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
|
/*
IMMS: Intelligent Multimedia Management System
Copyright (C) 2001-2009 Michael Grigoriev
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
*/
#include <assert.h>
#include <iostream>
#include <song.h>
#include <immsutil.h>
#include <string.h>
#include "distance.h"
#include "emd.h"
using std::cerr;
using std::endl;
float KL_Divergence(const Gaussian &g1, const Gaussian &g2)
{
float total = 0;
for (int i = 0; i < Gaussian::NumDimensions; ++i)
{
// Enforce a minimum for variences so we don't get huge distances
const float MinVariance = 10.0f;
float var1 = std::max(g1.vars[i], MinVariance);
float var2 = std::max(g2.vars[i], MinVariance);
float dist = var1 / var2 + var2 / var1 +
pow(g1.means[i] - g2.means[i], 2.0f) *
(1.0f / var1 + 1.0f / var2);
total += dist - 2;
}
return total;
}
float EMD::cost[NUMGAUSS][NUMGAUSS];
float EMD::raw_distance(const MixtureModel &m1, const MixtureModel &m2)
{
feature_t features[NUMGAUSS];
float w1[NUMGAUSS], w2[NUMGAUSS];
for (int i = 0; i < NUMGAUSS; ++i)
{
features[i] = i;
w1[i] = m1.gauss[i].weight;
w2[i] = m2.gauss[i].weight;
for (int j = 0; j < NUMGAUSS; ++j)
cost[i][j] = KL_Divergence(m1.gauss[i], m2.gauss[j]);
}
signature_t s1 = { NUMGAUSS, features, w1 };
signature_t s2 = { NUMGAUSS, features, w2 };
return emd(&s1, &s2, EMD::gauss_dist, 0, 0);
}
static bool normalize_beat_graph(float beats[BEATSSIZE], float *output, int comb)
{
float sum = 0, min = 1e100;
for (int i = 0; i < BEATSSIZE; ++i)
{
sum += beats[i];
if (beats[i] < min)
min = beats[i];
}
if (sum == 0)
return false;
// scale to keep the total area under the curve to be fixed
float scale = 100.0 / sum;
for (int i = 0; i < BEATSSIZE; ++i)
output[i / comb] += beats[i] * scale;
return true;
}
float EMD::raw_distance(float beats1[BEATSSIZE], float beats2[BEATSSIZE])
{
static const int comb = 5;
static const int OUTSIZE = DIVROUNDUP(BEATSSIZE, comb);
feature_t features[OUTSIZE];
for (int i = 0; i < OUTSIZE; ++i)
features[i] = i;
float b1[OUTSIZE], b2[OUTSIZE];
memset(b1, 0, sizeof(b1));
memset(b2, 0, sizeof(b2));
if (!normalize_beat_graph(beats1, b1, comb))
return -1;
if (!normalize_beat_graph(beats2, b2, comb))
return -1;
signature_t s1 = { OUTSIZE, features, b1 };
signature_t s2 = { OUTSIZE, features, b2 };
return emd(&s1, &s2, EMD::linear_dist, 0, 0);
}
float song_cepstr_distance(int uid1, int uid2)
{
Song song1("", uid1), song2("", uid2);
MixtureModel m1, m2;
if (!song1.get_acoustic(&m1, 0))
{
LOG(ERROR) << "warning: failed to load cepstrum data for uid "
<< uid1 << endl;
return -1;
}
if (!song2.get_acoustic(&m2, 0))
{
LOG(ERROR) << "warning: failed to load cepstrum data for uid "
<< uid2 << endl;
return -1;
}
return EMD::raw_distance(m1, m2);
}
float song_bpm_distance(int uid1, int uid2)
{
Song song1("", uid1), song2("", uid2);
MixtureModel m;
float beats1[BEATSSIZE], beats2[BEATSSIZE];
if (!song1.get_acoustic(&m, beats1))
{
LOG(ERROR) << "warning: failed to load bpm data for uid "
<< uid1 << endl;
return -1;
}
if (!song2.get_acoustic(&m, beats2))
{
LOG(ERROR) << "warning: failed to load bpm data for uid "
<< uid2 << endl;
return -1;
}
return EMD::raw_distance(beats1, beats2);
}
|