annotate examples/Automatic Music Transcription/SMALL_AMT_KSVD_Sparsity_test.m @ 128:8e660fd14774 ivand_dev

Feature 186
author Ivan Damnjanovic lnx <ivan.damnjanovic@eecs.qmul.ac.uk>
date Mon, 13 Jun 2011 14:55:45 +0100
parents cbf3521c25eb
children f42aa8bcb82f
rev   line source
ivan@128 1 %% Dictionary Learning for Automatic Music Transcription - KSVD sparsity
ivan@128 2 %% test
ivan@128 3 %
ivan@128 4 % *WARNING!* You should have SPAMS in your search path in order for this
ivan@128 5 % script to work.Due to licensing issues SPAMS can not be automatically
ivan@128 6 % provided in SMALLbox (http://www.di.ens.fr/willow/SPAMS/downloads.html).
idamnjanovic@25 7 %
idamnjanovic@6 8 % This file contains an example of how SMALLbox can be used to test diferent
idamnjanovic@6 9 % dictionary learning techniques in Automatic Music Transcription problem.
idamnjanovic@6 10 % It calls generateAMT_Learning_Problem that will let you to choose midi,
idamnjanovic@6 11 % wave or mat file to be transcribe. If file is midi it will be first
idamnjanovic@6 12 % converted to wave and original midi file will be used for comparison with
idamnjanovic@6 13 % results of dictionary learning and reconstruction.
idamnjanovic@6 14 % The function will generarte the Problem structure that is used to learn
idamnjanovic@6 15 % Problem.p notes spectrograms from training set Problem.b using
idamnjanovic@6 16 % dictionary learning technique defined in DL structure.
idamnjanovic@6 17 %
ivan@128 18
ivan@128 19 %
ivan@128 20 % Centre for Digital Music, Queen Mary, University of London.
ivan@128 21 % This file copyright 2010 Ivan Damnjanovic.
ivan@128 22 %
ivan@128 23 % This program is free software; you can redistribute it and/or
ivan@128 24 % modify it under the terms of the GNU General Public License as
ivan@128 25 % published by the Free Software Foundation; either version 2 of the
ivan@128 26 % License, or (at your option) any later version. See the file
ivan@128 27 % COPYING included with this distribution for more information.
idamnjanovic@6 28 %%
idamnjanovic@6 29
idamnjanovic@6 30 clear;
idamnjanovic@6 31
idamnjanovic@6 32
idamnjanovic@6 33 % Defining Automatic Transcription of Piano tune as Dictionary Learning
idamnjanovic@6 34 % Problem
idamnjanovic@6 35
idamnjanovic@6 36 SMALL.Problem = generateAMT_Learning_Problem();
idamnjanovic@6 37
idamnjanovic@6 38 TPmax=0;
idamnjanovic@6 39
idamnjanovic@6 40 for i=1:10
idamnjanovic@6 41
idamnjanovic@6 42 %%
idamnjanovic@6 43 % Use KSVD Dictionary Learning Algorithm to Learn 88 notes (defined in
idamnjanovic@6 44 % SMALL.Problem.p) using sparsity constrain only
idamnjanovic@6 45
idamnjanovic@6 46 % Initialising Dictionary structure
idamnjanovic@6 47 % Setting Dictionary structure fields (toolbox, name, param, D and time)
idamnjanovic@6 48 % to zero values
idamnjanovic@6 49
idamnjanovic@6 50 SMALL.DL(i)=SMALL_init_DL(i);
idamnjanovic@6 51
idamnjanovic@6 52 % Defining fields needed for dictionary learning
idamnjanovic@6 53
idamnjanovic@6 54 SMALL.DL(i).toolbox = 'KSVD';
idamnjanovic@6 55 SMALL.DL(i).name = 'ksvd';
idamnjanovic@6 56
idamnjanovic@6 57 % Defining the parameters for KSVD
idamnjanovic@6 58 % In this example we are learning 88 atoms in 100 iterations.
idamnjanovic@6 59 % our aim here is to show how individual parameters can be tested in
idamnjanovic@6 60 % the AMT problem. We test ten different values for sparity (Tdata)
idamnjanovic@6 61 % in KSVD algorithm.
idamnjanovic@6 62 % Type help ksvd in MATLAB prompt for more options.
idamnjanovic@6 63 Tdata(i)=i;
idamnjanovic@6 64 SMALL.DL(i).param=struct('Tdata', Tdata(i), 'dictsize', SMALL.Problem.p, 'iternum', 100);
idamnjanovic@6 65
idamnjanovic@6 66 % Learn the dictionary
idamnjanovic@6 67
idamnjanovic@6 68 SMALL.DL(i) = SMALL_learn(SMALL.Problem, SMALL.DL(i));
idamnjanovic@6 69
idamnjanovic@6 70 % Set SMALL.Problem.A dictionary and reconstruction function
idamnjanovic@6 71 % (backward compatiblity with SPARCO: solver structure communicate
idamnjanovic@6 72 % only with Problem structure, ie no direct communication between DL and
idamnjanovic@6 73 % solver structures)
idamnjanovic@6 74
idamnjanovic@6 75 SMALL.Problem.A = SMALL.DL(i).D;
idamnjanovic@6 76 SMALL.Problem.reconstruct = @(x) SMALL_midiGenerate(x, SMALL.Problem);
idamnjanovic@6 77
idamnjanovic@6 78 %%
idamnjanovic@6 79 % Initialising solver structure
idamnjanovic@6 80 % Setting solver structure fields (toolbox, name, param, solution,
idamnjanovic@6 81 % reconstructed and time) to zero values
idamnjanovic@6 82 % As an example, SPAMS (Julien Mairal 2009) implementation of LARS
idamnjanovic@6 83 % algorithm is used for representation of training set in the learned
idamnjanovic@6 84 % dictionary.
