------------------------------------------------------------------------------------------------------------------
Task: Professionally produced music recordings

Participants: Alexey Ozerov and Emmanuel Vincent

Data processed: All mixtures

Algorithm description:
A particular instance of general source separation framework described 
in [1].
It is implemented exactly as described in [1] (Section 5.C),
where the corresponding algorithm allows separating the 6 following 
components:
     - bass
     - drums
     - 4 harmonic spatial components (melody is selected as predominant 
harmonic component)
Here, we first separate in 6 components, as above, and then manually group
them to create desired components.

Average running time: Our Matlab implementation on 2.2 GHz CPU, runs 8.2 
minutes
                       to process the longest expert (25 seconds)
Time spent by human (to group components manually): 1-2 minutes per expert

URL: 
http://www.irisa.fr/metiss/ozerov/sisec_2011/Ozerov_Professionally_produced_music_recordings.zip

References:

[1] A. Ozerov, E. Vincent and F. Bimbot, "A general flexible framework 
for the handling of prior information in audio source separation,"  IEEE 
Trans. on Audio, Speech and Lang. Proc. (to appear)
------------------------------------------------------------------------------------------------------------------