Department of
Computational
Perception
|
Audio Fingerprinting Data Sets
The main data set that I use in my thesis (and in the ISMIR 2016 paper) consists of 428722 tracks
from the Jamendo corpus.
The data set can be reproduced by downloading the
reference
track IDs from the Jamendo service.
Our query track IDs can be obtained here.
An example on how to construct individual download URLs is given in the top two lines of the text
file contained in the reference .zip.
In addition, I make use of the so-called mixotic-set, that we described in
R. Sonnleitner, A. Arzt and G. Widmer.
Landmark-based Audio Fingerprinting for DJ Mix Monitoring
.
In Proceedings of the 17th
International Society for Music Information Retrieval Conference (ISMIR),
New York, USA, 2016.
Please cite this paper if you make use of this data set in your work.
The mixotic data set consists of 10 DJ mixes and 723 reference tracks.
Download the mixotic data set: