FFmpeg
af_astats.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2009 Rob Sykes <robs@users.sourceforge.net>
3  * Copyright (c) 2013 Paul B Mahol
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include <float.h>
23 #include <math.h>
24 
25 #include "libavutil/opt.h"
26 #include "audio.h"
27 #include "avfilter.h"
28 #include "internal.h"
29 
30 #define HISTOGRAM_SIZE 8192
31 #define HISTOGRAM_MAX (HISTOGRAM_SIZE-1)
32 
33 #define MEASURE_ALL UINT_MAX
34 #define MEASURE_NONE 0
35 
36 #define MEASURE_DC_OFFSET (1 << 0)
37 #define MEASURE_MIN_LEVEL (1 << 1)
38 #define MEASURE_MAX_LEVEL (1 << 2)
39 #define MEASURE_MIN_DIFFERENCE (1 << 3)
40 #define MEASURE_MAX_DIFFERENCE (1 << 4)
41 #define MEASURE_MEAN_DIFFERENCE (1 << 5)
42 #define MEASURE_RMS_DIFFERENCE (1 << 6)
43 #define MEASURE_PEAK_LEVEL (1 << 7)
44 #define MEASURE_RMS_LEVEL (1 << 8)
45 #define MEASURE_RMS_PEAK (1 << 9)
46 #define MEASURE_RMS_TROUGH (1 << 10)
47 #define MEASURE_CREST_FACTOR (1 << 11)
48 #define MEASURE_FLAT_FACTOR (1 << 12)
49 #define MEASURE_PEAK_COUNT (1 << 13)
50 #define MEASURE_BIT_DEPTH (1 << 14)
51 #define MEASURE_DYNAMIC_RANGE (1 << 15)
52 #define MEASURE_ZERO_CROSSINGS (1 << 16)
53 #define MEASURE_ZERO_CROSSINGS_RATE (1 << 17)
54 #define MEASURE_NUMBER_OF_SAMPLES (1 << 18)
55 #define MEASURE_NUMBER_OF_NANS (1 << 19)
56 #define MEASURE_NUMBER_OF_INFS (1 << 20)
57 #define MEASURE_NUMBER_OF_DENORMALS (1 << 21)
58 #define MEASURE_NOISE_FLOOR (1 << 22)
59 #define MEASURE_NOISE_FLOOR_COUNT (1 << 23)
60 #define MEASURE_ENTROPY (1 << 24)
61 
62 #define MEASURE_MINMAXPEAK (MEASURE_MIN_LEVEL | MEASURE_MAX_LEVEL | MEASURE_PEAK_LEVEL)
63 
64 typedef struct ChannelStats {
65  double last;
66  double last_non_zero;
67  double min_non_zero;
68  double sigma_x, sigma_x2;
69  double avg_sigma_x2, min_sigma_x2, max_sigma_x2;
70  double min, max;
71  double nmin, nmax;
72  double min_run, max_run;
73  double min_runs, max_runs;
74  double min_diff, max_diff;
75  double diff1_sum;
76  double diff1_sum_x2;
77  uint64_t mask, imask;
78  uint64_t min_count, max_count;
79  uint64_t noise_floor_count;
80  uint64_t zero_runs;
81  uint64_t nb_samples;
82  uint64_t nb_nans;
83  uint64_t nb_infs;
84  uint64_t nb_denormals;
85  double *win_samples;
86  uint64_t histogram[HISTOGRAM_SIZE];
87  uint64_t ehistogram[HISTOGRAM_SIZE];
88  int win_pos;
89  int max_index;
90  double noise_floor;
91  double entropy;
92 } ChannelStats;
93 
94 typedef struct AudioStatsContext {
95  const AVClass *class;
96  ChannelStats *chstats;
97  int nb_channels;
98  uint64_t tc_samples;
99  double time_constant;
100  double mult;
101  int metadata;
102  int used;
103  int reset_count;
104  int nb_frames;
105  int maxbitdepth;
106  int measure_perchannel;
107  int measure_overall;
108  int is_float;
109  int is_double;
110 } AudioStatsContext;
111 
112 #define OFFSET(x) offsetof(AudioStatsContext, x)
113 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
114 
115 static const AVOption astats_options[] = {
116  { "length", "set the window length", OFFSET(time_constant), AV_OPT_TYPE_DOUBLE, {.dbl=.05}, 0, 10, FLAGS },
117  { "metadata", "inject metadata in the filtergraph", OFFSET(metadata), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
118  { "reset", "Set the number of frames over which cumulative stats are calculated before being reset", OFFSET(reset_count), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS },
119  { "measure_perchannel", "Select the parameters which are measured per channel", OFFSET(measure_perchannel), AV_OPT_TYPE_FLAGS, {.i64=MEASURE_ALL}, 0, UINT_MAX, FLAGS, "measure" },
120  { "none" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NONE }, 0, 0, FLAGS, "measure" },
121  { "all" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_ALL }, 0, 0, FLAGS, "measure" },
122  { "Bit_depth" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_BIT_DEPTH }, 0, 0, FLAGS, "measure" },
123  { "Crest_factor" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_CREST_FACTOR }, 0, 0, FLAGS, "measure" },
124  { "DC_offset" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_DC_OFFSET }, 0, 0, FLAGS, "measure" },
125  { "Dynamic_range" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_DYNAMIC_RANGE }, 0, 0, FLAGS, "measure" },
126  { "Entropy" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_ENTROPY }, 0, 0, FLAGS, "measure" },
127  { "Flat_factor" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_FLAT_FACTOR }, 0, 0, FLAGS, "measure" },
128  { "Max_difference" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_MAX_DIFFERENCE }, 0, 0, FLAGS, "measure" },
129  { "Max_level" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_MAX_LEVEL }, 0, 0, FLAGS, "measure" },
130  { "Mean_difference" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_MEAN_DIFFERENCE }, 0, 0, FLAGS, "measure" },
131  { "Min_difference" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_MIN_DIFFERENCE }, 0, 0, FLAGS, "measure" },
132  { "Min_level" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_MIN_LEVEL }, 0, 0, FLAGS, "measure" },
133  { "Noise_floor" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NOISE_FLOOR }, 0, 0, FLAGS, "measure" },
134  { "Noise_floor_count" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NOISE_FLOOR_COUNT }, 0, 0, FLAGS, "measure" },
135  { "Number_of_Infs" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NUMBER_OF_INFS }, 0, 0, FLAGS, "measure" },
136  { "Number_of_NaNs" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NUMBER_OF_NANS }, 0, 0, FLAGS, "measure" },
137  { "Number_of_denormals" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NUMBER_OF_DENORMALS }, 0, 0, FLAGS, "measure" },
138  { "Number_of_samples" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_NUMBER_OF_SAMPLES }, 0, 0, FLAGS, "measure" },
139  { "Peak_count" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_PEAK_COUNT }, 0, 0, FLAGS, "measure" },
140  { "Peak_level" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_PEAK_LEVEL }, 0, 0, FLAGS, "measure" },
141  { "RMS_difference" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_RMS_DIFFERENCE }, 0, 0, FLAGS, "measure" },
142  { "RMS_level" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_RMS_LEVEL }, 0, 0, FLAGS, "measure" },
143  { "RMS_peak" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_RMS_PEAK }, 0, 0, FLAGS, "measure" },
144  { "RMS_trough" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_RMS_TROUGH }, 0, 0, FLAGS, "measure" },
145  { "Zero_crossings" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_ZERO_CROSSINGS }, 0, 0, FLAGS, "measure" },
146  { "Zero_crossings_rate" , "", 0, AV_OPT_TYPE_CONST, {.i64=MEASURE_ZERO_CROSSINGS_RATE }, 0, 0, FLAGS, "measure" },
147  { "measure_overall", "Select the parameters which are measured overall", OFFSET(measure_overall), AV_OPT_TYPE_FLAGS, {.i64=MEASURE_ALL}, 0, UINT_MAX, FLAGS, "measure" },
148  { NULL }
149 };
150 
151 AVFILTER_DEFINE_CLASS(astats);
152 
153 static void reset_stats(AudioStatsContext *s)
154 {
155  int c;
156 
157  for (c = 0; c < s->nb_channels; c++) {
158  ChannelStats *p = &s->chstats[c];
159 
160  p->min = p->nmin = p->min_sigma_x2 = DBL_MAX;
161  p->max = p->nmax = p->max_sigma_x2 =-DBL_MAX;
162  p->min_non_zero = DBL_MAX;
163  p->min_diff = DBL_MAX;
164  p->max_diff = 0;
165  p->sigma_x = 0;
166  p->sigma_x2 = 0;
167  p->avg_sigma_x2 = 0;
168  p->min_run = 0;
169  p->max_run = 0;
170  p->min_runs = 0;
171  p->max_runs = 0;
172  p->diff1_sum = 0;
173  p->diff1_sum_x2 = 0;
174  p->mask = 0;
175  p->imask = 0xFFFFFFFFFFFFFFFF;
176  p->min_count = 0;
177  p->max_count = 0;
178  p->zero_runs = 0;
179  p->nb_samples = 0;
180  p->nb_nans = 0;
181  p->nb_infs = 0;
182  p->nb_denormals = 0;
183  p->last = NAN;
184  p->noise_floor = NAN;
185  p->noise_floor_count = 0;
186  p->entropy = 0;
187  p->win_pos = 0;
188  memset(p->win_samples, 0, s->tc_samples * sizeof(*p->win_samples));
189  memset(p->histogram, 0, sizeof(p->histogram));
190  memset(p->ehistogram, 0, sizeof(p->ehistogram));
191  }
192 }
193 
194 static int config_output(AVFilterLink *outlink)
195 {
196  AudioStatsContext *s = outlink->src->priv;
197 
198  s->chstats = av_calloc(sizeof(*s->chstats), outlink->ch_layout.nb_channels);
199  if (!s->chstats)
200  return AVERROR(ENOMEM);
201 
202  s->tc_samples = FFMAX(s->time_constant * outlink->sample_rate + .5, 1);
203  s->nb_channels = outlink->ch_layout.nb_channels;
204 
205  for (int i = 0; i < s->nb_channels; i++) {
206  ChannelStats *p = &s->chstats[i];
207 
208  p->win_samples = av_calloc(s->tc_samples, sizeof(*p->win_samples));
209  if (!p->win_samples)
210  return AVERROR(ENOMEM);
211  }
212 
213  s->mult = exp((-1 / s->time_constant / outlink->sample_rate));
214  s->nb_frames = 0;
215  s->maxbitdepth = av_get_bytes_per_sample(outlink->format) * 8;
216  s->is_double = outlink->format == AV_SAMPLE_FMT_DBL ||
217  outlink->format == AV_SAMPLE_FMT_DBLP;
218 
219  s->is_float = outlink->format == AV_SAMPLE_FMT_FLT ||
220  outlink->format == AV_SAMPLE_FMT_FLTP;
221 
222  reset_stats(s);
223 
224  return 0;
225 }
226 
227 static void bit_depth(AudioStatsContext *s, uint64_t mask, uint64_t imask, AVRational *depth)
228 {
229  unsigned result = s->maxbitdepth;
230 
231  mask = mask & (~imask);
232 
233  for (; result && !(mask & 1); --result, mask >>= 1);
234 
235  depth->den = result;
236  depth->num = 0;
237 
238  for (; result; --result, mask >>= 1)
239  if (mask & 1)
240  depth->num++;
241 }
242 
243 static double calc_entropy(AudioStatsContext *s, ChannelStats *p)
244 {
245  double entropy = 0.;
246 
247  for (int i = 0; i < HISTOGRAM_SIZE; i++) {
248  double entry = p->ehistogram[i] / ((double)p->nb_samples);
249 
250  if (entry > 1e-8)
251  entropy += entry * log2(entry);
252  }
253 
254  return -entropy / log2(HISTOGRAM_SIZE);
255 }
256 
257 static inline void update_minmax(AudioStatsContext *s, ChannelStats *p, double d)
258 {
259  if (d < p->min)
260  p->min = d;
261  if (d > p->max)
262  p->max = d;
263 }
264 
265 static inline void update_stat(AudioStatsContext *s, ChannelStats *p, double d, double nd, int64_t i)
266 {
267  double drop;
268  int index;
269 
270  if (d < p->min) {
271  p->min = d;
272  p->nmin = nd;
273  p->min_run = 1;
274  p->min_runs = 0;
275  p->min_count = 1;
276  } else if (d == p->min) {
277  p->min_count++;
278  p->min_run = d == p->last ? p->min_run + 1 : 1;
279  } else if (p->last == p->min) {
280  p->min_runs += p->min_run * p->min_run;
281  }
282 
283  if (d != 0 && FFABS(d) < p->min_non_zero)
284  p->min_non_zero = FFABS(d);
285 
286  if (d > p->max) {
287  p->max = d;
288  p->nmax = nd;
289  p->max_run = 1;
290  p->max_runs = 0;
291  p->max_count = 1;
292  } else if (d == p->max) {
293  p->max_count++;
294  p->max_run = d == p->last ? p->max_run + 1 : 1;
295  } else if (p->last == p->max) {
296  p->max_runs += p->max_run * p->max_run;
297  }
298 
299  if (d != 0) {
300  p->zero_runs += FFSIGN(d) != FFSIGN(p->last_non_zero);
301  p->last_non_zero = d;
302  }
303 
304  p->sigma_x += nd;
305  p->sigma_x2 += nd * nd;
306  p->avg_sigma_x2 = p->avg_sigma_x2 * s->mult + (1.0 - s->mult) * nd * nd;
307  if (!isnan(p->last)) {
308  p->min_diff = FFMIN(p->min_diff, fabs(d - p->last));
309  p->max_diff = FFMAX(p->max_diff, fabs(d - p->last));
310  p->diff1_sum += fabs(d - p->last);
311  p->diff1_sum_x2 += (d - p->last) * (d - p->last);
312  }
313  p->last = d;
314  p->mask |= i;
315  p->imask &= i;
316 
317  drop = p->win_samples[p->win_pos];
318  p->win_samples[p->win_pos] = nd;
319  index = av_clip(lrint(av_clipd(FFABS(nd), 0.0, 1.0) * HISTOGRAM_MAX), 0, HISTOGRAM_MAX);
320  p->max_index = FFMAX(p->max_index, index);
321  p->histogram[index]++;
322  p->ehistogram[index]++;
323  if (!isnan(p->noise_floor))
324  p->histogram[av_clip(lrint(av_clipd(FFABS(drop), 0.0, 1.0) * HISTOGRAM_MAX), 0, HISTOGRAM_MAX)]--;
325  p->win_pos++;
326 
327  while (p->histogram[p->max_index] == 0)
328  p->max_index--;
329  if (p->win_pos >= s->tc_samples || !isnan(p->noise_floor)) {
330  double noise_floor = 1.;
331 
332  for (int i = p->max_index; i >= 0; i--) {
333  if (p->histogram[i]) {
334  noise_floor = i / (double)HISTOGRAM_MAX;
335  break;
336  }
337  }
338 
339  if (isnan(p->noise_floor)) {
340  p->noise_floor = noise_floor;
341  p->noise_floor_count = 1;
342  } else {
343  if (noise_floor < p->noise_floor) {
344  p->noise_floor = noise_floor;
345  p->noise_floor_count = 1;
346  } else if (noise_floor == p->noise_floor) {
347  p->noise_floor_count++;
348  }
349  }
350  }
351 
352  if (p->win_pos >= s->tc_samples) {
353  p->win_pos = 0;
354  }
355 
356  if (p->nb_samples >= s->tc_samples) {
357  p->max_sigma_x2 = FFMAX(p->max_sigma_x2, p->avg_sigma_x2);
358  p->min_sigma_x2 = FFMIN(p->min_sigma_x2, p->avg_sigma_x2);
359  }
360  p->nb_samples++;
361 }
362 
363 static inline void update_float_stat(AudioStatsContext *s, ChannelStats *p, float d)
364 {
365  int type = fpclassify(d);
366 
367  p->nb_nans += type == FP_NAN;
368  p->nb_infs += type == FP_INFINITE;
369  p->nb_denormals += type == FP_SUBNORMAL;
370 }
371 
372 static inline void update_double_stat(AudioStatsContext *s, ChannelStats *p, double d)
373 {
374  int type = fpclassify(d);
375 
376  p->nb_nans += type == FP_NAN;
377  p->nb_infs += type == FP_INFINITE;
378  p->nb_denormals += type == FP_SUBNORMAL;
379 }
380 
381 static void set_meta(AVDictionary **metadata, int chan, const char *key,
382  const char *fmt, double val)
383 {
384  uint8_t value[128];
385  uint8_t key2[128];
386 
387  snprintf(value, sizeof(value), fmt, val);
388  if (chan)
389  snprintf(key2, sizeof(key2), "lavfi.astats.%d.%s", chan, key);
390  else
391  snprintf(key2, sizeof(key2), "lavfi.astats.%s", key);
392  av_dict_set(metadata, key2, value, 0);
393 }
394 
395 #define LINEAR_TO_DB(x) (log10(x) * 20)
396 
397 static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
398 {
399  uint64_t mask = 0, imask = 0xFFFFFFFFFFFFFFFF, min_count = 0, max_count = 0, nb_samples = 0, noise_floor_count = 0;
400  uint64_t nb_nans = 0, nb_infs = 0, nb_denormals = 0;
401  double min_runs = 0, max_runs = 0,
402  min = DBL_MAX, max =-DBL_MAX, min_diff = DBL_MAX, max_diff = 0,
403  nmin = DBL_MAX, nmax =-DBL_MAX,
404  max_sigma_x = 0,
405  diff1_sum = 0,
406  diff1_sum_x2 = 0,
407  sigma_x2 = 0,
408  noise_floor = 0,
409  entropy = 0,
410  min_sigma_x2 = DBL_MAX,
411  max_sigma_x2 =-DBL_MAX;
412  AVRational depth;
413  int c;
414 
415  for (c = 0; c < s->nb_channels; c++) {
416  ChannelStats *p = &s->chstats[c];
417 
418  if (p->nb_samples < s->tc_samples)
419  p->min_sigma_x2 = p->max_sigma_x2 = p->sigma_x2 / p->nb_samples;
420 
421  min = FFMIN(min, p->min);
422  max = FFMAX(max, p->max);
423  nmin = FFMIN(nmin, p->nmin);
424  nmax = FFMAX(nmax, p->nmax);
425  min_diff = FFMIN(min_diff, p->min_diff);
426  max_diff = FFMAX(max_diff, p->max_diff);
427  diff1_sum += p->diff1_sum;
428  diff1_sum_x2 += p->diff1_sum_x2;
429  min_sigma_x2 = FFMIN(min_sigma_x2, p->min_sigma_x2);
430  max_sigma_x2 = FFMAX(max_sigma_x2, p->max_sigma_x2);
431  sigma_x2 += p->sigma_x2;
432  noise_floor = FFMAX(noise_floor, p->noise_floor);
433  noise_floor_count += p->noise_floor_count;
434  p->entropy = calc_entropy(s, p);
435  entropy += p->entropy;
436  min_count += p->min_count;
437  max_count += p->max_count;
438  min_runs += p->min_runs;
439  max_runs += p->max_runs;
440  mask |= p->mask;
441  imask &= p->imask;
442  nb_samples += p->nb_samples;
443  nb_nans += p->nb_nans;
444  nb_infs += p->nb_infs;
445  nb_denormals += p->nb_denormals;
446  if (fabs(p->sigma_x) > fabs(max_sigma_x))
447  max_sigma_x = p->sigma_x;
448 
449  if (s->measure_perchannel & MEASURE_DC_OFFSET)
450  set_meta(metadata, c + 1, "DC_offset", "%f", p->sigma_x / p->nb_samples);
451  if (s->measure_perchannel & MEASURE_MIN_LEVEL)
452  set_meta(metadata, c + 1, "Min_level", "%f", p->min);
453  if (s->measure_perchannel & MEASURE_MAX_LEVEL)
454  set_meta(metadata, c + 1, "Max_level", "%f", p->max);
455  if (s->measure_perchannel & MEASURE_MIN_DIFFERENCE)
456  set_meta(metadata, c + 1, "Min_difference", "%f", p->min_diff);
457  if (s->measure_perchannel & MEASURE_MAX_DIFFERENCE)
458  set_meta(metadata, c + 1, "Max_difference", "%f", p->max_diff);
459  if (s->measure_perchannel & MEASURE_MEAN_DIFFERENCE)
460  set_meta(metadata, c + 1, "Mean_difference", "%f", p->diff1_sum / (p->nb_samples - 1));
461  if (s->measure_perchannel & MEASURE_RMS_DIFFERENCE)
462  set_meta(metadata, c + 1, "RMS_difference", "%f", sqrt(p->diff1_sum_x2 / (p->nb_samples - 1)));
463  if (s->measure_perchannel & MEASURE_PEAK_LEVEL)
464  set_meta(metadata, c + 1, "Peak_level", "%f", LINEAR_TO_DB(FFMAX(-p->nmin, p->nmax)));
465  if (s->measure_perchannel & MEASURE_RMS_LEVEL)
466  set_meta(metadata, c + 1, "RMS_level", "%f", LINEAR_TO_DB(sqrt(p->sigma_x2 / p->nb_samples)));
467  if (s->measure_perchannel & MEASURE_RMS_PEAK)
468  set_meta(metadata, c + 1, "RMS_peak", "%f", LINEAR_TO_DB(sqrt(p->max_sigma_x2)));
469  if (s->measure_perchannel & MEASURE_RMS_TROUGH)
470  set_meta(metadata, c + 1, "RMS_trough", "%f", LINEAR_TO_DB(sqrt(p->min_sigma_x2)));
471  if (s->measure_perchannel & MEASURE_CREST_FACTOR)
472  set_meta(metadata, c + 1, "Crest_factor", "%f", p->sigma_x2 ? FFMAX(-p->min, p->max) / sqrt(p->sigma_x2 / p->nb_samples) : 1);
473  if (s->measure_perchannel & MEASURE_FLAT_FACTOR)
474  set_meta(metadata, c + 1, "Flat_factor", "%f", LINEAR_TO_DB((p->min_runs + p->max_runs) / (p->min_count + p->max_count)));
475  if (s->measure_perchannel & MEASURE_PEAK_COUNT)
476  set_meta(metadata, c + 1, "Peak_count", "%f", (float)(p->min_count + p->max_count));
477  if (s->measure_perchannel & MEASURE_NOISE_FLOOR)
478  set_meta(metadata, c + 1, "Noise_floor", "%f", LINEAR_TO_DB(p->noise_floor));
479  if (s->measure_perchannel & MEASURE_NOISE_FLOOR_COUNT)
480  set_meta(metadata, c + 1, "Noise_floor_count", "%f", p->noise_floor_count);
481  if (s->measure_perchannel & MEASURE_ENTROPY)
482  set_meta(metadata, c + 1, "Entropy", "%f", p->entropy);
483  if (s->measure_perchannel & MEASURE_BIT_DEPTH) {
484  bit_depth(s, p->mask, p->imask, &depth);
485  set_meta(metadata, c + 1, "Bit_depth", "%f", depth.num);
486  set_meta(metadata, c + 1, "Bit_depth2", "%f", depth.den);
487  }
488  if (s->measure_perchannel & MEASURE_DYNAMIC_RANGE)
489  set_meta(metadata, c + 1, "Dynamic_range", "%f", LINEAR_TO_DB(2 * FFMAX(FFABS(p->min), FFABS(p->max))/ p->min_non_zero));
490  if (s->measure_perchannel & MEASURE_ZERO_CROSSINGS)
491  set_meta(metadata, c + 1, "Zero_crossings", "%f", p->zero_runs);
492  if (s->measure_perchannel & MEASURE_ZERO_CROSSINGS_RATE)
493  set_meta(metadata, c + 1, "Zero_crossings_rate", "%f", p->zero_runs/(double)p->nb_samples);
494  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_NANS)
495  set_meta(metadata, c + 1, "Number of NaNs", "%f", p->nb_nans);
496  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_INFS)
497  set_meta(metadata, c + 1, "Number of Infs", "%f", p->nb_infs);
498  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_DENORMALS)
499  set_meta(metadata, c + 1, "Number of denormals", "%f", p->nb_denormals);
500  }
501 
502  if (s->measure_overall & MEASURE_DC_OFFSET)
503  set_meta(metadata, 0, "Overall.DC_offset", "%f", max_sigma_x / (nb_samples / s->nb_channels));
504  if (s->measure_overall & MEASURE_MIN_LEVEL)
505  set_meta(metadata, 0, "Overall.Min_level", "%f", min);
506  if (s->measure_overall & MEASURE_MAX_LEVEL)
507  set_meta(metadata, 0, "Overall.Max_level", "%f", max);
508  if (s->measure_overall & MEASURE_MIN_DIFFERENCE)
509  set_meta(metadata, 0, "Overall.Min_difference", "%f", min_diff);
510  if (s->measure_overall & MEASURE_MAX_DIFFERENCE)
511  set_meta(metadata, 0, "Overall.Max_difference", "%f", max_diff);
512  if (s->measure_overall & MEASURE_MEAN_DIFFERENCE)
513  set_meta(metadata, 0, "Overall.Mean_difference", "%f", diff1_sum / (nb_samples - s->nb_channels));
514  if (s->measure_overall & MEASURE_RMS_DIFFERENCE)
515  set_meta(metadata, 0, "Overall.RMS_difference", "%f", sqrt(diff1_sum_x2 / (nb_samples - s->nb_channels)));
516  if (s->measure_overall & MEASURE_PEAK_LEVEL)
517  set_meta(metadata, 0, "Overall.Peak_level", "%f", LINEAR_TO_DB(FFMAX(-nmin, nmax)));
518  if (s->measure_overall & MEASURE_RMS_LEVEL)
519  set_meta(metadata, 0, "Overall.RMS_level", "%f", LINEAR_TO_DB(sqrt(sigma_x2 / nb_samples)));
520  if (s->measure_overall & MEASURE_RMS_PEAK)
521  set_meta(metadata, 0, "Overall.RMS_peak", "%f", LINEAR_TO_DB(sqrt(max_sigma_x2)));
522  if (s->measure_overall & MEASURE_RMS_TROUGH)
523  set_meta(metadata, 0, "Overall.RMS_trough", "%f", LINEAR_TO_DB(sqrt(min_sigma_x2)));
524  if (s->measure_overall & MEASURE_FLAT_FACTOR)
525  set_meta(metadata, 0, "Overall.Flat_factor", "%f", LINEAR_TO_DB((min_runs + max_runs) / (min_count + max_count)));
526  if (s->measure_overall & MEASURE_PEAK_COUNT)
527  set_meta(metadata, 0, "Overall.Peak_count", "%f", (float)(min_count + max_count) / (double)s->nb_channels);
528  if (s->measure_overall & MEASURE_NOISE_FLOOR)
529  set_meta(metadata, 0, "Overall.Noise_floor", "%f", LINEAR_TO_DB(noise_floor));
530  if (s->measure_overall & MEASURE_NOISE_FLOOR_COUNT)
531  set_meta(metadata, 0, "Overall.Noise_floor_count", "%f", noise_floor_count / (double)s->nb_channels);
532  if (s->measure_overall & MEASURE_ENTROPY)
533  set_meta(metadata, 0, "Overall.Entropy", "%f", entropy / (double)s->nb_channels);
534  if (s->measure_overall & MEASURE_BIT_DEPTH) {
535  bit_depth(s, mask, imask, &depth);
536  set_meta(metadata, 0, "Overall.Bit_depth", "%f", depth.num);
537  set_meta(metadata, 0, "Overall.Bit_depth2", "%f", depth.den);
538  }
539  if (s->measure_overall & MEASURE_NUMBER_OF_SAMPLES)
540  set_meta(metadata, 0, "Overall.Number_of_samples", "%f", nb_samples / s->nb_channels);
541  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_NANS)
542  set_meta(metadata, 0, "Number of NaNs", "%f", nb_nans / (float)s->nb_channels);
543  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_INFS)
544  set_meta(metadata, 0, "Number of Infs", "%f", nb_infs / (float)s->nb_channels);
545  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_DENORMALS)
546  set_meta(metadata, 0, "Number of denormals", "%f", nb_denormals / (float)s->nb_channels);
547 }
548 
549 #define UPDATE_STATS_P(type, update_func, update_float, channel_func) \
550  for (int c = start; c < end; c++) { \
551  ChannelStats *p = &s->chstats[c]; \
552  const type *src = (const type *)data[c]; \
553  const type * const srcend = src + samples; \
554  for (; src < srcend; src++) { \
555  update_func; \
556  update_float; \
557  } \
558  channel_func; \
559  }
560 
561 #define UPDATE_STATS_I(type, update_func, update_float, channel_func) \
562  for (int c = start; c < end; c++) { \
563  ChannelStats *p = &s->chstats[c]; \
564  const type *src = (const type *)data[0]; \
565  const type * const srcend = src + samples * channels; \
566  for (src += c; src < srcend; src += channels) { \
567  update_func; \
568  update_float; \
569  } \
570  channel_func; \
571  }
572 
573 #define UPDATE_STATS(planar, type, sample, normalizer_suffix, int_sample) \
574  if ((s->measure_overall | s->measure_perchannel) & ~MEASURE_MINMAXPEAK) { \
575  UPDATE_STATS_##planar(type, update_stat(s, p, sample, sample normalizer_suffix, int_sample), s->is_float ? update_float_stat(s, p, sample) : s->is_double ? update_double_stat(s, p, sample) : (void)NULL, ); \
576  } else { \
577  UPDATE_STATS_##planar(type, update_minmax(s, p, sample), , p->nmin = p->min normalizer_suffix; p->nmax = p->max normalizer_suffix;); \
578  }
579 
580 static int filter_channel(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
581 {
582  AudioStatsContext *s = ctx->priv;
583  AVFilterLink *inlink = ctx->inputs[0];
584  AVFrame *buf = arg;
585  const uint8_t * const * const data = (const uint8_t * const *)buf->extended_data;
586  const int channels = s->nb_channels;
587  const int samples = buf->nb_samples;
588  const int start = (buf->ch_layout.nb_channels * jobnr) / nb_jobs;
589  const int end = (buf->ch_layout.nb_channels * (jobnr+1)) / nb_jobs;
590 
591  switch (inlink->format) {
592  case AV_SAMPLE_FMT_DBLP:
593  UPDATE_STATS(P, double, *src, , llrint(*src * (UINT64_C(1) << 63)));
594  break;
595  case AV_SAMPLE_FMT_DBL:
596  UPDATE_STATS(I, double, *src, , llrint(*src * (UINT64_C(1) << 63)));
597  break;
598  case AV_SAMPLE_FMT_FLTP:
599  UPDATE_STATS(P, float, *src, , llrint(*src * (UINT64_C(1) << 31)));
600  break;
601  case AV_SAMPLE_FMT_FLT:
602  UPDATE_STATS(I, float, *src, , llrint(*src * (UINT64_C(1) << 31)));
603  break;
604  case AV_SAMPLE_FMT_S64P:
605  UPDATE_STATS(P, int64_t, *src, / (double)INT64_MAX, *src);
606  break;
607  case AV_SAMPLE_FMT_S64:
608  UPDATE_STATS(I, int64_t, *src, / (double)INT64_MAX, *src);
609  break;
610  case AV_SAMPLE_FMT_S32P:
611  UPDATE_STATS(P, int32_t, *src, / (double)INT32_MAX, *src);
612  break;
613  case AV_SAMPLE_FMT_S32:
614  UPDATE_STATS(I, int32_t, *src, / (double)INT32_MAX, *src);
615  break;
616  case AV_SAMPLE_FMT_S16P:
617  UPDATE_STATS(P, int16_t, *src, / (double)INT16_MAX, *src);
618  break;
619  case AV_SAMPLE_FMT_S16:
620  UPDATE_STATS(I, int16_t, *src, / (double)INT16_MAX, *src);
621  break;
622  }
623 
624  return 0;
625 }
626 
627 static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
628 {
629  AVFilterContext *ctx = inlink->dst;
630  AudioStatsContext *s = ctx->priv;
631  AVDictionary **metadata = &buf->metadata;
632 
633  if (s->reset_count > 0) {
634  if (s->nb_frames >= s->reset_count) {
635  reset_stats(s);
636  s->nb_frames = 0;
637  }
638  s->nb_frames++;
639  }
640 
641  if (s->used == 0)
642  s->used = buf->nb_samples > 0;
643  ff_filter_execute(ctx, filter_channel, buf, NULL,
644  FFMIN(inlink->ch_layout.nb_channels, ff_filter_get_nb_threads(ctx)));
645 
646  if (s->metadata)
647  set_metadata(s, metadata);
648 
649  return ff_filter_frame(inlink->dst->outputs[0], buf);
650 }
651 
652 static void print_stats(AVFilterContext *ctx)
653 {
654  AudioStatsContext *s = ctx->priv;
655  uint64_t mask = 0, imask = 0xFFFFFFFFFFFFFFFF, min_count = 0, max_count = 0, nb_samples = 0, noise_floor_count = 0;
656  uint64_t nb_nans = 0, nb_infs = 0, nb_denormals = 0;
657  double min_runs = 0, max_runs = 0,
658  min = DBL_MAX, max =-DBL_MAX, min_diff = DBL_MAX, max_diff = 0,
659  nmin = DBL_MAX, nmax =-DBL_MAX,
660  max_sigma_x = 0,
661  diff1_sum_x2 = 0,
662  diff1_sum = 0,
663  sigma_x2 = 0,
664  noise_floor = 0,
665  entropy = 0,
666  min_sigma_x2 = DBL_MAX,
667  max_sigma_x2 =-DBL_MAX;
668  AVRational depth;
669  int c;
670 
671  for (c = 0; c < s->nb_channels; c++) {
672  ChannelStats *p = &s->chstats[c];
673 
674  if (p->nb_samples == 0 && !s->used)
675  continue;
676 
677  if (p->nb_samples < s->tc_samples)
678  p->min_sigma_x2 = p->max_sigma_x2 = p->sigma_x2 / p->nb_samples;
679 
680  min = FFMIN(min, p->min);
681  max = FFMAX(max, p->max);
682  nmin = FFMIN(nmin, p->nmin);
683  nmax = FFMAX(nmax, p->nmax);
684  min_diff = FFMIN(min_diff, p->min_diff);
685  max_diff = FFMAX(max_diff, p->max_diff);
686  diff1_sum_x2 += p->diff1_sum_x2;
687  diff1_sum += p->diff1_sum;
688  min_sigma_x2 = FFMIN(min_sigma_x2, p->min_sigma_x2);
689  max_sigma_x2 = FFMAX(max_sigma_x2, p->max_sigma_x2);
690  sigma_x2 += p->sigma_x2;
691  noise_floor = FFMAX(noise_floor, p->noise_floor);
692  p->entropy = calc_entropy(s, p);
693  entropy += p->entropy;
694  min_count += p->min_count;
695  max_count += p->max_count;
696  noise_floor_count += p->noise_floor_count;
697  min_runs += p->min_runs;
698  max_runs += p->max_runs;
699  mask |= p->mask;
700  imask &= p->imask;
701  nb_samples += p->nb_samples;
702  nb_nans += p->nb_nans;
703  nb_infs += p->nb_infs;
704  nb_denormals += p->nb_denormals;
705  if (fabs(p->sigma_x) > fabs(max_sigma_x))
706  max_sigma_x = p->sigma_x;
707 
708  if (s->measure_perchannel != MEASURE_NONE)
709  av_log(ctx, AV_LOG_INFO, "Channel: %d\n", c + 1);
710  if (s->measure_perchannel & MEASURE_DC_OFFSET)
711  av_log(ctx, AV_LOG_INFO, "DC offset: %f\n", p->sigma_x / p->nb_samples);
712  if (s->measure_perchannel & MEASURE_MIN_LEVEL)
713  av_log(ctx, AV_LOG_INFO, "Min level: %f\n", p->min);
714  if (s->measure_perchannel & MEASURE_MAX_LEVEL)
715  av_log(ctx, AV_LOG_INFO, "Max level: %f\n", p->max);
716  if (s->measure_perchannel & MEASURE_MIN_DIFFERENCE)
717  av_log(ctx, AV_LOG_INFO, "Min difference: %f\n", p->min_diff);
718  if (s->measure_perchannel & MEASURE_MAX_DIFFERENCE)
719  av_log(ctx, AV_LOG_INFO, "Max difference: %f\n", p->max_diff);
720  if (s->measure_perchannel & MEASURE_MEAN_DIFFERENCE)
721  av_log(ctx, AV_LOG_INFO, "Mean difference: %f\n", p->diff1_sum / (p->nb_samples - 1));
722  if (s->measure_perchannel & MEASURE_RMS_DIFFERENCE)
723  av_log(ctx, AV_LOG_INFO, "RMS difference: %f\n", sqrt(p->diff1_sum_x2 / (p->nb_samples - 1)));
724  if (s->measure_perchannel & MEASURE_PEAK_LEVEL)
725  av_log(ctx, AV_LOG_INFO, "Peak level dB: %f\n", LINEAR_TO_DB(FFMAX(-p->nmin, p->nmax)));
726  if (s->measure_perchannel & MEASURE_RMS_LEVEL)
727  av_log(ctx, AV_LOG_INFO, "RMS level dB: %f\n", LINEAR_TO_DB(sqrt(p->sigma_x2 / p->nb_samples)));
728  if (s->measure_perchannel & MEASURE_RMS_PEAK)
729  av_log(ctx, AV_LOG_INFO, "RMS peak dB: %f\n", LINEAR_TO_DB(sqrt(p->max_sigma_x2)));
730  if (s->measure_perchannel & MEASURE_RMS_TROUGH)
731  if (p->min_sigma_x2 != 1)
732  av_log(ctx, AV_LOG_INFO, "RMS trough dB: %f\n",LINEAR_TO_DB(sqrt(p->min_sigma_x2)));
733  if (s->measure_perchannel & MEASURE_CREST_FACTOR)
734  av_log(ctx, AV_LOG_INFO, "Crest factor: %f\n", p->sigma_x2 ? FFMAX(-p->nmin, p->nmax) / sqrt(p->sigma_x2 / p->nb_samples) : 1);
735  if (s->measure_perchannel & MEASURE_FLAT_FACTOR)
736  av_log(ctx, AV_LOG_INFO, "Flat factor: %f\n", LINEAR_TO_DB((p->min_runs + p->max_runs) / (p->min_count + p->max_count)));
737  if (s->measure_perchannel & MEASURE_PEAK_COUNT)
738  av_log(ctx, AV_LOG_INFO, "Peak count: %"PRId64"\n", p->min_count + p->max_count);
739  if (s->measure_perchannel & MEASURE_NOISE_FLOOR)
740  av_log(ctx, AV_LOG_INFO, "Noise floor dB: %f\n", LINEAR_TO_DB(p->noise_floor));
741  if (s->measure_perchannel & MEASURE_NOISE_FLOOR_COUNT)
742  av_log(ctx, AV_LOG_INFO, "Noise floor count: %"PRId64"\n", p->noise_floor_count);
743  if (s->measure_perchannel & MEASURE_ENTROPY)
744  av_log(ctx, AV_LOG_INFO, "Entropy: %f\n", p->entropy);
745  if (s->measure_perchannel & MEASURE_BIT_DEPTH) {
746  bit_depth(s, p->mask, p->imask, &depth);
747  av_log(ctx, AV_LOG_INFO, "Bit depth: %u/%u\n", depth.num, depth.den);
748  }
749  if (s->measure_perchannel & MEASURE_DYNAMIC_RANGE)
750  av_log(ctx, AV_LOG_INFO, "Dynamic range: %f\n", LINEAR_TO_DB(2 * FFMAX(FFABS(p->min), FFABS(p->max))/ p->min_non_zero));
751  if (s->measure_perchannel & MEASURE_ZERO_CROSSINGS)
752  av_log(ctx, AV_LOG_INFO, "Zero crossings: %"PRId64"\n", p->zero_runs);
753  if (s->measure_perchannel & MEASURE_ZERO_CROSSINGS_RATE)
754  av_log(ctx, AV_LOG_INFO, "Zero crossings rate: %f\n", p->zero_runs/(double)p->nb_samples);
755  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_NANS)
756  av_log(ctx, AV_LOG_INFO, "Number of NaNs: %"PRId64"\n", p->nb_nans);
757  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_INFS)
758  av_log(ctx, AV_LOG_INFO, "Number of Infs: %"PRId64"\n", p->nb_infs);
759  if ((s->is_float || s->is_double) && s->measure_perchannel & MEASURE_NUMBER_OF_DENORMALS)
760  av_log(ctx, AV_LOG_INFO, "Number of denormals: %"PRId64"\n", p->nb_denormals);
761  }
762 
763  if (nb_samples == 0 && !s->used)
764  return;
765 
766  if (s->measure_overall != MEASURE_NONE)
767  av_log(ctx, AV_LOG_INFO, "Overall\n");
768  if (s->measure_overall & MEASURE_DC_OFFSET)
769  av_log(ctx, AV_LOG_INFO, "DC offset: %f\n", max_sigma_x / (nb_samples / s->nb_channels));
770  if (s->measure_overall & MEASURE_MIN_LEVEL)
771  av_log(ctx, AV_LOG_INFO, "Min level: %f\n", min);
772  if (s->measure_overall & MEASURE_MAX_LEVEL)
773  av_log(ctx, AV_LOG_INFO, "Max level: %f\n", max);
774  if (s->measure_overall & MEASURE_MIN_DIFFERENCE)
775  av_log(ctx, AV_LOG_INFO, "Min difference: %f\n", min_diff);
776  if (s->measure_overall & MEASURE_MAX_DIFFERENCE)
777  av_log(ctx, AV_LOG_INFO, "Max difference: %f\n", max_diff);
778  if (s->measure_overall & MEASURE_MEAN_DIFFERENCE)
779  av_log(ctx, AV_LOG_INFO, "Mean difference: %f\n", diff1_sum / (nb_samples - s->nb_channels));
780  if (s->measure_overall & MEASURE_RMS_DIFFERENCE)
781  av_log(ctx, AV_LOG_INFO, "RMS difference: %f\n", sqrt(diff1_sum_x2 / (nb_samples - s->nb_channels)));
782  if (s->measure_overall & MEASURE_PEAK_LEVEL)
783  av_log(ctx, AV_LOG_INFO, "Peak level dB: %f\n", LINEAR_TO_DB(FFMAX(-nmin, nmax)));
784  if (s->measure_overall & MEASURE_RMS_LEVEL)
785  av_log(ctx, AV_LOG_INFO, "RMS level dB: %f\n", LINEAR_TO_DB(sqrt(sigma_x2 / nb_samples)));
786  if (s->measure_overall & MEASURE_RMS_PEAK)
787  av_log(ctx, AV_LOG_INFO, "RMS peak dB: %f\n", LINEAR_TO_DB(sqrt(max_sigma_x2)));
788  if (s->measure_overall & MEASURE_RMS_TROUGH)
789  if (min_sigma_x2 != 1)
790  av_log(ctx, AV_LOG_INFO, "RMS trough dB: %f\n", LINEAR_TO_DB(sqrt(min_sigma_x2)));
791  if (s->measure_overall & MEASURE_FLAT_FACTOR)
792  av_log(ctx, AV_LOG_INFO, "Flat factor: %f\n", LINEAR_TO_DB((min_runs + max_runs) / (min_count + max_count)));
793  if (s->measure_overall & MEASURE_PEAK_COUNT)
794  av_log(ctx, AV_LOG_INFO, "Peak count: %f\n", (min_count + max_count) / (double)s->nb_channels);
795  if (s->measure_overall & MEASURE_NOISE_FLOOR)
796  av_log(ctx, AV_LOG_INFO, "Noise floor dB: %f\n", LINEAR_TO_DB(noise_floor));
797  if (s->measure_overall & MEASURE_NOISE_FLOOR_COUNT)
798  av_log(ctx, AV_LOG_INFO, "Noise floor count: %f\n", noise_floor_count / (double)s->nb_channels);
799  if (s->measure_overall & MEASURE_ENTROPY)
800  av_log(ctx, AV_LOG_INFO, "Entropy: %f\n", entropy / (double)s->nb_channels);
801  if (s->measure_overall & MEASURE_BIT_DEPTH) {
802  bit_depth(s, mask, imask, &depth);
803  av_log(ctx, AV_LOG_INFO, "Bit depth: %u/%u\n", depth.num, depth.den);
804  }
805  if (s->measure_overall & MEASURE_NUMBER_OF_SAMPLES)
806  av_log(ctx, AV_LOG_INFO, "Number of samples: %"PRId64"\n", nb_samples / s->nb_channels);
807  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_NANS)
808  av_log(ctx, AV_LOG_INFO, "Number of NaNs: %f\n", nb_nans / (float)s->nb_channels);
809  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_INFS)
810  av_log(ctx, AV_LOG_INFO, "Number of Infs: %f\n", nb_infs / (float)s->nb_channels);
811  if ((s->is_float || s->is_double) && s->measure_overall & MEASURE_NUMBER_OF_DENORMALS)
812  av_log(ctx, AV_LOG_INFO, "Number of denormals: %f\n", nb_denormals / (float)s->nb_channels);
813 }
814 
815 static av_cold void uninit(AVFilterContext *ctx)
816 {
817  AudioStatsContext *s = ctx->priv;
818 
819  if (s->nb_channels)
820  print_stats(ctx);
821  if (s->chstats) {
822  for (int i = 0; i < s->nb_channels; i++) {
823  ChannelStats *p = &s->chstats[i];
824 
825  av_freep(&p->win_samples);
826  }
827  }
828  av_freep(&s->chstats);
829 }
830 
831 static const AVFilterPad astats_inputs[] = {
832  {
833  .name = "default",
834  .type = AVMEDIA_TYPE_AUDIO,
835  .filter_frame = filter_frame,
836  },
837 };
838 
839 static const AVFilterPad astats_outputs[] = {
840  {
841  .name = "default",
842  .type = AVMEDIA_TYPE_AUDIO,
843  .config_props = config_output,
844  },
845 };
846 
847 const AVFilter ff_af_astats = {
848  .name = "astats",
849  .description = NULL_IF_CONFIG_SMALL("Show time domain statistics about audio frames."),
850  .priv_size = sizeof(AudioStatsContext),
851  .priv_class = &astats_class,
852  .uninit = uninit,
853  FILTER_INPUTS(astats_inputs),
854  FILTER_OUTPUTS(astats_outputs),
855  FILTER_SAMPLEFMTS(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P,
856  AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32P,
857  AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64P,
858  AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
859  AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBLP),
860  .flags = AVFILTER_FLAG_SLICE_THREADS | AVFILTER_FLAG_METADATA_ONLY,
861 };
ChannelStats::nmax
double nmax
Definition: af_astats.c:71
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:66
bit_depth
static void bit_depth(AudioStatsContext *s, uint64_t mask, uint64_t imask, AVRational *depth)
Definition: af_astats.c:227
av_clip
#define av_clip
Definition: common.h:95
AudioStatsContext::nb_channels
int nb_channels
Definition: af_astats.c:97
set_meta
static void set_meta(AVDictionary **metadata, int chan, const char *key, const char *fmt, double val)
Definition: af_astats.c:381
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
opt.h
MEASURE_PEAK_COUNT
#define MEASURE_PEAK_COUNT
Definition: af_astats.c:49
AudioStatsContext::is_double
int is_double
Definition: af_astats.c:109
ff_filter_frame
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:969
astats_options
static const AVOption astats_options[]
Definition: af_astats.c:115
AudioStatsContext::maxbitdepth
int maxbitdepth
Definition: af_astats.c:105
inlink
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
Definition: filter_design.txt:212
MEASURE_RMS_TROUGH
#define MEASURE_RMS_TROUGH
Definition: af_astats.c:46
MEASURE_MIN_LEVEL
#define MEASURE_MIN_LEVEL
Definition: af_astats.c:37
ChannelStats::diff1_sum
double diff1_sum
Definition: af_astats.c:75
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:330
MEASURE_ALL
#define MEASURE_ALL
Definition: af_astats.c:33
MEASURE_NONE
#define MEASURE_NONE
Definition: af_astats.c:34
AudioStatsContext::is_float
int is_float
Definition: af_astats.c:108
AVOption
AVOption.
