In order to run the separation baseline, please :

- Download and extract voiceHome-2 corpus (https://doi.org/10.5281/zenodo.1252143)
- Download and install FASST toolbox (https://gitlab.inria.fr/bass-db/fasst)
- Run main_compute_est_loc.m to estimate DOAs (see localization baseline)

Then :

1) Run main_compute_speaker_models to compute clean speaker models
2) Run main_separation to compute the separation baseline on the noisy part of the corpus

Please refer to function headers for more details.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Copyright 2017 Ewen Camberlein and Romain Lebarbenchon
% This software is distributed under the terms of the GNU Public License
% version 3 (http://www.gnu.org/licenses/gpl.txt)
% If you find it useful, please cite the following reference:
% - Nancy Bertin, Ewen Camberlein, Romain Lebarbenchon, Emmanuel Vincent,
%   Sunit Sivasankaran, Irina Illina, Frédéric Bimbot 
%   "VoiceHome-2, an extended corpus for multichannelspeech processing in
%    real homes", submitted to Speech Communication, Elsevier, 2017

% Contact : nancy.bertin[at]irisa.fr

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%