comparison util/SMALL_AudioDeNoiseResult.m @ 8:33850553b702

(none)

+function SMALL_AudioDeNoiseResult(SMALL)
+
+fMain=figure('Name', sprintf('File %s (training set size- %d, sigma - %d)',SMALL.Problem.name, SMALL.Problem.n, SMALL.Problem.sigma));
+m=size(SMALL.solver,2);
+maxval=SMALL.Problem.maxval;
+au=SMALL.Problem.Original;
+aunoise=SMALL.Problem.Noisy;
+
+subplot(2, m, 1); plot(au/maxval);
+title('Original audio');
+
+subplot(2,m,2); plot(aunoise/maxval);
+title(sprintf('Noisy audio, PSNR = %.2fdB', 20*log10(maxval * sqrt(numel(au)) / norm(au(:)-aunoise(:))) ));
+
+for i=1:m
+    params=SMALL.solver(i).param;
+    sWav=subplot(2, m, m+i, 'Parent', fMain); plot(SMALL.solver(i).reconstructed.Image/maxval, 'Parent', sWav);
+    title(sprintf('%s Denoised audio, PSNR: %.2fdB', SMALL.DL(i).name, SMALL.solver(i).reconstructed.psnr),'Parent', sWav );
+    if strcmpi(SMALL.DL(i).name,'ksvds')
+        D = kron(SMALL.Problem.basedict{2},SMALL.Problem.basedict{1})*SMALL.DL(i).D;
+    else
+        D = SMALL.DL(i).D;
+    end
+    figure('Name', sprintf('%s dictionary in %.2f s', SMALL.DL(i).name, SMALL.DL(i).time));
+    imshow(D*255);
+%     n= size(D,2);
+%     sqrtn=round(sqrt(size(D,2)));
+%     for j=1:n
+%     subplot(sqrtn,sqrtn,j); plot(D(:,j));
+%     end
+%     dictimg = showdict(D,[params.blocksize 1],round(sqrt(size(D,2))),round(sqrt(size(D,2))),'lines','highcontrast');
+%     
+%     subplot(2,m,m+i);imshow(imresize(dictimg,2,'nearest'));
+%     title(sprintf('%s dictionary in %.2f s', SMALL.DL(i-1).name, SMALL.DL(i-1).time));
+    
+end