idamnjanovic@6 85
idamnjanovic@6 86 SMALL.solver(1)=SMALL_init_solver;
idamnjanovic@6 87
idamnjanovic@6 88 % Defining the parameters needed for sparse representation
idamnjanovic@6 89
idamnjanovic@6 90 SMALL.solver(1).toolbox='SPAMS';
idamnjanovic@6 91 SMALL.solver(1).name='mexLasso';
idamnjanovic@6 92
idamnjanovic@6 93 %%
idamnjanovic@6 94 % Initialising solver structure
idamnjanovic@6 95 % Setting solver structure fields (toolbox, name, param, solution,
idamnjanovic@6 96 % reconstructed and time) to zero values
idamnjanovic@6 97 % As an example, SPAMS (Julien Mairal 2009) implementation of LARS
idamnjanovic@6 98 % algorithm is used for representation of training set in the learned
idamnjanovic@6 99 % dictionary.
idamnjanovic@6 100
idamnjanovic@6 101 SMALL.solver(1).param=struct(...
idamnjanovic@6 102 'lambda', 2,...
idamnjanovic@6 103 'pos', 1,...
idamnjanovic@6 104 'mode', 2);
idamnjanovic@6 105
idamnjanovic@6 106 % Call SMALL_soolve to represent the signal in the given dictionary.
idamnjanovic@6 107 % As a final command SMALL_solve will call above defined reconstruction
idamnjanovic@6 108 % function to reconstruct the training set (Problem.b) in the learned
idamnjanovic@6 109 % dictionary (Problem.A)
idamnjanovic@6 110
idamnjanovic@6 111
idamnjanovic@6 112 SMALL.solver(1)=SMALL_solve(SMALL.Problem, SMALL.solver(1));
idamnjanovic@6 113
idamnjanovic@6 114 %%
idamnjanovic@6 115 % Analysis of the result of automatic music transcription. If groundtruth
idamnjanovic@6 116 % exists, we can compare transcribed notes and original and get usual
idamnjanovic@6 117 % True Positives, False Positives and False Negatives measures.
idamnjanovic@6 118
idamnjanovic@6 119 AMT_res(i) = AMT_analysis(SMALL.Problem, SMALL.solver(1));
idamnjanovic@6 120 if AMT_res(i).TP>TPmax
idamnjanovic@6 121 TPmax=AMT_res(i).TP;
idamnjanovic@6 122 BLmidi=SMALL.solver(1).reconstructed.midi;
idamnjanovic@6 123 max=i;
idamnjanovic@6 124 end
idamnjanovic@6 125 end % end of for loop
idamnjanovic@6 126
idamnjanovic@6 127 %%
idamnjanovic@6 128 % Plot results and save midi files
idamnjanovic@6 129
idamnjanovic@6 130 figAMTbest=SMALL_AMT_plot(SMALL, AMT_res(max));
idamnjanovic@6 131
idamnjanovic@6 132 resFig=figure('Name', 'Automatic Music Transcription KSVD Sparsity TEST');
idamnjanovic@6 133
idamnjanovic@6 134 subplot (3,1,1); plot(Tdata(:), [AMT_res(:).TP], 'ro-');
idamnjanovic@6 135 title('True Positives vs Tdata');
idamnjanovic@6 136
idamnjanovic@6 137 subplot (3,1,2); plot(Tdata(:), [AMT_res(:).FN], 'ro-');
idamnjanovic@6 138 title('False Negatives vs Tdata');
idamnjanovic@6 139
idamnjanovic@6 140 subplot (3,1,3); plot(Tdata(:), [AMT_res(:).FP], 'ro-');
idamnjanovic@6 141 title('False Positives vs Tdata');
idamnjanovic@6 142
idamnjanovic@6 143 FS=filesep;
idamnjanovic@6 144 [pathstr1, name, ext, versn] = fileparts(which('SMALLboxSetup.m'));
idamnjanovic@6 145 cd([pathstr1,FS,'results']);
idamnjanovic@6 146 [filename,pathname] = uiputfile({' *.mid;' },'Save midi');
idamnjanovic@6 147 if filename~=0 writemidi(BLmidi, [pathname,FS,filename]);end
idamnjanovic@6 148 [filename,pathname] = uiputfile({' *.fig;' },'Save figure TP/FN/FP vs Tdata');
idamnjanovic@6 149 if filename~=0 saveas(resFig, [pathname,FS,filename]);end
idamnjanovic@6 150
idamnjanovic@6 151 [filename,pathname] = uiputfile({' *.fig;' },'Save BEST AMT figure');
idamnjanovic@6 152 if filename~=0 saveas(figAMTbest, [pathname,FS,filename]);end
idamnjanovic@6 153