The dsd100 is a dataset of 100 full lengths of music tracks of different styles along with their isolated drums, bass, vocals, and other stems.
dsd100 contains two folders, a folder with a training set: "train", composed of 50 songs, and a folder with a test set: "test", composed of 50 songs. Supervised approaches should be trained on the training set and tested on both sets.
For each file, the mixture corresponds to the sum of all the signals. All signals are stereophonic and encoded at 44.1kHz.
Variants: DSD100
This dataset is used in 1 benchmark:
| Task | Model | Paper | Date |
|---|---|---|---|
| Audio Super-Resolution | U-Net and ResNet | On Filter Generalization for Music … | 2020-11-14 |
Recent papers with results on this dataset: