-
Notifications
You must be signed in to change notification settings - Fork 0
/
transit_validation0.m
228 lines (207 loc) · 7.79 KB
/
transit_validation0.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
% Validation
% Given A, B, C and D test sentences, compare the mapping
% A -> X -> B for all sentences X and
% A ------> B
% B -> X -> C for all sentences X,
% B ------> C
% ... (all 12 combinations)
%
clear all;
load FileWav16kHz
datKind = {
'N1W180R0S0EdN0'
'N1W180R0S0EdN1'
'N3W180R0S0EdN1'
'N4W180R0S0EdN1'
'N5W180R0S0EdN1'
};
datroot = '.\MappingTable\GT100old';
Files = dir(fullfile(datroot, datKind{1}));
Files = {Files(3:end).name}; % Remove . and ..
Files = {
% 'english]english10.female.N_english.R_usa.Y35.A35';
% 'english]english165.female.N_english.R_usa.Y43.A43';
% 'english]english208.male.N_english.R_usa.Y29.A46';
% 'english]english239.male.N_english.R_usa.Y18.A19';
% 'english]english325.male.N_english.R_usa.Y32.A32';
% 'english]english376.male.N_english.R_usa.Y46.A46';
'english]english451.male.N_english.R_usa.Y44.A44';
% 'english]english462.male.N_english.R_usa.Y20.A20';
'japanese]japanese4.male.N_japanese.R_usa.Y1.A20';
% 'japanese]japanese26.female.N_japanese.R_usa.Y6.A44';
'spanish]spanish35.male.N_spanish.R_usa.Y4.A28'
% 'mankanya]mankanya1.male.N_mankanya.R_.Y0.A27'
% 'maltese]maltese1.female.N_maltese.R_.Y0.A30'
'russian]russian44.female.N_russian.R_usa.Y10.A29'
% 'malayalam]malayalam3.female.N_malayalam.R_usa.Y15.A19'
'lao]lao3.male.N_lao.R_usa.Y33.A52'
% 'arabic]arabic80.male.N_arabic.R_usa.Y2.A25'
% 'spanish]spanish83.male.N_spanish.R_usa.Y1.A48'
'english]english108.male.N_english.R_uk.Y22.A21'
% 'mandarin]mandarin27.male.N_mandarin.R_usa.Y3.A18'
% 'english]english490.male.N_english.R_new zealand.Y19.A18'
% 'russian]russian9.male.N_russian.R_.Y0.A23'
% % 'russian]russian41.female.N_russian.R_usa.Y10.A30'
% 'mortlockese]mortlockese1.male.N_mortlockese.R_usa.Y3.A21'
% 'ga]ga3.female.N_ga.R_usa.Y1.A60'
% 'english]english40.male.N_english.R_uk.Y53.A53'
% % 'panjabi]punjabi4.female.N_punjabi.R_australia.Y3.A33'
% 'kikuyu]kikuyu3.female.N_kikuyu.R_kenya, usa.Y45.A45'
% % 'lithuanian]lithuanian4.female.N_lithuanian.R_.Y0.A23'
% 'panjabi]punjabi6.male.N_punjabi.R_usa.Y2.A25'
% % 'bulgarian]bulgarian4.male.N_bulgarian.R_usa.Y2.A22'
% 'spanish]spanish46.male.N_spanish.R_usa.Y10.A33'
% % % 'amharic]amharic15.male.N_amharic.R_usa.Y1.A23'
% 'english]english305.female.N_english.R_uk, usa.Y7.A24'
% % % 'dutch]dutch28.male.N_dutch.R_.Y0.A22'
% % 'tagalog]tagalog7.male.N_tagalog.R_australia.Y22.A60'
% 'bahasa indonesia]indonesian6.male.N_indonesian.R_usa.Y3.A20'
% % 'japanese]japanese13.male.N_japanese.R_usa.Y0.A28'
% 'krio]krio5.male.N_krio.R_sierra leone, usa.Y26.A36'
% 'lithuanian]lithuanian5.female.N_lithuanian.R_ireland.Y1.A24'
'spanish]spanish2.male.N_spanish.R_usa.Y1.A20'
% 'arabic]arabic84.female.N_arabic.R_usa.Y2.A26'
% 'serbian]serbian1.female.N_serbian.R_usa.Y1.A47'
% 'mandarin]mandarin57.male.N_mandarin.R_uk.Y0.A20'
% 'english]english138.female.N_english.R_usa.Y25.A25'
% 'hindi]hindi11.male.N_hindi.R_usa.Y7.A20'
% 'mende]mende3.male.N_mende.R_usa.Y3.A32'
% 'turkish]turkish24.male.N_turkish.R_usa.Y10.A31'
% 'farsi]farsi17.female.N_farsi.R_usa.Y3.A27'
'bosnian]bosnian1.male.N_bosnian.R_usa.Y2.A39'
% 'arabic]arabic87.female.N_arabic.R_usa.Y1.A49'
'korean]korean39.male.N_korean.R_usa.Y41.A55'
% 'russian]russian13.male.N_russian.R_.Y0.A26'
'mandarin]mandarin50.male.N_mandarin.R_usa.Y0.A25'
% 'english]english409.female.N_english.R_australia.Y60.A60'
% 'tigrigna]tigrigna5.male.N_tigrigna.R_uk.Y27.A45'