Definition: opt.h:251
ChannelStats::diff1_sum_x2
double diff1_sum_x2
Definition: af_astats.c:76
AV_SAMPLE_FMT_S32P
@ AV_SAMPLE_FMT_S32P
signed 32 bits, planar
Definition: samplefmt.h:65
data
const char data[16]
Definition: mxf.c:146
config_output
static int config_output(AVFilterLink *outlink)
Definition: af_astats.c:194
AudioStatsContext::tc_samples
uint64_t tc_samples
Definition: af_astats.c:98
float.h
max
#define max(a, b)
Definition: cuda_runtime.h:33
AVDictionary
Definition: dict.c:32
FFMAX
#define FFMAX(a, b)
Definition: macros.h:47
AVFilter::name
const char * name
Filter name.
Definition: avfilter.h:165
ChannelStats::max_sigma_x2
double max_sigma_x2
Definition: af_astats.c:69
AVChannelLayout::nb_channels
int nb_channels
Number of channels in this layout.
Definition: channel_layout.h:311
reset_stats
static void reset_stats(AudioStatsContext *s)
Definition: af_astats.c:153
ChannelStats::last_non_zero
double last_non_zero
Definition: af_astats.c:66
update_stat
static void update_stat(AudioStatsContext *s, ChannelStats *p, double d, double nd, int64_t i)
Definition: af_astats.c:265
AVFilterContext::priv
void * priv
private data for use by the filter
Definition: avfilter.h:407
FFSIGN
#define FFSIGN(a)
Definition: common.h:65
ChannelStats::nb_nans
uint64_t nb_nans
Definition: af_astats.c:82
update_minmax
static void update_minmax(AudioStatsContext *s, ChannelStats *p, double d)
Definition: af_astats.c:257
val
static double val(void *priv, double ch)
Definition: aeval.c:77
AVFrame::ch_layout
AVChannelLayout ch_layout
Channel layout of the audio data.
Definition: frame.h:723
type
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf type
Definition: writing_filters.txt:86
AV_SAMPLE_FMT_S64P
@ AV_SAMPLE_FMT_S64P
signed 64 bits, planar
Definition: samplefmt.h:69
ChannelStats::ehistogram
uint64_t ehistogram[HISTOGRAM_SIZE]
Definition: af_astats.c:87
OFFSET
#define OFFSET(x)
Definition: af_astats.c:112
ChannelStats::avg_sigma_x2
double avg_sigma_x2
Definition: af_astats.c:69
AVRational::num
int num
Numerator.
Definition: rational.h:59
ChannelStats::mask
uint64_t mask
Definition: af_astats.c:77
AVFilterPad
A filter pad used for either input or output.
Definition: internal.h:49
ChannelStats::nmin
double nmin
Definition: af_astats.c:71
LINEAR_TO_DB
#define LINEAR_TO_DB(x)
Definition: af_astats.c:395
lrint
#define lrint
Definition: tablegen.h:53
av_cold
#define av_cold
Definition: attributes.h:90
ChannelStats::win_pos
int win_pos
Definition: af_astats.c:88
mask
static const uint16_t mask[17]
Definition: lzw.c:38
ChannelStats::entropy
double entropy
Definition: af_astats.c:91
ChannelStats::last
double last
Definition: af_astats.c:65
AudioStatsContext::mult
double mult
Definition: af_astats.c:100
s
#define s(width, name)
Definition: cbs_vp9.c:256
AudioStatsContext::measure_overall
int measure_overall
Definition: af_astats.c:107
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Definition: opt.h:227
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
filter_frame
static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
Definition: af_astats.c:627
calc_entropy
static double calc_entropy(AudioStatsContext *s, ChannelStats *p)
Definition: af_astats.c:243
ctx
AVFormatContext * ctx
Definition: movenc.c:48
UPDATE_STATS
#define UPDATE_STATS(planar, type, sample, normalizer_suffix, int_sample)
Definition: af_astats.c:573
channels
channels
Definition: aptx.h:31
ChannelStats::max
double max
Definition: af_astats.c:70
update_double_stat
static void update_double_stat(AudioStatsContext *s, ChannelStats *p, double d)
Definition: af_astats.c:372
ChannelStats::max_index
int max_index
Definition: af_astats.c:89
HISTOGRAM_MAX
#define HISTOGRAM_MAX
Definition: af_astats.c:31
update_float_stat
static void update_float_stat(AudioStatsContext *s, ChannelStats *p, float d)
Definition: af_astats.c:363
key
const char * key
Definition: hwcontext_opencl.c:174
uninit
static av_cold void uninit(AVFilterContext *ctx)
Definition: af_astats.c:815
NAN
#define NAN
Definition: mathematics.h:64
ChannelStats::min_diff
double min_diff
Definition: af_astats.c:74
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition: internal.h:194
arg
const char * arg
Definition: jacosubdec.c:67
FFABS
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:64
ChannelStats::min_count
uint64_t min_count
Definition: af_astats.c:78
MEASURE_RMS_DIFFERENCE
#define MEASURE_RMS_DIFFERENCE
Definition: af_astats.c:42
ChannelStats::sigma_x
double sigma_x
Definition: af_astats.c:68
ChannelStats::noise_floor_count
uint64_t noise_floor_count
Definition: af_astats.c:79
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
result
and forward the result(frame or status change) to the corresponding input. If nothing is possible
fabs
static __device__ float fabs(float a)
Definition: cuda_runtime.h:182
NULL
#define NULL
Definition: coverity.c:32
MEASURE_NOISE_FLOOR_COUNT
#define MEASURE_NOISE_FLOOR_COUNT
Definition: af_astats.c:59
ChannelStats::win_samples
double * win_samples
Definition: af_astats.c:85
MEASURE_ZERO_CROSSINGS_RATE
#define MEASURE_ZERO_CROSSINGS_RATE
Definition: af_astats.c:53
MEASURE_DC_OFFSET
#define MEASURE_DC_OFFSET
Definition: af_astats.c:36
AVRational
Rational number (pair of numerator and denominator).
Definition: rational.h:58
isnan
#define isnan(x)
Definition: libm.h:340
ChannelStats::min_runs
double min_runs
Definition: af_astats.c:73
AudioStatsContext::used
int used
Definition: af_astats.c:102
double
double
Definition: af_crystalizer.c:132
exp
int8_t exp
Definition: eval.c:72
index
int index
Definition: gxfenc.c:89
c
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
FLAGS
#define FLAGS
Definition: af_astats.c:113
MEASURE_FLAT_FACTOR
#define MEASURE_FLAT_FACTOR
Definition: af_astats.c:48
AudioStatsContext::measure_perchannel
int measure_perchannel
Definition: af_astats.c:106
HISTOGRAM_SIZE
#define HISTOGRAM_SIZE
Definition: af_astats.c:30
AudioStatsContext
Definition: af_astats.c:94
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:115
P
#define P
MEASURE_DYNAMIC_RANGE
#define MEASURE_DYNAMIC_RANGE
Definition: af_astats.c:51
ChannelStats::nb_samples
uint64_t nb_samples
Definition: af_astats.c:81
MEASURE_NUMBER_OF_SAMPLES
#define MEASURE_NUMBER_OF_SAMPLES
Definition: af_astats.c:54
MEASURE_ENTROPY
#define MEASURE_ENTROPY
Definition: af_astats.c:60
MEASURE_RMS_PEAK
#define MEASURE_RMS_PEAK
Definition: af_astats.c:45
AV_SAMPLE_FMT_S16P
@ AV_SAMPLE_FMT_S16P
signed 16 bits, planar
Definition: samplefmt.h:64
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:191
internal.h
ChannelStats::max_diff
double max_diff
Definition: af_astats.c:74
astats_inputs
static const AVFilterPad astats_inputs[]
Definition: af_astats.c:831
ChannelStats::min_sigma_x2
double min_sigma_x2
Definition: af_astats.c:69
MEASURE_MEAN_DIFFERENCE
#define MEASURE_MEAN_DIFFERENCE
Definition: af_astats.c:41
AVFrame::nb_samples
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:410
ChannelStats
Definition: af_astats.c:64
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:269
MEASURE_NUMBER_OF_DENORMALS
#define MEASURE_NUMBER_OF_DENORMALS
Definition: af_astats.c:57
av_get_bytes_per_sample
int av_get_bytes_per_sample(enum AVSampleFormat sample_fmt)
Return number of bytes per sample.
