Coverage Report

Created: 2025-08-29 06:24

/src/speex/libspeex/vbr.c
Line
Count
Source (jump to first uncovered line)
1
/* Copyright (C) 2002 Jean-Marc Valin
2
   File: vbr.c
3
4
   VBR-related routines
5
6
   Redistribution and use in source and binary forms, with or without
7
   modification, are permitted provided that the following conditions
8
   are met:
9
10
   - Redistributions of source code must retain the above copyright
11
   notice, this list of conditions and the following disclaimer.
12
13
   - Redistributions in binary form must reproduce the above copyright
14
   notice, this list of conditions and the following disclaimer in the
15
   documentation and/or other materials provided with the distribution.
16
17
   - Neither the name of the Xiph.org Foundation nor the names of its
18
   contributors may be used to endorse or promote products derived from
19
   this software without specific prior written permission.
20
21
   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22
   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23
   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24
   A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
25
   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
26
   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
27
   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
28
   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
29
   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
30
   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
31
   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
32
33
*/
34
35
#ifdef HAVE_CONFIG_H
36
#include "config.h"
37
#endif
38
39
#include "vbr.h"
40
#include <math.h>
41
42
43
113k
#define sqr(x) ((x)*(x))
44
45
73.7k
#define MIN_ENERGY 6000
46
26.6k
#define NOISE_POW .3
47
48
#ifndef DISABLE_VBR
49
50
const float vbr_nb_thresh[9][11]={
51
   {-1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f}, /*   CNG   */
52
   { 4.0f,  2.5f,  2.0f,  1.2f,  0.5f,-0.25f, -0.5f, -0.7f, -0.8f, -0.9f, -1.0f}, /*  2 kbps */
53
   {10.0f,  6.5f,  5.2f,  4.5f,  3.9f, 3.7f,  3.0f,  2.5f,  2.3f,  1.8f,  1.0f}, /*  6 kbps */
54
   {11.0f,  8.8f,  7.5f,  6.5f,  5.0f,  4.2f,  3.9f,  3.9f,  3.5f,  3.0f,  1.0f}, /*  8 kbps */
55
   {11.0f, 11.0f,  9.9f,  8.5f,  7.0f, 5.25f,  4.5f,  4.0f,  4.0f,  4.0f,  2.0f}, /* 11 kbps */
56
   {11.0f, 11.0f, 11.0f, 11.0f,  9.5f, 9.25f,  8.0f,  7.0f,  5.0f,  4.0f,  3.0f}, /* 15 kbps */
57
   {11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f,  9.5f,  8.5f,  6.2f,  5.2f,  5.0f}, /* 18 kbps */
58
   {11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 10.0f,  9.8f,  7.5f}, /* 24 kbps */
59
   { 7.0f,  4.5f,  3.7f,  3.0f,  2.5f,  1.0f,  1.8f,  1.5f,  1.0f,  0.0f,  0.0f}  /*  4 kbps */
60
};
61
62
63
const float vbr_hb_thresh[5][11]={
64
   {-1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f}, /* silence */
65
   {-1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f}, /*  2 kbps */
66
   {11.0f, 11.0f,  9.5f,  8.5f,  7.5f,  6.0f,  5.0f,  3.9f,  3.0f,  2.0f,  1.0f}, /*  6 kbps */
67
   {11.0f, 11.0f, 11.0f, 11.0f, 11.0f,  9.5f,  8.7f,  7.8f,  7.0f,  6.5f,  4.0f}, /* 10 kbps */
68
   {11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f, 11.0f,  9.8f,  7.5f,  5.5f}  /* 18 kbps */
69
};
70
71
const float vbr_uhb_thresh[2][11]={
72
   {-1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f}, /* silence */
73
   { 3.9f,  2.5f,  0.0f,  0.0f,  0.0f,  0.0f,  0.0f,  0.0f,  0.0f,  0.0f, -1.0f}  /*  2 kbps */
74
};
75
76
void vbr_init(VBRState *vbr)
77
4.05k
{
78
4.05k
   int i;
79
80
4.05k
   vbr->average_energy=1600000;
81
4.05k
   vbr->last_energy=1;
82
4.05k
   vbr->accum_sum=0;
83
4.05k
   vbr->soft_pitch=0;
84
4.05k
   vbr->last_pitch_coef=0;
85
4.05k
   vbr->last_quality=0;
86
87
4.05k
   vbr->noise_accum = .05*pow(MIN_ENERGY, NOISE_POW);
88
4.05k
   vbr->noise_accum_count=.05;
89
4.05k
   vbr->noise_level=vbr->noise_accum/vbr->noise_accum_count;
90
4.05k
   vbr->consec_noise=0;
91
92
93
24.3k
   for (i=0;i<VBR_MEMORY_SIZE;i++)
94
20.2k
      vbr->last_log_energy[i] = log(MIN_ENERGY);
95
4.05k
}
96
97
98
/*
99
  This function should analyse the signal and decide how critical the
100
  coding error will be perceptually. The following factors should be
101
  taken into account:
102
103
  -Attacks (positive energy derivative) should be coded with more bits
104
105
  -Stationary voiced segments should receive more bits
106
107
  -Segments with (very) low absolute energy should receive less bits (maybe
108
  only shaped noise?)