'hadiyya]hadiyya1.male.N_hadiyya.R_uk, usa.Y9.A51'
};
offset = 100;
gridwidth = 10;
%% Pick one reference file and use the detour files to get the validation values
ii = 1;
for iy = 1 :length(Files)
for ix = 1 : length(Files)
for iu = 1 : length(Files)
if iy == iu | ix == iu | ix == iy
continue;
end
file(ii).X2Y = fullfile(Files{iy}, [Files{ix}, '.mat']);
file(ii).X2U = fullfile(Files{iu}, [Files{ix}, '.mat']);
file(ii).U2Y = fullfile(Files{iy}, [Files{iu}, '.mat']);
for k = 1:length(datKind)
load(fullfile(datroot, datKind{k}, file(ii).X2Y));
M.X2Y = MT;
load(fullfile(datroot, datKind{k}, file(ii).X2U));
M.X2U = MT;
load(fullfile(datroot, datKind{k}, file(ii).U2Y));
M.U2Y = MT;
MaxY = max(M.X2Y(end, 2), M.U2Y(end, 2));
Grid = gridwidth/2:gridwidth:MaxY-gridwidth;
% Resampling
G.X2Y = mt2gt(M.X2Y, Grid);
G.U2Y = mt2gt(M.U2Y, Grid);
G.X2U2Y = mt2gt(M.X2U, G.U2Y(:, 1));
G.X2U2Y(:, 2) = G.U2Y(:, 2);
N = size(G.X2U2Y, 1);
Intv = offset:N-offset;
Err1(ix, iy, k) = simmx2(G.X2U2Y(Intv, 1),G.X2Y(Intv, 1), 'cosine');
Err2(ix, iy, k) = pow(G.X2U2Y(Intv, 1) - G.X2Y(Intv, 1));
Err3(ix, iy, k) = simmx2(G.X2U2Y(Intv, 1),G.X2Y(Intv, 1), 'tanimoto');
% Err4(ix, iy, k) = simmx2(G.X2U2Y(Intv, 1),G.X2Y(Intv, 1), 'mahal');
% Err5(ix, iy, k) = simmx2(G.X2U2Y(Intv, 1),G.X2Y(Intv, 1), 'corrcoef');
% Plotting
% figure(k);
% plot(G.X2Y(Intv, 1), G.X2Y(Intv, 2));
% hold on;
% plot(G.X2U2Y(Intv, 1), G.X2U2Y(Intv, 2), 'r');
% hold off;
% title(k);
% pause;
% % file(i).X2U
% plot(Diff); pause;
end
ii = ii + 1;
% pause;
end
fprintf('Completed [ix %d, iy %d]\n', ix, iy);
end
end
%%
Err = Err2;
MaxErr = max(Err(:));
Vmean = 0; Vvar = 0;
for k = 1 : length(datKind)
ee = Err(:, :, k);
Vmean(k) = mean(ee(:));
Vmed(k) = median(ee(:));
Vvar(k) = var(ee(:));
end
Vmean = Vmean/Vmean(2);
% Vvar = Vvar/Vvar(1);
Vmean
Vmed
Vvar
%%
for k = 1 : length(datKind)
figure(k);
bar3(Err(:, :, k)');
xlabel('Y; Reference'); ylabel('X; Test');
zlabel('Error');
% set(gca, 'YDir', 'normal');
% set(gca, 'XDir', 'reverse');
set(gca, 'XDir', 'normal');
end
view([-40, -30]);
for k = 1:length(datKind)
% Average
ydat(k, :) = var(Err(:, :, k), [], 1);
mean(ydat(k, :))
% ydat(k, :) = mean(Err(:, :, k), 1);
end
figure(k+1);
plot(ydat', 'o-');
legend(datKind);
title('Distance Measure comparison')
% Winner
% figure(k+2);
% ydat = mean(Err(:, :, 1) > Err(:, :, 2), 1);
% plot(ydat'>0.5, 'o-');
% %%
% GT.ca2 = mt2gt(MT.cb, GT.ba(:, 1));
% GT.ca2(:, 2) = GT.ba(:, 2);
%
% figure(4);
% plot(GT.ca(:, 1), GT.ca(:, 2), 'LineSmoothing', 'On'); hold on;
% plot(GT.ca2(:, 1), GT.ca2(:, 2), 'r', 'LineSmoothing', 'On'); hold off;
%
% legend('C - A', 'C - B & B - A');
%
% figure(5);
% % error plot
% idxs = 50:size(GT.ca, 1)-50;
% I1 = GT.ca(idxs, 1); I2 = GT.ca2(idxs, 1);
% err = I1 - I2;
% plot(GT.ca(idxs, 2), err);
%
% normerr = sqrt(sum(err.^2))
% coserr = simmx2(I1, I2)
% dum = corrcoef(I1, I2);
% correlation = dum(1, 2)
% Previously the best aligned indexes
% 529 : 'english_ usa_ male_ west covina_71.wav'
% 527 : 'english_ usa_ male_ washington_208.wav'
% 497 : 'english_ usa_ male_ reading_325.wav'
% 484 : 'english_ usa_ male_ pasadena_376.wav'
% 438 : 'english_ usa_ male_ hazlehurst_451.wav'
% 483 : 'english_ usa_ male_ paducah_462.wav'
% Pick a file as a reference, test1, test2
% fidx.ref = 244; % english_ usa_ female_ davenport_10.wav
%
% fidx.test1 = 529; % Pick a file as a test
% fidx.test2 = 527; % Pick another file as intermediate test file
%
% zrunByIndex(fidx.ref, fidx.test1, false);
% zrunByIndex(fidx.test1, fidx.test2, false);
% zrunByIndex(fidx.ref, fidx.test2, false);