int64
1
119
dance
1
1
dance
1
2
dance
1
3
dance
1
4
dance
1
5
dance
1
6
dance
1
7
dance
1
8
dance
1
9
dance
2
1
dance
2
2
dance
2
3
dance
2
4
dance
2
5
dance
2
6
dance
2
7
dance
2
8
dance
2
9
dance
2
10
dance
2
11
dance
2
12
dance
2
13
dance
2
14
dance
2
15
dance
2
16
dance
2
17
dance
2
18
dance
2
19
dance
2
20
dance
3
1
dance
3
2
dance
3
3
dance
3
4
dance
3
5
dance
3
6
dance
3
7
dance
3
8
dance
3
9
dance
3
10
dance
3
11
dance
3
12
dance
3
13
dance
3
14
dance
3
15
dance
3
16
dance
3
17
dance
3
18
dance
3
19
dance
3
20
dance
4
1
dance
4
2
dance
4
3
dance
4
4
dance
4
5
dance
4
6
dance
4
7
dance
4
8
dance
4
9
dance
4
10
dance
4
11
dance
4
12
dance
4
13
dance
4
14
dance
4
15
dance
4
16
dance
5
1
dance
5
2
dance
5
3
dance
5
4
dance
5
5
dance
5
6
dance
5
7
dance
5
8
dance
5
9
dance
5
10
dance
5
11
dance
5
12
dance
5
13
dance
5
14
dance
5
15
dance
5
16
dance
5
17
dance
5
18
dance
5
19
dance
5
20
dance
5
21
dance
6
1
dance
6
2
dance
6
3
dance
6
4
dance
6
5
dance
6
6
dance
6
7
dance
7
1
dance
7
2
dance
7
3
dance
7
4
dance
7
5
dance
7
6
dance
7
7
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

This dataset contains mel spectrograms of songs from five different genres, each represented as 512x512 shaped images. The dataset is part of my music generation AI project Amuse. Each genre includes approximately 11 hours of audio, making this dataset a comprehensive resource for training and evaluating music generation models.

Downloads last month
5
Edit dataset card