FFmpeg  4.4
af_dynaudnorm.c
Go to the documentation of this file.
1 /*
2  * Dynamic Audio Normalizer
3  * Copyright (c) 2015 LoRd_MuldeR <mulder2@gmx.de>. Some rights reserved.
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 /**
23  * @file
24  * Dynamic Audio Normalizer
25  */
26 
27 #include <float.h>
28 
29 #include "libavutil/avassert.h"
30 #include "libavutil/opt.h"
31 
32 #define MIN_FILTER_SIZE 3
33 #define MAX_FILTER_SIZE 301
34 
35 #define FF_BUFQUEUE_SIZE (MAX_FILTER_SIZE + 1)
37 
38 #include "audio.h"
39 #include "avfilter.h"
40 #include "filters.h"
41 #include "internal.h"
42 
43 typedef struct local_gain {
44  double max_gain;
45  double threshold;
46 } local_gain;
47 
48 typedef struct cqueue {
49  double *elements;
50  int size;
51  int max_size;
53 } cqueue;
54 
56  const AVClass *class;
57 
58  struct FFBufQueue queue;
59 
60  int frame_len;
66 
67  double peak_value;
69  double target_rms;
71  double threshold;
75  double *weights;
76 
77  int channels;
78  int eof;
79  int64_t pts;
80 
85 
88 
89 #define OFFSET(x) offsetof(DynamicAudioNormalizerContext, x)
90 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
91 
92 static const AVOption dynaudnorm_options[] = {
93  { "framelen", "set the frame length in msec", OFFSET(frame_len_msec), AV_OPT_TYPE_INT, {.i64 = 500}, 10, 8000, FLAGS },
94  { "f", "set the frame length in msec", OFFSET(frame_len_msec), AV_OPT_TYPE_INT, {.i64 = 500}, 10, 8000, FLAGS },
95  { "gausssize", "set the filter size", OFFSET(filter_size), AV_OPT_TYPE_INT, {.i64 = 31}, 3, 301, FLAGS },
96  { "g", "set the filter size", OFFSET(filter_size), AV_OPT_TYPE_INT, {.i64 = 31}, 3, 301, FLAGS },
97  { "peak", "set the peak value", OFFSET(peak_value), AV_OPT_TYPE_DOUBLE, {.dbl = 0.95}, 0.0, 1.0, FLAGS },
98  { "p", "set the peak value", OFFSET(peak_value), AV_OPT_TYPE_DOUBLE, {.dbl = 0.95}, 0.0, 1.0, FLAGS },
99  { "maxgain", "set the max amplification", OFFSET(max_amplification), AV_OPT_TYPE_DOUBLE, {.dbl = 10.0}, 1.0, 100.0, FLAGS },
100  { "m", "set the max amplification", OFFSET(max_amplification), AV_OPT_TYPE_DOUBLE, {.dbl = 10.0}, 1.0, 100.0, FLAGS },
101  { "targetrms", "set the target RMS", OFFSET(target_rms), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 1.0, FLAGS },
102  { "r", "set the target RMS", OFFSET(target_rms), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 1.0, FLAGS },
103  { "coupling", "set channel coupling", OFFSET(channels_coupled), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, FLAGS },
104  { "n", "set channel coupling", OFFSET(channels_coupled), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, FLAGS },
105  { "correctdc", "set DC correction", OFFSET(dc_correction), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
106  { "c", "set DC correction", OFFSET(dc_correction), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
107  { "altboundary", "set alternative boundary mode", OFFSET(alt_boundary_mode), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
108  { "b", "set alternative boundary mode", OFFSET(alt_boundary_mode), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS },
109  { "compress", "set the compress factor", OFFSET(compress_factor), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 30.0, FLAGS },
110  { "s", "set the compress factor", OFFSET(compress_factor), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 30.0, FLAGS },
111  { "threshold", "set the threshold value", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 1.0, FLAGS },
112  { "t", "set the threshold value", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 0.0}, 0.0, 1.0, FLAGS },
113  { NULL }
114 };
115 
117 
119 {
121 
122  if (!(s->filter_size & 1)) {
123  av_log(ctx, AV_LOG_WARNING, "filter size %d is invalid. Changing to an odd value.\n", s->filter_size);
124  s->filter_size |= 1;
125  }
126 
127  return 0;
128 }
129 
131 {
134  static const enum AVSampleFormat sample_fmts[] = {
137  };
138  int ret;
139 
141  if (!layouts)
142  return AVERROR(ENOMEM);
144  if (ret < 0)
145  return ret;
146 
148  if (!formats)
149  return AVERROR(ENOMEM);
151  if (ret < 0)
152  return ret;
153 
155  if (!formats)
156  return AVERROR(ENOMEM);
158 }
159 
160 static inline int frame_size(int sample_rate, int frame_len_msec)
161 {
162  const int frame_size = lrint((double)sample_rate * (frame_len_msec / 1000.0));
163  return frame_size + (frame_size % 2);
