DREANSS: DRum Event ANnotations for Source Separation
The purpose of the annotations is to help develop research in source separation methods for polyphonic audio music mixtures containing drums.
We provide a dataset that contains annotations for 22 excerpts of songs taken from different multi-track audio datasets publicly available for research purposes. These multi-track excerpts range from several genres including Rock, Reggae, electronic, Indie and Metal. The excerpts have a duration of 10 seconds each in average. This annotations dataset is divided into four folders, each of which contains the annotations of a given audio source separation dataset.
The dataset provides several files for each song extract. One file for each percussive instrument that has been annotated (e.g. Bass drum, snare drum,...)
The files that are exported are ´.txt´ here is an example of the format, where each row corresponds to a drum event containing <time stamp> <label>.
Chinese ride cymbal
Crash cymbal 2
Snare drum brush
Snare rim shot
Timbala rim shot
The software used to carry out the manual annotations was Sonic Visualiser (http://www.sonicvisualiser.org/). Using time instants layers, each part of the drumset was annotated separately. The annotator used the audio tracks corresponding to the drums to perform the annotations.
The dataset annotation was carried out by Jack Kirwan (MTG internship student from http://www.ittralee.ie/en/) under the supervision of Ricard Marxer and Jordi Janer. This work was in the scope of the current research on Audio Source Separation at the MTG.
Please Acknowledge DREANSS in Academic Research
When the DREANSS dataset is used for academic research, we would highly appreciate if scientific publications of works partly based on the DREANSS dataset cite the following publication:
The DREANSS dataset is offered free of charge for internal non-commercial use only. You may not redistribute, publically communicate or modify it. Please see the license terms in the README file within the dataset for applicable conditions.
Problems, positive feedback, negative feedback... it is all welcome! Please help us improve DREANSS by sending your feedback to: [email protected]
In case of a problem report please include as many details as possible.