ivan@161: function reconstructed=AudioDenoise_reconstruct(y, Problem)
ivan@161: %%  Audio denoising Problem reconstruction function
ivan@161: %   
ivan@161: %   This reconstruction function is using sparse representation y 
ivan@161: %   in dictionary Problem.A to reconstruct denoised audio.
ivan@161: %   The output structre has following fields:
ivan@161: %       audio   - denoised audio signal
ivan@161: %       psnr    - psnr of the reconstructed audio signal
ivan@161: %
ivan@161: %   [1] I. Damnjanovic, M. E. P. Davies, and M. P. Plumbley "SMALLbox - an 
ivan@161: %   evaluation framework for sparse representations and dictionary 
ivan@161: %   learning algorithms," V. Vigneron et al. (Eds.): LVA/ICA 2010, 
ivan@161: %   Springer-Verlag, Berlin, Germany, LNCS 6365, pp. 418-425
ivan@161: 
ivan@161: %
ivan@161: %   Centre for Digital Music, Queen Mary, University of London.
ivan@161: %   This file copyright 2011 Ivan Damnjanovic.
ivan@161: %
ivan@161: %   This program is free software; you can redistribute it and/or
ivan@161: %   modify it under the terms of the GNU General Public License as
ivan@161: %   published by the Free Software Foundation; either version 2 of the
ivan@161: %   License, or (at your option) any later version.  See the file
ivan@161: %   COPYING included with this distribution for more information.
ivan@161: %%
ivan@161: 
ivan@161: windowSize = Problem.windowSize;
ivan@161: overlap = Problem.overlap;
ivan@161: ws = Problem.ws(windowSize);
ivan@161: wa = Problem.wa(windowSize);
ivan@161: 
ivan@161: A = Problem.A;
ivan@161: 
ivan@161: orig   = Problem.Original;
ivan@161: noisy  = Problem.Noisy;
ivan@161: 
ivan@161: 
ivan@161: % reconstruct audio frames
ivan@161: 
ivan@161: xFrames = diag(ws)*(A*y);
ivan@161: wNormFrames = (ws.*wa)'*ones(1,size(xFrames,2));
ivan@161: 
ivan@161: %   overlap and add
ivan@161: 
ivan@161: rec   = col2imstep(xFrames, size(noisy), [windowSize 1], [windowSize*overlap 1]);
ivan@161: wNorm = col2imstep(wNormFrames, size(noisy), [windowSize 1], [windowSize*overlap 1]); 
ivan@161: wNorm(find(wNorm==0)) = 1; 
ivan@161: recN  = rec./wNorm;
ivan@161: 
ivan@161: %% output structure image+psnr %%
ivan@161: reconstructed.audio  = recN;
ivan@161: reconstructed.psnr = 20*log10(sqrt(numel(orig)) / norm(orig - reconstructed.audio));
ivan@161: 
ivan@161: end