164 }
165 
166 static cqueue *cqueue_create(int size, int max_size)
167 {
168  cqueue *q;
169 
170  if (max_size < size)
171  return NULL;
172 
173  q = av_malloc(sizeof(cqueue));
174  if (!q)
175  return NULL;
176 
177  q->max_size = max_size;
178  q->size = size;
179  q->nb_elements = 0;
180 
181  q->elements = av_malloc_array(max_size, sizeof(double));
182  if (!q->elements) {
183  av_free(q);
184  return NULL;
185  }
186 
187  return q;
188 }
189 
190 static void cqueue_free(cqueue *q)
191 {
192  if (q)
193  av_free(q->elements);
194  av_free(q);
195 }
196 
197 static int cqueue_size(cqueue *q)
198 {
199  return q->nb_elements;
200 }
201 
202 static int cqueue_empty(cqueue *q)
203 {
204  return q->nb_elements <= 0;
205 }
206 
207 static int cqueue_enqueue(cqueue *q, double element)
208 {
210 
211  q->elements[q->nb_elements] = element;
212  q->nb_elements++;
213 
214  return 0;
215 }
216 
217 static double cqueue_peek(cqueue *q, int index)
218 {
219  av_assert2(index < q->nb_elements);
220  return q->elements[index];
221 }
222 
223 static int cqueue_dequeue(cqueue *q, double *element)
224 {
226 
227  *element = q->elements[0];
228  memmove(&q->elements[0], &q->elements[1], (q->nb_elements - 1) * sizeof(double));
229  q->nb_elements--;
230 
231  return 0;
232 }
233 
234 static int cqueue_pop(cqueue *q)
235 {
237 
238  memmove(&q->elements[0], &q->elements[1], (q->nb_elements - 1) * sizeof(double));
239  q->nb_elements--;
240 
241  return 0;
242 }
243 
244 static void cqueue_resize(cqueue *q, int new_size)
245 {
246  av_assert2(q->max_size >= new_size);
247  av_assert2(MIN_FILTER_SIZE <= new_size);
248 
249  if (new_size > q->nb_elements) {
250  const int side = (new_size - q->nb_elements) / 2;
251 
252  memmove(q->elements + side, q->elements, sizeof(double) * q->nb_elements);
253  for (int i = 0; i < side; i++)
254  q->elements[i] = q->elements[side];
255  q->nb_elements = new_size - 1 - side;
256  } else {
257  int count = (q->size - new_size + 1) / 2;
258 
259  while (count-- > 0)
260  cqueue_pop(q);
261  }
262 
263  q->size = new_size;
264 }
265 
267 {
268  double total_weight = 0.0;
269  const double sigma = (((s->filter_size / 2.0) - 1.0) / 3.0) + (1.0 / 3.0);
270  double adjust;
271  int i;
272 
273  // Pre-compute constants
274  const int offset = s->filter_size / 2;
275  const double c1 = 1.0 / (sigma * sqrt(2.0 * M_PI));
276  const double c2 = 2.0 * sigma * sigma;
277 
278  // Compute weights
279  for (i = 0; i < s->filter_size; i++) {
280  const int x = i - offset;
281 
282  s->weights[i] = c1 * exp(-x * x / c2);
283  total_weight += s->weights[i];
284  }
285 
286  // Adjust weights
287  adjust = 1.0 / total_weight;
288  for (i = 0; i < s->filter_size; i++) {
289  s->weights[i] *= adjust;
290  }
291 }
292 
294 {
296  int c;
297 
298  av_freep(&s->prev_amplification_factor);
299  av_freep(&s->dc_correction_value);
300  av_freep(&s->compress_threshold);
301 
302  for (c = 0; c < s->channels; c++) {
303  if (s->gain_history_original)
304  cqueue_free(s->gain_history_original[c]);
305  if (s->gain_history_minimum)
306  cqueue_free(s->gain_history_minimum[c]);
307  if (s->gain_history_smoothed)
308  cqueue_free(s->gain_history_smoothed[c]);
309  if (s->threshold_history)
310  cqueue_free(s->threshold_history[c]);
311  }
312 
313  av_freep(&s->gain_history_original);
314  av_freep(&s->gain_history_minimum);
315  av_freep(&s->gain_history_smoothed);
316  av_freep(&s->threshold_history);
317 
318  cqueue_free(s->is_enabled);
319  s->is_enabled = NULL;
320 
321  av_freep(&s->weights);
322 
323  ff_bufqueue_discard_all(&s->queue);
324 }
325 
326 static int config_input(AVFilterLink *inlink)
327 {
328  AVFilterContext *ctx = inlink->dst;
330  int c;
331 
332  uninit(ctx);
333 
334  s->channels = inlink->channels;
335  s->frame_len = frame_size(inlink->sample_rate, s->frame_len_msec);
336  av_log(ctx, AV_LOG_DEBUG, "frame len %d\n", s->frame_len);
337 
338  s->prev_amplification_factor = av_malloc_array(inlink->channels, sizeof(*s->prev_amplification_factor));
339  s->dc_correction_value = av_calloc(inlink->channels, sizeof(*s->dc_correction_value));
340  s->compress_threshold = av_calloc(inlink->channels, sizeof(*s->compress_threshold));
341  s->gain_history_original = av_calloc(inlink->channels, sizeof(*s->gain_history_original));
342  s->gain_history_minimum = av_calloc(inlink->channels, sizeof(*s->gain_history_minimum));
