Implementation of the algorithm in: 

Aggelos Pikrakis, Yannis Kopsinis, Nadine Kroher, Jose Miguel Diaz-Banez, 
"Unsupervised Singing Voice Detection Using Dictionary Learning", 
24th European Signal Processing Conference (EUSIPCO), Budapest, Hungary, 2016.

If the code is useful for you, please cite the above paper.

The software is provided by the authors for research purposes only, in the hope that it will be useful, but without any warranty; without even the implied warranty of merchantability or fitness for a particular purpose.

Note that, the method was developed for audio recordings where the accompaniment of the singing voice exhibits 
consistent behaviour throughout the recording. If you try it on arbitrary instrumentation, it will, most likely return 
results of low recall and high precision. 

To use the package:
a) Install the K-SVD implementation from 
http://www.cs.technion.ac.il/~ronrubin/software.html
b) Read the instructions in file mainFunction.m

Best wishes,

Aggelos Pikrakis