This code solves multichannel sparse linear prediction using the Least Absolute Deviation (LAD) and Group LASSO (GL) algorithms with applications in dereverberation
License: GNU General Public License v3.0
MATLAB 100.00%
groupsparselp_dereverberation's Introduction
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% %
% Speech Dereverberation based on Convex Optimization %
% Algorithms for Group Sparse Linear Prediction %
% %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
This code is designed around solving the following convex optimization
problems
minimize ||X-X_tau*G||_1,1
minimize ||X-X_tau*G||_1,1 + gamma || G ||_1,1
minimize ||X-X_tau*G||_2,1
minimize ||X-X_tau*G||_2,1 + gamma || G ||_1,1
The algorithmic background, motivation, etc. is described in
SPEECH DEREVERBERATION BASED ON CONVEX OPTIMIZATION ALGORITHMS
FOR GROUP SPARSE LINEAR PREDICTION
D. Giacobello, T.L. Jensen
ICASSP2018
If you use this implementation then please give reference to the
above paper.