The Latin Music Database (LMD)

The Latin Music Database was developed when I was still a Masters' by Research student at the Graduate Program of Computer Science (PPGIA) at Pontifical Catholic University of Parana (PUCPR), in Curitiba, Brazil.

The files avaiable in this page are exactly the same as they were made available back in 2007 in my personal student homepage using the Marsyas framework for feature extraction.

Due to copyright issues, only the feature vectors of the files are available in the weka compatible "arff" format. In total there are 3.160 songs classified in one of the following 10 music genres: Tango, Bolero, Batchata, Salsa, Merengue, Axé, Forró, Sertaneja, Gaúcha or Pagode. More information about the database and the genre classification process can be found in the following paper:

  • C. N. Silla Jr., A. L. Koerich & C. A. A. Kaestner. The Latin Music Database. 9th International Conference on Music Information Retrieval (ISMIR), Philadelphia, PA, USA. pp.451-456, September 2008. [PDF]

The following files represent the features extracted from different parts of the music:

    • Feature Vectors extracted from the initial (30 seconds) of the music pieces: LMD-Begin.zip

    • Feature Vectors extracted from the middle (30 seconds) of the music pieces: LMD-Middle.zip

    • Feature Vectors extracted from the end (30 seconds) of the music pieces:: LMD-End.zip

Note: In our previous experiments we used 10-fold cross-validation and for that reason the files are identificated regarding which fold was used. Also more information concerning the use of three feature vectores instead of only one in our experiments can be found in the following paper:

  • C. N. Silla Jr., C. A. A. Kaestner & A. L. Koerich. Automatic Music Genre Classification Using Ensemble of Classifiers. IEEE International Conference on Systems, Man and Cybernetics (SMC), Montreal, Canada, pp.1687-1692, October 2007. ISSN/ISBN 1424409918. [PDF]

Further information about the Latin Music Database as well as informationg concerning joint research can be obtained at the following e-mail address: silla at ppgia.pucpr.br.