109
110
  -DTX for near-zero energy?
111
112
  -Stationary fricative segments should have less bits
113
114
  -Temporal masking: when energy slope is decreasing, decrease the bit-rate
115
116
  -Decrease bit-rate for males (low pitch)?
117
118
  -(wideband only) less bits in the high-band when signal is very
119
  non-stationary (harder to notice high-frequency noise)???
120
121
*/
122
123
float vbr_analysis(VBRState *vbr, spx_word16_t *sig, int len, int pitch, float pitch_coef)
124
22.6k
{
125
22.6k
   int i;
126
22.6k
   float ener=0, ener1=0, ener2=0;
127
22.6k
   float qual=7;
128
22.6k
   float log_energy;
129
22.6k
   float non_st=0;
130
22.6k
   float voicing;
131
22.6k
   float pow_ener;
132
133
1.83M
   for (i=0;i<len>>1;i++)
134
1.80M
      ener1 += ((float)sig[i])*sig[i];
135
136
1.83M
   for (i=len>>1;i<len;i++)
137
1.80M
      ener2 += ((float)sig[i])*sig[i];
138
22.6k
   ener=ener1+ener2;
139
140
22.6k
   log_energy = log(ener+MIN_ENERGY);
141
135k
   for (i=0;i<VBR_MEMORY_SIZE;i++)
142
113k
      non_st += sqr(log_energy-vbr->last_log_energy[i]);
143
22.6k
   non_st =  non_st/(30*VBR_MEMORY_SIZE);
144
22.6k
   if (non_st>1)
145
10.0k
      non_st=1;
146
147
22.6k
   voicing = 3*(pitch_coef-.4)*fabs(pitch_coef-.4);
148
22.6k
   vbr->average_energy = 0.9*vbr->average_energy + .1*ener;
149
22.6k
   vbr->noise_level=vbr->noise_accum/vbr->noise_accum_count;
150
22.6k
   pow_ener = pow(ener,NOISE_POW);
151
22.6k
   if (vbr->noise_accum_count<.06 && ener>MIN_ENERGY)
152
5.44k
      vbr->noise_accum = .05*pow_ener;
153
154
22.6k
   if ((voicing<.3 && non_st < .2 && pow_ener < 1.2*vbr->noise_level)
155
22.6k
       || (voicing<.3 && non_st < .05 && pow_ener < 1.5*vbr->noise_level)
156
22.6k
       || (voicing<.4 && non_st < .05 && pow_ener < 1.2*vbr->noise_level)
157
22.6k
       || (voicing<0 && non_st < .05))
158
6.87k
   {
159
6.87k
      float tmp;
160
161
6.87k
      vbr->consec_noise++;
162
6.87k
      if (pow_ener > 3*vbr->noise_level)
163
14
         tmp = 3*vbr->noise_level;
164
6.85k
      else
165
6.85k
         tmp = pow_ener;
166
6.87k
      if (vbr->consec_noise>=4)
167
2.57k
      {
168
2.57k
         vbr->noise_accum = .95*vbr->noise_accum + .05*tmp;
169
2.57k
         vbr->noise_accum_count = .95*vbr->noise_accum_count + .05;
170
2.57k
      }
171
15.7k
   } else {
172
15.7k
      vbr->consec_noise=0;
173
15.7k
   }
174
175
22.6k
   if (pow_ener < vbr->noise_level && ener>MIN_ENERGY)
176
8.62k
   {
177
8.62k
      vbr->noise_accum = .95*vbr->noise_accum + .05*pow_ener;
178
8.62k
      vbr->noise_accum_count = .95*vbr->noise_accum_count + .05;
179
8.62k
   }
180
181
   /* Checking for very low absolute energy */
182
22.6k
   if (ener < 30000)
183
9.60k
   {
184
9.60k
      qual -= .7;
185
9.60k
      if (ener < 10000)
186
8.90k
