Datasets

MagTag5k Autotagging dataset

A pre-processed version of the original Magnatagatune data set, coping for issues such as duplication, synonymy, etc. The data is here.

Details about the pre-processing applied on the Magtag5k can be found in the paper [Marques G., Domingues M.A., Langlois T., Gouyon F. “Three current issues in music autotagging“. In 12th International Society for Music Information Retrieval Conference (ISMIR 2011), 2011.]

(Audio files for the original Magnatagatune data set can be downloaded from the original URL.)


Beat tracking dataset

From the paper [“Selective Sampling for Beat Tracking Evaluation” Holzapfel A., Davies M. E. P., Zapata J. R., Oliveira J., Gouyon F., IEEE Transactions on Audio, Speech and Language Processing, 20(9), pp.2539-2548, 2012.] This data includes:

  • Beat annotations and supporting meta-data
  • Audio excerpts
  • Beat annotation protocol for tapping to musical pieces using Audacity

You can find the data here.

For more information on this dataset, contact us.

“Groove” datasets

From the ShakeIt project