343  s->gain_history_smoothed = av_calloc(inlink->channels, sizeof(*s->gain_history_smoothed));
344  s->threshold_history = av_calloc(inlink->channels, sizeof(*s->threshold_history));
345  s->weights = av_malloc_array(MAX_FILTER_SIZE, sizeof(*s->weights));
346  s->is_enabled = cqueue_create(s->filter_size, MAX_FILTER_SIZE);
347  if (!s->prev_amplification_factor || !s->dc_correction_value ||
348  !s->compress_threshold ||
349  !s->gain_history_original || !s->gain_history_minimum ||
350  !s->gain_history_smoothed || !s->threshold_history ||
351  !s->is_enabled || !s->weights)
352  return AVERROR(ENOMEM);
353 
354  for (c = 0; c < inlink->channels; c++) {
355  s->prev_amplification_factor[c] = 1.0;
356 
357  s->gain_history_original[c] = cqueue_create(s->filter_size, MAX_FILTER_SIZE);
358  s->gain_history_minimum[c] = cqueue_create(s->filter_size, MAX_FILTER_SIZE);
359  s->gain_history_smoothed[c] = cqueue_create(s->filter_size, MAX_FILTER_SIZE);
360  s->threshold_history[c] = cqueue_create(s->filter_size, MAX_FILTER_SIZE);
361 
362  if (!s->gain_history_original[c] || !s->gain_history_minimum[c] ||
363  !s->gain_history_smoothed[c] || !s->threshold_history[c])
364  return AVERROR(ENOMEM);
365  }
366 
368 
369  return 0;
370 }
371 
372 static inline double fade(double prev, double next, int pos, int length)
373 {
374  const double step_size = 1.0 / length;
375  const double f0 = 1.0 - (step_size * (pos + 1.0));
376  const double f1 = 1.0 - f0;
377  return f0 * prev + f1 * next;
378 }
379 
380 static inline double pow_2(const double value)
381 {
382  return value * value;
383 }
384 
385 static inline double bound(const double threshold, const double val)
386 {
387  const double CONST = 0.8862269254527580136490837416705725913987747280611935; //sqrt(PI) / 2.0
388  return erf(CONST * (val / threshold)) * threshold;
389 }
390 
392 {
393  double max = DBL_EPSILON;
394  int c, i;
395 
396  if (channel == -1) {
397  for (c = 0; c < frame->channels; c++) {
398  double *data_ptr = (double *)frame->extended_data[c];
399 
400  for (i = 0; i < frame->nb_samples; i++)
401  max = FFMAX(max, fabs(data_ptr[i]));
402  }
403  } else {
404  double *data_ptr = (double *)frame->extended_data[channel];
405 
406  for (i = 0; i < frame->nb_samples; i++)
407  max = FFMAX(max, fabs(data_ptr[i]));
408  }
409 
410  return max;
411 }
412 
414 {
415  double rms_value = 0.0;
416  int c, i;
417 
418  if (channel == -1) {
419  for (c = 0; c < frame->channels; c++) {
420  const double *data_ptr = (double *)frame->extended_data[c];
421 
422  for (i = 0; i < frame->nb_samples; i++) {
423  rms_value += pow_2(data_ptr[i]);
424  }
425  }
426 
427  rms_value /= frame->nb_samples * frame->channels;
428  } else {
429  const double *data_ptr = (double *)frame->extended_data[channel];
430  for (i = 0; i < frame->nb_samples; i++) {
431  rms_value += pow_2(data_ptr[i]);
432  }
433 
434  rms_value /= frame->nb_samples;
435  }
436 
437  return FFMAX(sqrt(rms_value), DBL_EPSILON);
438 }
439 
441  int channel)
442 {
443  const double peak_magnitude = find_peak_magnitude(frame, channel);
444  const double maximum_gain = s->peak_value / peak_magnitude;
445  const double rms_gain = s->target_rms > DBL_EPSILON ? (s->target_rms / compute_frame_rms(frame, channel)) : DBL_MAX;
446  local_gain gain;
447 
448  gain.threshold = peak_magnitude > s->threshold;
449  gain.max_gain = bound(s->max_amplification, FFMIN(maximum_gain, rms_gain));
450 
451  return gain;
452 }
453 
454 static double minimum_filter(cqueue *q)
455 {
456  double min = DBL_MAX;
457  int i;
458 
459  for (i = 0; i < cqueue_size(q); i++) {
460  min = FFMIN(min, cqueue_peek(q, i));
461  }
462 
463  return min;
464 }
465 
467 {
468  double result = 0.0, tsum = 0.0;
469  int i;
470 
471  for (i = 0; i < cqueue_size(q); i++) {
472  tsum += cqueue_peek(tq, i) * s->weights[i];
473  result += cqueue_peek(q, i) * s->weights[i] * cqueue_peek(tq, i);
474  }
475 
476  if (tsum == 0.0)
477  result = 1.0;
478 
479  return result;
480 }
481 
483  local_gain gain)
484 {
485  if (cqueue_empty(s->gain_history_original[channel])) {
486  const int pre_fill_size = s->filter_size / 2;
487  const double initial_value = s->alt_boundary_mode ? gain.max_gain : s->peak_value;
488 
489  s->prev_amplification_factor[channel] = initial_value;
490 
491  while (cqueue_size(s->gain_history_original[channel]) < pre_fill_size) {
492  cqueue_enqueue(s->gain_history_original[channel], initial_value);
493  cqueue_enqueue(s->threshold_history[channel], gain.threshold);
494  }
495  }
496 
497  cqueue_enqueue(s->gain_history_original[channel], gain.max_gain);
498 
499  while (cqueue_size(s->gain_history_original[channel]) >= s->filter_size) {
500  double minimum;
501 
502  if (cqueue_empty(s->gain_history_minimum[channel])) {
503  const int pre_fill_size = s->filter_size / 2;
504  double initial_value = s->alt_boundary_mode ? cqueue_peek(s->gain_history_original[channel], 0) : 1.0;
505  int input = pre_fill_size;
506 
507  while (cqueue_size(s->gain_history_minimum[channel]) < pre_fill_size) {
508  input++;
509  initial_value = FFMIN(initial_value, cqueue_peek(s->gain_history_original[channel], input));
510  cqueue_enqueue(s->gain_history_minimum[channel], initial_value);
511  }
512  }
513 
514  minimum = minimum_filter(s->gain_history_original[channel]);
515 
516  cqueue_enqueue(s->gain_history_minimum[channel], minimum);
517 
518  cqueue_enqueue(s->threshold_history[channel], gain.threshold);
519 
520  cqueue_pop(s->gain_history_original[channel]);
521  }
522 
523  while (cqueue_size(s->gain_history_minimum[channel]) >= s->filter_size) {
524  double smoothed, limit;
525 
526  smoothed = gaussian_filter(s, s->gain_history_minimum[channel], s->threshold_history[channel]);
527  limit = cqueue_peek(s->gain_history_original[channel], 0);
528  smoothed = FFMIN(smoothed, limit);
529 
530  cqueue_enqueue(s->gain_history_smoothed[channel], smoothed);
531 
532  cqueue_pop(s->gain_history_minimum[channel]);
533  cqueue_pop(s->threshold_history[channel]);
534  }
535 }
536 
537 static inline double update_value(double new, double old, double aggressiveness)
538 {
539  av_assert0((aggressiveness >= 0.0) && (aggressiveness <= 1.0));
540  return aggressiveness * new + (1.0 - aggressiveness) * old;
541 }
542 
544 {
545  const double diff = 1.0 / frame->nb_samples;
546  int is_first_frame = cqueue_empty(s->gain_history_original[0]);
547  int c, i;
548 
549  for (c = 0; c < s->channels; c++) {
550  double *dst_ptr = (double *)frame->extended_data[c];
551  double current_average_value = 0.0;
552  double prev_value;
553 
554  for (i = 0; i < frame->nb_samples; i++)
555  current_average_value += dst_ptr[i] * diff;
556 
557  prev_value = is_first_frame ? current_average_value : s->dc_correction_value[c];
558  s->dc_correction_value[c] = is_first_frame ? current_average_value : update_value(current_average_value, s->dc_correction_value[c], 0.1);
559 
560  for (i = 0; i < frame->nb_samples; i++) {
561  dst_ptr[i] -= fade(prev_value, s->dc_correction_value[c], i, frame->nb_samples);
562  }
563  }
564 }
565 
566 static double setup_compress_thresh(double threshold)
567 {
568  if ((threshold > DBL_EPSILON) && (threshold < (1.0 - DBL_EPSILON))) {
569  double current_threshold = threshold;
570  double step_size = 1.0;
571 
572  while (step_size > DBL_EPSILON) {
573  while ((llrint((current_threshold + step_size) * (UINT64_C(1) << 63)) >
574  llrint(current_threshold * (UINT64_C(1) << 63))) &&
575  (bound(current_threshold + step_size, 1.0) <= threshold)) {
576  current_threshold += step_size;
577  }
578 
579  step_size /= 2.0;
580  }
581 
582  return current_threshold;
583  } else {
584  return threshold;
585  }
586 }
587 
589  AVFrame *frame, int channel)
590 {
591  double variance = 0.0;
592  int i, c;
593 
594  if (channel == -1) {
595  for (c = 0; c < s->channels; c++) {
596  const double *data_ptr = (double *)frame->extended_data[c];
597 
598  for (i = 0; i < frame->nb_samples; i++) {
599  variance += pow_2(data_ptr[i]); // Assume that MEAN is *zero*
600  }
601  }
602  variance /= (s->channels * frame->nb_samples) - 1;
603  } else {
604  const double *data_ptr = (double *)frame->extended_data[channel];
605 
606  for (i = 0; i < frame->nb_samples; i++) {
607  variance += pow_2(data_ptr[i]); // Assume that MEAN is *zero*
608  }
609  variance /= frame->nb_samples - 1;
610  }
611 
612  return FFMAX(sqrt(variance), DBL_EPSILON);
613 }
614 
616 {
617  int is_first_frame = cqueue_empty(s->gain_history_original[0]);
618  int c, i;
619 
620  if (s->channels_coupled) {
621  const double standard_deviation = compute_frame_std_dev(s, frame, -1);
622  const double current_threshold = FFMIN(1.0, s->compress_factor * standard_deviation);
623 
624  const double prev_value = is_first_frame ? current_threshold : s->compress_threshold[0];
625  double prev_actual_thresh, curr_actual_thresh;
626  s->compress_threshold[0] = is_first_frame ? current_threshold : update_value(current_threshold, s->compress_threshold[0], (1.0/3.0));
627 
628  prev_actual_thresh = setup_compress_thresh(prev_value);
629  curr_actual_thresh = setup_compress_thresh(s->compress_threshold[0]);
630 
631  for (c = 0; c < s->channels; c++) {
632  double *const dst_ptr = (double *)frame->extended_data[c];
633  for (i = 0; i < frame->nb_samples; i++) {
634  const double localThresh = fade(prev_actual_thresh, curr_actual_thresh, i, frame->nb_samples);
635  dst_ptr[i] = copysign(bound(localThresh, fabs(dst_ptr[i])), dst_ptr[i]);
636  }
637  }
638  } else {
639  for (c = 0; c < s->channels; c++) {
640  const double standard_deviation = compute_frame_std_dev(s, frame, c);
641  const double current_threshold = setup_compress_thresh(FFMIN(1.0, s->compress_factor * standard_deviation));
642 
643  const double prev_value = is_first_frame ? current_threshold : s->compress_threshold[c];
644  double prev_actual_thresh, curr_actual_thresh;
645  double *dst_ptr;
646  s->compress_threshold[c] = is_first_frame ? current_threshold : update_value(current_threshold, s->compress_threshold[c], 1.0/3.0);
647 
648  prev_actual_thresh = setup_compress_thresh(prev_value);
649  curr_actual_thresh = setup_compress_thresh(s->compress_threshold[c]);
650 
651  dst_ptr = (double *)frame->extended_data[c];
652  for (i = 0; i < frame->nb_samples; i++) {
653  const double localThresh = fade(prev_actual_thresh, curr_actual_thresh, i, frame->nb_samples);
654  dst_ptr[i] = copysign(bound(localThresh, fabs(dst_ptr[i])), dst_ptr[i]);
655  }
656  }
657  }
658 }
659 
661 {
662  if (s->dc_correction) {
664  }
665 
666  if (s->compress_factor > DBL_EPSILON) {
668  }
669 
670  if (s->channels_coupled) {
671  const local_gain gain = get_max_local_gain(s, frame, -1);
672  int c;
673 
674  for (c = 0; c < s->channels; c++)
675  update_gain_history(s, c, gain);
676  } else {
677  int c;
678 
679  for (c = 0; c < s->channels; c++)
681  }
682 }
683 
685 {
686  int c, i;
687 
688  for (c = 0; c < s->channels; c++) {
689  double *dst_ptr = (double *)frame->extended_data[c];
690  double current_amplification_factor;
691 
692  cqueue_dequeue(s->gain_history_smoothed[c], &current_amplification_factor);
693 
694  for (i = 0; i < frame->nb_samples && enabled; i++) {
695  const double amplification_factor = fade(s->prev_amplification_factor[c],
696  current_amplification_factor, i,
697  frame->nb_samples);
698 
699  dst_ptr[i] *= amplification_factor;
700  }
701 
702  s->prev_amplification_factor[c] = current_amplification_factor;
703  }
704 }
705 
706 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
707 {
708  AVFilterContext *ctx = inlink->dst;
710  AVFilterLink *outlink = ctx->outputs[0];
711  int ret = 1;
712 
713  while (((s->queue.available >= s->filter_size) ||
714  (s->eof && s->queue.available)) &&
715  !cqueue_empty(s->gain_history_smoothed[0])) {
716  AVFrame *out = ff_bufqueue_get(&s->queue);
717  double is_enabled;
718 
719  cqueue_dequeue(s->is_enabled, &is_enabled);
720 
721  amplify_frame(s, out, is_enabled > 0.);
722  s->pts = out->pts + out->nb_samples;
723  ret = ff_filter_frame(outlink, out);
724  }
725 
727  analyze_frame(s, in);
728  if (!s->eof) {
729  ff_bufqueue_add(ctx, &s->queue, in);
730  cqueue_enqueue(s->is_enabled, !ctx->is_disabled);
731  } else {
732  av_frame_free(&in);
733  }
734 
735  return ret;
736 }
737 
739  AVFilterLink *outlink)
740 {
741  AVFrame *out = ff_get_audio_buffer(outlink, s->frame_len);
742  int c, i;
743 
744  if (!out)
745  return AVERROR(ENOMEM);
746 
747  for (c = 0; c < s->channels; c++) {
748  double *dst_ptr = (double *)out->extended_data[c];
749 
750  for (i = 0; i < out->nb_samples; i++) {
751  dst_ptr[i] = s->alt_boundary_mode ? DBL_EPSILON : ((s->target_rms > DBL_EPSILON) ? FFMIN(s->peak_value, s->target_rms) : s->peak_value);
752  if (s->dc_correction) {
753  dst_ptr[i] *= ((i % 2) == 1) ? -1 : 1;
754  dst_ptr[i] += s->dc_correction_value[c];
755  }
756  }
757  }
758 
759  return filter_frame(inlink, out);
760 }
761 
762 static int flush(AVFilterLink *outlink)
763 {
764  AVFilterContext *ctx = outlink->src;
766  int ret = 0;
767 
768  if (!cqueue_empty(s->gain_history_smoothed[0])) {
769  ret = flush_buffer(s, ctx->inputs[0], outlink);
770  } else if (s->queue.available) {
771  AVFrame *out = ff_bufqueue_get(&s->queue);
772 
773  s->pts = out->pts + out->nb_samples;
774  ret = ff_filter_frame(outlink, out);
775  }
776 
777  return ret;
778 }
779 
781 {
782  AVFilterLink *inlink = ctx->inputs[0];
783  AVFilterLink *outlink = ctx->outputs[0];
785  AVFrame *in = NULL;
786  int ret = 0, status;
787  int64_t pts;
788 
789  FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
790 
791  if (!s->eof) {
792  ret = ff_inlink_consume_samples(inlink, s->frame_len, s->frame_len, &in);
793  if (ret < 0)
794  return ret;
795  if (ret > 0) {
796  ret = filter_frame(inlink, in);
797  if (ret <= 0)
798  return ret;
799  }
800 
801  if (ff_inlink_check_available_samples(inlink, s->frame_len) > 0) {
803  return 0;
804  }
805  }
806 
807  if (!s->eof && ff_inlink_acknowledge_status(inlink, &status, &pts)) {
808  if (status == AVERROR_EOF)
809  s->eof = 1;
810  }
811 
812  if (s->eof && s->queue.available)
813  return flush(outlink);
814 
815  if (s->eof && !s->queue.available) {
816  ff_outlink_set_status(outlink, AVERROR_EOF, s->pts);
817  return 0;
818  }
819 
820  if (!s->eof)
821  FF_FILTER_FORWARD_WANTED(outlink, inlink);
822 
823  return FFERROR_NOT_READY;
824 }
825 
826 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
827  char *res, int res_len, int flags)
828 {
830  AVFilterLink *inlink = ctx->inputs[0];
831  int prev_filter_size = s->filter_size;
832  int ret;
833 
834  ret = ff_filter_process_command(ctx, cmd, args, res, res_len, flags);
835  if (ret < 0)
836  return ret;
837 
838  s->filter_size |= 1;
839  if (prev_filter_size != s->filter_size) {
841 
842  for (int c = 0; c < s->channels; c++) {
843  cqueue_resize(s->gain_history_original[c], s->filter_size);
844  cqueue_resize(s->gain_history_minimum[c], s->filter_size);
845  cqueue_resize(s->threshold_history[c], s->filter_size);
846  }
847  }
848 
849  s->frame_len = frame_size(inlink->sample_rate, s->frame_len_msec);
850 
851  return 0;
852 }
853 
855  {
856  .name = "default",
857  .type = AVMEDIA_TYPE_AUDIO,
858  .config_props = config_input,
859  },
860  { NULL }
861 };
862 
864  {
865  .name = "default",
866  .type = AVMEDIA_TYPE_AUDIO,
867  },
868  { NULL }
869 };
870 
872  .name = "dynaudnorm",
873  .description = NULL_IF_CONFIG_SMALL("Dynamic Audio Normalizer."),
874  .query_formats = query_formats,
875  .priv_size = sizeof(DynamicAudioNormalizerContext),
876  .init = init,
877  .uninit = uninit,
878  .activate = activate,
881  .priv_class = &dynaudnorm_class,
884 };
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:925
static double val(void *priv, double ch)
Definition: aeval.c:76
static const AVFilterPad inputs[]
Definition: af_acontrast.c:193
static const AVFilterPad outputs[]
Definition: af_acontrast.c:203
static int flush(AVFilterLink *outlink)
static int cqueue_pop(cqueue *q)
static void cqueue_resize(cqueue *q, int new_size)
AVFilter ff_af_dynaudnorm
static cqueue * cqueue_create(int size, int max_size)
static double bound(const double threshold, const double val)
static void perform_compression(DynamicAudioNormalizerContext *s, AVFrame *frame)
static int cqueue_empty(cqueue *q)
static double update_value(double new, double old, double aggressiveness)
static double compute_frame_rms(AVFrame *frame, int channel)
static int flush_buffer(DynamicAudioNormalizerContext *s, AVFilterLink *inlink, AVFilterLink *outlink)
static int cqueue_enqueue(cqueue *q, double element)
static const AVOption dynaudnorm_options[]
Definition: af_dynaudnorm.c:92
static void perform_dc_correction(DynamicAudioNormalizerContext *s, AVFrame *frame)
static void analyze_frame(DynamicAudioNormalizerContext *s, AVFrame *frame)
static double setup_compress_thresh(double threshold)
static local_gain get_max_local_gain(DynamicAudioNormalizerContext *s, AVFrame *frame, int channel)
static double compute_frame_std_dev(DynamicAudioNormalizerContext *s, AVFrame *frame, int channel)
static int cqueue_size(cqueue *q)
static double minimum_filter(cqueue *q)
static double cqueue_peek(cqueue *q, int index)
static void update_gain_history(DynamicAudioNormalizerContext *s, int channel, local_gain gain)
static int query_formats(AVFilterContext *ctx)
static void cqueue_free(cqueue *q)
static double find_peak_magnitude(AVFrame *frame, int channel)
static int config_input(AVFilterLink *inlink)
#define FLAGS
Definition: af_dynaudnorm.c:90
static int frame_size(int sample_rate, int frame_len_msec)
static double pow_2(const double value)
static void init_gaussian_filter(DynamicAudioNormalizerContext *s)
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
#define MIN_FILTER_SIZE
Definition: af_dynaudnorm.c:32
static const AVFilterPad avfilter_af_dynaudnorm_outputs[]
static int cqueue_dequeue(cqueue *q, double *element)
AVFILTER_DEFINE_CLASS(dynaudnorm)
static int process_command(AVFilterContext *ctx, const char *cmd, const char *args, char *res, int res_len, int flags)
static int activate(AVFilterContext *ctx)
static av_cold int init(AVFilterContext *ctx)
static av_cold void uninit(AVFilterContext *ctx)
static double fade(double prev, double next, int pos, int length)
static double gaussian_filter(DynamicAudioNormalizerContext *s, cqueue *q, cqueue *tq)
#define MAX_FILTER_SIZE
Definition: af_dynaudnorm.c:33
#define OFFSET(x)
Definition: af_dynaudnorm.c:89
static void amplify_frame(DynamicAudioNormalizerContext *s, AVFrame *frame, int enabled)
static const AVFilterPad avfilter_af_dynaudnorm_inputs[]
#define av_cold
Definition: attributes.h:88
AVFrame * ff_get_audio_buffer(AVFilterLink *link, int nb_samples)
Request an audio samples buffer with a specific set of permissions.
Definition: audio.c:86
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(const int16_t *) pi >> 8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(const int32_t *) pi >> 24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) #define SET_CONV_FUNC_GROUP(ofmt, ifmt) static void set_generic_function(AudioConvert *ac) { } void ff_audio_convert_free(AudioConvert **ac) { if(! *ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);} AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, int sample_rate, int apply_map) { AudioConvert *ac;int in_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) return NULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method !=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt) > 2) { ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc) { av_free(ac);return NULL;} return ac;} in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar) { ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar ? ac->channels :1;} else if(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;else ac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);return ac;} int ff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in) { int use_generic=1;int len=in->nb_samples;int p;if(ac->dc) { av_log(ac->avr, AV_LOG_TRACE, "%d samples - audio_convert: %s to %s (dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));return ff_convert_dither(ac-> in
simple assert() macros that are a bit more flexible than ISO C assert().
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
Definition: avassert.h:64
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
int ff_inlink_acknowledge_status(AVFilterLink *link, int *rstatus, int64_t *rpts)
Test and acknowledge the change of status on the link.
Definition: avfilter.c:1447
int ff_inlink_check_available_samples(AVFilterLink *link, unsigned min)
Test if enough samples are available on the link.
Definition: avfilter.c:1477
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1094
int ff_filter_process_command(AVFilterContext *ctx, const char *cmd, const char *arg, char *res, int res_len, int flags)
Generic processing of user supplied commands that are set in the same way as the filter options.
Definition: avfilter.c:882
int ff_inlink_consume_samples(AVFilterLink *link, unsigned min, unsigned max, AVFrame **rframe)
Take samples from the link's FIFO and update the link's stats.
Definition: avfilter.c:1511
void ff_filter_set_ready(AVFilterContext *filter, unsigned priority)
Mark a filter ready and schedule it for activation.
Definition: avfilter.c:193
Main libavfilter public API header.
static void ff_bufqueue_add(void *log, struct FFBufQueue *queue, AVFrame *buf)
Add a buffer to the queue.
Definition: bufferqueue.h:71
static void ff_bufqueue_discard_all(struct FFBufQueue *queue)
Unref and remove all buffers from the queue.
Definition: bufferqueue.h:111
static AVFrame * ff_bufqueue_get(struct FFBufQueue *queue)
Get the first buffer from the queue and remove it.
Definition: bufferqueue.h:98
#define flags(name, subs,...)
Definition: cbs_av1.c:561
#define s(width, name)
Definition: cbs_vp9.c:257
#define FFMIN(a, b)
Definition: common.h:105
#define FFMAX(a, b)
Definition: common.h:103
#define NULL
Definition: coverity.c:32
static __device__ float fabs(float a)
Definition: cuda_runtime.h:182
#define max(a, b)
Definition: cuda_runtime.h:33
static AVFrame * frame
static float minimum(float src0, float src1)
channel
Use these values when setting the channel map with ebur128_set_channel().
Definition: ebur128.h:39
double value
Definition: eval.c:98
int8_t exp
Definition: eval.c:72
sample_rate
#define FF_FILTER_FORWARD_WANTED(outlink, inlink)
Forward the frame_wanted_out flag from an output link to an input link.
Definition: filters.h:254
static void ff_outlink_set_status(AVFilterLink *link, int status, int64_t pts)
Set the status field of a link from the source filter.
Definition: filters.h:189
#define FFERROR_NOT_READY
Filters implementation helper functions.
Definition: filters.h:34
#define FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink)
Forward the status on an output link to an input link.
Definition: filters.h:199
AVFilterChannelLayouts * ff_all_channel_counts(void)
Construct an AVFilterChannelLayouts coding for any channel layout, with known or unknown disposition.
Definition: formats.c:436
int ff_set_common_formats(AVFilterContext *ctx, AVFilterFormats *formats)
A helper for query_formats() which sets all links to the same list of formats.
Definition: formats.c:587
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:286
int ff_set_common_samplerates(AVFilterContext *ctx, AVFilterFormats *samplerates)
Definition: formats.c:575
int ff_set_common_channel_layouts(AVFilterContext *ctx, AVFilterChannelLayouts *channel_layouts)
A helper for query_formats() which sets all links to the same list of channel layouts/sample rates.
Definition: formats.c:568
AVFilterFormats * ff_all_samplerates(void)
Definition: formats.c:421
@ AV_OPT_TYPE_INT
Definition: opt.h:225
@ AV_OPT_TYPE_DOUBLE
Definition: opt.h:227
@ AV_OPT_TYPE_BOOL
Definition: opt.h:242
#define AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL
Same as AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC, except that the filter will have its filter_frame() c...
Definition: avfilter.h:134
#define AVERROR_EOF
End of file.
Definition: error.h:55
#define AVERROR(e)
Definition: error.h:43
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:203
int av_frame_make_writable(AVFrame *frame)
Ensure that the frame data is writable, avoiding data copy if possible.
Definition: frame.c:611
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:215
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:200
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
Definition: mem.c:245
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
@ AV_SAMPLE_FMT_NONE
Definition: samplefmt.h:59
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition: samplefmt.h:70
int index
Definition: gxfenc.c:89
for(j=16;j >0;--j)
int i
Definition: input.c:407
common internal API header
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:117
static double erf(double z)
erf function Algorithm taken from the Boost project, source: http://www.boost.org/doc/libs/1_46_1/boo...
Definition: libm.h:121
#define llrint(x)
Definition: libm.h:394
static av_always_inline double copysign(double x, double y)
Definition: libm.h:68
#define M_PI
Definition: mathematics.h:52
static int adjust(int x, int size)
Definition: mobiclip.c:515
enum MovChannelLayoutTag * layouts
Definition: mov_chan.c:434
static const uint64_t c2
Definition: murmur3.c:52
static const uint64_t c1
Definition: murmur3.c:51
AVOptions.
formats
Definition: signature.h:48
unsigned int pos
Definition: spdifenc.c:412
Describe the class of an AVClass context structure.
Definition: log.h:67
A list of supported channel layouts.
Definition: formats.h:86
An instance of a filter.
Definition: avfilter.h:341
A list of supported formats for one end of a filter link.
Definition: formats.h:65
A filter pad used for either input or output.
Definition: internal.h:54
const char * name
Pad name.
Definition: internal.h:60
Filter definition.
Definition: avfilter.h:145
const char * name
Filter name.
Definition: avfilter.h:149
This structure describes decoded (raw) audio or video data.
Definition: frame.h:318
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:384
int channels
number of audio channels, only used for audio.
Definition: frame.h:624
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:365
AVOption.
Definition: opt.h:248
Structure holding the queue.
Definition: bufferqueue.h:49
double * elements
Definition: af_dynaudnorm.c:49
int size
Definition: af_dynaudnorm.c:50
int max_size
Definition: af_dynaudnorm.c:51
int nb_elements
Definition: af_dynaudnorm.c:52
double threshold
Definition: af_dynaudnorm.c:45
double max_gain
Definition: af_dynaudnorm.c:44
#define lrint
Definition: tablegen.h:53
#define av_free(p)
#define av_malloc_array(a, b)
#define av_freep(p)
#define av_malloc(s)
#define av_log(a,...)
FILE * out
Definition: movenc.c:54
AVFormatContext * ctx
Definition: movenc.c:48
static int64_t pts
int size
#define CONST(name, help, val, unit)
Definition: vf_bwdif.c:373
static av_always_inline int diff(const uint32_t a, const uint32_t b)
static const uint8_t offset[127][2]
Definition: vf_spp.c:107
float min
static double c[64]