Definition: samplefmt.c:108
ff_af_astats
const AVFilter ff_af_astats
Definition: af_astats.c:847
ChannelStats::min_run
double min_run
Definition: af_astats.c:72
filter_channel
static int filter_channel(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
Definition: af_astats.c:580
AVFrame::extended_data
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:391
ff_filter_get_nb_threads
int ff_filter_get_nb_threads(AVFilterContext *ctx)
Get number of threads for current filter instance.
Definition: avfilter.c:777
value
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default value
Definition: writing_filters.txt:86
MEASURE_PEAK_LEVEL
#define MEASURE_PEAK_LEVEL
Definition: af_astats.c:43
FFMIN
#define FFMIN(a, b)
Definition: macros.h:49
ChannelStats::max_run
double max_run
Definition: af_astats.c:72
AV_SAMPLE_FMT_S16
@ AV_SAMPLE_FMT_S16
signed 16 bits
Definition: samplefmt.h:58
set_metadata
static void set_metadata(AudioStatsContext *s, AVDictionary **metadata)
Definition: af_astats.c:397
AVFilterPad::name
const char * name
Pad name.
Definition: internal.h:55
av_calloc
void * av_calloc(size_t nmemb, size_t size)
Definition: mem.c:262
ChannelStats::max_runs
double max_runs
Definition: af_astats.c:73
log2
#define log2(x)
Definition: libm.h:404
AVFilter
Filter definition.
Definition: avfilter.h:161
MEASURE_RMS_LEVEL
#define MEASURE_RMS_LEVEL
Definition: af_astats.c:44
MEASURE_NUMBER_OF_NANS
#define MEASURE_NUMBER_OF_NANS
Definition: af_astats.c:55
MEASURE_BIT_DEPTH
#define MEASURE_BIT_DEPTH
Definition: af_astats.c:50
AudioStatsContext::nb_frames
int nb_frames
Definition: af_astats.c:104
AVRational::den
int den
Denominator.
Definition: rational.h:60
AV_OPT_TYPE_INT
@ AV_OPT_TYPE_INT
Definition: opt.h:225
avfilter.h
AVFrame::metadata
AVDictionary * metadata
metadata.
Definition: frame.h:639
AVFILTER_FLAG_METADATA_ONLY
#define AVFILTER_FLAG_METADATA_ONLY
The filter is a "metadata" filter - it does not modify the frame data in any way.
Definition: avfilter.h:133
AV_SAMPLE_FMT_DBLP
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition: samplefmt.h:67
AudioStatsContext::reset_count
int reset_count
Definition: af_astats.c:103
ChannelStats::histogram
uint64_t histogram[HISTOGRAM_SIZE]
Definition: af_astats.c:86
samples
Filter the word “frame” indicates either a video frame or a group of audio samples
Definition: filter_design.txt:8
AVFilterContext
An instance of a filter.
Definition: avfilter.h:392
MEASURE_NUMBER_OF_INFS
#define MEASURE_NUMBER_OF_INFS
Definition: af_astats.c:56
AVFILTER_FLAG_SLICE_THREADS
#define AVFILTER_FLAG_SLICE_THREADS
The filter supports multithreading by splitting frames into multiple parts and processing them concur...
Definition: avfilter.h:117
ChannelStats::noise_floor
double noise_floor
Definition: af_astats.c:90
MEASURE_NOISE_FLOOR
#define MEASURE_NOISE_FLOOR
Definition: af_astats.c:58
audio.h
AudioStatsContext::metadata
int metadata
Definition: af_astats.c:101
llrint
#define llrint(x)
Definition: libm.h:394
ChannelStats::nb_denormals
uint64_t nb_denormals
Definition: af_astats.c:84
ChannelStats::sigma_x2
double sigma_x2
Definition: af_astats.c:68
AVFILTER_DEFINE_CLASS
AVFILTER_DEFINE_CLASS(astats)
print_stats
static void print_stats(AVFilterContext *ctx)
Definition: af_astats.c:652
MEASURE_MAX_LEVEL
#define MEASURE_MAX_LEVEL
Definition: af_astats.c:38
MEASURE_MIN_DIFFERENCE
#define MEASURE_MIN_DIFFERENCE
Definition: af_astats.c:39
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Definition: opt.h:244
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition: internal.h:195
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:34
av_dict_set
int av_dict_set(AVDictionary **pm, const char *key, const char *value, int flags)
Set the given entry in *pm, overwriting an existing entry.
Definition: dict.c:86
src
INIT_CLIP pixel * src
Definition: h264pred_template.c:418
AudioStatsContext::chstats
ChannelStats * chstats
Definition: af_astats.c:96
d
d
Definition: ffmpeg_filter.c:156
AV_OPT_TYPE_FLAGS
@ AV_OPT_TYPE_FLAGS
Definition: opt.h:224
int32_t
int32_t
Definition: audioconvert.c:56
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
ChannelStats::zero_runs
uint64_t zero_runs
Definition: af_astats.c:80
MEASURE_CREST_FACTOR
#define MEASURE_CREST_FACTOR
Definition: af_astats.c:47
ChannelStats::imask
uint64_t imask
Definition: af_astats.c:77
AV_SAMPLE_FMT_DBL
@ AV_SAMPLE_FMT_DBL
double
Definition: samplefmt.h:61
ff_filter_execute
static av_always_inline int ff_filter_execute(AVFilterContext *ctx, avfilter_action_func *func, void *arg, int *ret, int nb_jobs)
Definition: internal.h:146
ChannelStats::min
double min
Definition: af_astats.c:70
AudioStatsContext::time_constant
double time_constant
Definition: af_astats.c:99
AV_SAMPLE_FMT_S32
@ AV_SAMPLE_FMT_S32
signed 32 bits
Definition: samplefmt.h:59
ChannelStats::max_count
uint64_t max_count
Definition: af_astats.c:78
ChannelStats::min_non_zero
double min_non_zero
Definition: af_astats.c:67
AV_OPT_TYPE_CONST
@ AV_OPT_TYPE_CONST
Definition: opt.h:234
snprintf
#define snprintf
Definition: snprintf.h:34
AV_SAMPLE_FMT_FLT
@ AV_SAMPLE_FMT_FLT
float
Definition: samplefmt.h:60
MEASURE_MAX_DIFFERENCE
#define MEASURE_MAX_DIFFERENCE
Definition: af_astats.c:40
astats_outputs
static const AVFilterPad astats_outputs[]
Definition: af_astats.c:839
ChannelStats::nb_infs
uint64_t nb_infs
Definition: af_astats.c:83
AV_SAMPLE_FMT_S64
@ AV_SAMPLE_FMT_S64
signed 64 bits
Definition: samplefmt.h:68
av_clipd
av_clipd
Definition: af_crystalizer.c:132
MEASURE_ZERO_CROSSINGS
#define MEASURE_ZERO_CROSSINGS
Definition: af_astats.c:52
FILTER_SAMPLEFMTS
#define FILTER_SAMPLEFMTS(...)
Definition: internal.h:182
min
float min
Definition: vorbis_enc_data.h:429
Generated on Tue Feb 28 2023 21:33:47 for FFmpeg by   doxygen 1.8.17