         qual-=.7;
187
9.60k
      if (ener < 3000)
188
8.28k
         qual-=.7;
189
13.0k
   } else {
190
13.0k
      float short_diff, long_diff;
191
13.0k
      short_diff = log((ener+1)/(1+vbr->last_energy));
192
13.0k
      long_diff = log((ener+1)/(1+vbr->average_energy));
193
      /*fprintf (stderr, "%f %f\n", short_diff, long_diff);*/
194
195
13.0k
      if (long_diff<-5)
196
2.65k
         long_diff=-5;
197
13.0k
      if (long_diff>2)
198
2.31k
         long_diff=2;
199
200
13.0k
      if (long_diff>0)
201
5.71k
         qual += .6*long_diff;
202
13.0k
      if (long_diff<0)
203
7.29k
         qual += .5*long_diff;
204
13.0k
      if (short_diff>0)
205
5.51k
      {
206
5.51k
         if (short_diff>5)
207
2.64k
            short_diff=5;
208
5.51k
         qual += 1*short_diff;
209
5.51k
      }
210
      /* Checking for energy increases */
211
13.0k
      if (ener2 > 1.6*ener1)
212
3.28k
         qual += .5;
213
13.0k
   }
214
22.6k
   vbr->last_energy = ener;
215
22.6k
   vbr->soft_pitch = .8*vbr->soft_pitch + .2*pitch_coef;
216
22.6k
   qual += 2.2*((pitch_coef-.4) + (vbr->soft_pitch-.4));
217
218
22.6k
   if (qual < vbr->last_quality)
219
9.40k
      qual = .5*qual + .5*vbr->last_quality;
220
22.6k
   if (qual<4)
221
7.74k
      qual=4;
222
22.6k
   if (qual>10)
223
3.81k
      qual=10;
224
225
   /*
226
   if (vbr->consec_noise>=2)
227
      qual-=1.3;
228
   if (vbr->consec_noise>=5)
229
      qual-=1.3;
230
   if (vbr->consec_noise>=12)
231
      qual-=1.3;
232
   */
233
22.6k
   if (vbr->consec_noise>=3)
234
3.71k
      qual=4;
235
236
22.6k
   if (vbr->consec_noise)
237
6.87k
      qual -= 1.0 * (log(3.0 + vbr->consec_noise)-log(3));
238
22.6k
   if (qual<0)
239
0
      qual=0;
240
241
22.6k
   if (ener<1600000)
242
12.0k
   {
243
12.0k
      if (vbr->consec_noise>2)
244
3.53k
         qual-=0.5*(log(3.0 + vbr->consec_noise)-log(3));
245
12.0k
      if (ener<10000&&vbr->consec_noise>2)
246
3.44k
         qual-=0.5*(log(3.0 + vbr->consec_noise)-log(3));
247
12.0k
      if (qual<0)
248
0
         qual=0;
249
12.0k
      qual += .3*log(.0001+ener/1600000.0);
250
12.0k
   }
251
22.6k
   if (qual<-1)
252
972
      qual=-1;
253
254
   /*printf ("%f %f %f %f\n", qual, voicing, non_st, pow_ener/(.01+vbr->noise_level));*/
255
256
22.6k
   vbr->last_pitch_coef = pitch_coef;
257
22.6k
   vbr->last_quality = qual;
258
259
113k
   for (i=VBR_MEMORY_SIZE-1;i>0;i--)
260
90.4k
      vbr->last_log_energy[i] = vbr->last_log_energy[i-1];
261
22.6k
   vbr->last_log_energy[0] = log_energy;
262
263
   /*printf ("VBR: %f %f %f %f\n", (float)(log_energy-log(vbr->average_energy+MIN_ENERGY)), non_st, voicing, vbr->noise_level);*/
264
265
22.6k
   return qual;
266
22.6k
}
267
268
void vbr_destroy(VBRState *vbr)
269
4.05k
{
270
4.05k
}
271
272
#endif /* #ifndef DISABLE_VBR */