FFmpeg
resample.c
Go to the documentation of this file.
1 /*
2  * audio resampling
3  * Copyright (c) 2004-2012 Michael Niedermayer <michaelni@gmx.at>
4  * bessel function: Copyright (c) 2006 Xiaogang Zhang
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * FFmpeg is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with FFmpeg; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 /**
24  * @file
25  * audio resampling
26  * @author Michael Niedermayer <michaelni@gmx.at>
27  */
28 
29 #include "libavutil/avassert.h"
30 #include "resample.h"
31 
32 static inline double eval_poly(const double *coeff, int size, double x) {
33  double sum = coeff[size-1];
34  int i;
35  for (i = size-2; i >= 0; --i) {
36  sum *= x;
37  sum += coeff[i];
38  }
39  return sum;
40 }
41 
42 /**
43  * 0th order modified bessel function of the first kind.
44  * Algorithm taken from the Boost project, source:
45  * https://searchcode.com/codesearch/view/14918379/
46  * Use, modification and distribution are subject to the
47  * Boost Software License, Version 1.0 (see notice below).
48  * Boost Software License - Version 1.0 - August 17th, 2003
49 Permission is hereby granted, free of charge, to any person or organization
50 obtaining a copy of the software and accompanying documentation covered by
51 this license (the "Software") to use, reproduce, display, distribute,
52 execute, and transmit the Software, and to prepare derivative works of the
53 Software, and to permit third-parties to whom the Software is furnished to
54 do so, all subject to the following:
55 
56 The copyright notices in the Software and this entire statement, including
57 the above license grant, this restriction and the following disclaimer,
58 must be included in all copies of the Software, in whole or in part, and
59 all derivative works of the Software, unless such copies or derivative
60 works are solely in the form of machine-executable object code generated by
61 a source language processor.
62 
63 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
64 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
65 FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
66 SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
67 FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
68 ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
69 DEALINGS IN THE SOFTWARE.
70  */
71 
72 static double bessel(double x) {
73 // Modified Bessel function of the first kind of order zero
74 // minimax rational approximations on intervals, see
75 // Blair and Edwards, Chalk River Report AECL-4928, 1974
76  static const double p1[] = {
77  -2.2335582639474375249e+15,
78  -5.5050369673018427753e+14,
79  -3.2940087627407749166e+13,
80  -8.4925101247114157499e+11,
81  -1.1912746104985237192e+10,
82  -1.0313066708737980747e+08,
83  -5.9545626019847898221e+05,
84  -2.4125195876041896775e+03,
85  -7.0935347449210549190e+00,
86  -1.5453977791786851041e-02,
87  -2.5172644670688975051e-05,
88  -3.0517226450451067446e-08,
89  -2.6843448573468483278e-11,
90  -1.5982226675653184646e-14,
91  -5.2487866627945699800e-18,
92  };
93  static const double q1[] = {
94  -2.2335582639474375245e+15,
95  7.8858692566751002988e+12,
96  -1.2207067397808979846e+10,
97  1.0377081058062166144e+07,
98  -4.8527560179962773045e+03,
99  1.0,
100  };
101  static const double p2[] = {
102  -2.2210262233306573296e-04,
103  1.3067392038106924055e-02,
104  -4.4700805721174453923e-01,
105  5.5674518371240761397e+00,
106  -2.3517945679239481621e+01,
107  3.1611322818701131207e+01,
108  -9.6090021968656180000e+00,
109  };
110  static const double q2[] = {
111  -5.5194330231005480228e-04,
112  3.2547697594819615062e-02,
113  -1.1151759188741312645e+00,
114  1.3982595353892851542e+01,
115  -6.0228002066743340583e+01,
116  8.5539563258012929600e+01,
117  -3.1446690275135491500e+01,
118  1.0,
119  };
120  double y, r, factor;
121  if (x == 0)
122  return 1.0;
123  x = fabs(x);
124  if (x <= 15) {
125  y = x * x;
126  return eval_poly(p1, FF_ARRAY_ELEMS(p1), y) / eval_poly(q1, FF_ARRAY_ELEMS(q1), y);
127  }
128  else {
129  y = 1 / x - 1.0 / 15;
130  r = eval_poly(p2, FF_ARRAY_ELEMS(p2), y) / eval_poly(q2, FF_ARRAY_ELEMS(q2), y);
131  factor = exp(x) / sqrt(x);
132  return factor * r;
133  }
134 }
135 
136 /**
137  * builds a polyphase filterbank.
138  * @param factor resampling factor
139  * @param scale wanted sum of coefficients for each filter
140  * @param filter_type filter type
141  * @param kaiser_beta kaiser window beta
142  * @return 0 on success, negative on error
143  */
144 static int build_filter(ResampleContext *c, void *filter, double factor, int tap_count, int alloc, int phase_count, int scale,
145  int filter_type, double kaiser_beta){
146  int ph, i;
147  int ph_nb = phase_count % 2 ? phase_count : phase_count / 2 + 1;
148  double x, y, w, t, s;
149  double *tab = av_malloc_array(tap_count+1, sizeof(*tab));
150  double *sin_lut = av_malloc_array(ph_nb, sizeof(*sin_lut));
151  const int center= (tap_count-1)/2;
152  double norm = 0;
153  int ret = AVERROR(ENOMEM);
154 
155  if (!tab || !sin_lut)
156  goto fail;
157 
158  av_assert0(tap_count == 1 || tap_count % 2 == 0);
159 
160  /* if upsampling, only need to interpolate, no filter */
161  if (factor > 1.0)
162  factor = 1.0;
163 
164  if (factor == 1.0) {
165  for (ph = 0; ph < ph_nb; ph++)
166  sin_lut[ph] = sin(M_PI * ph / phase_count) * (center & 1 ? 1 : -1);
167  }
168  for(ph = 0; ph < ph_nb; ph++) {
169  s = sin_lut[ph];
170  for(i=0;i<tap_count;i++) {
171  x = M_PI * ((double)(i - center) - (double)ph / phase_count) * factor;
172  if (x == 0) y = 1.0;
173  else if (factor == 1.0)
174  y = s / x;
175  else
176  y = sin(x) / x;
177  switch(filter_type){
178  case SWR_FILTER_TYPE_CUBIC:{
179  const float d= -0.5; //first order derivative = -0.5
180  x = fabs(((double)(i - center) - (double)ph / phase_count) * factor);
181  if(x<1.0) y= 1 - 3*x*x + 2*x*x*x + d*( -x*x + x*x*x);
182  else y= d*(-4 + 8*x - 5*x*x + x*x*x);
183  break;}
184  case SWR_FILTER_TYPE_BLACKMAN_NUTTALL:
185  w = 2.0*x / (factor*tap_count);
186  t = -cos(w);
187  y *= 0.3635819 - 0.4891775 * t + 0.1365995 * (2*t*t-1) - 0.0106411 * (4*t*t*t - 3*t);
188  break;
189  case SWR_FILTER_TYPE_KAISER:
190  w = 2.0*x / (factor*tap_count*M_PI);
191  y *= bessel(kaiser_beta*sqrt(FFMAX(1-w*w, 0)));
192  break;
193  default:
194  av_assert0(0);
195  }
196 
197  tab[i] = y;
198  s = -s;
199  if (!ph)
200  norm += y;
201  }
202 
203  /* normalize so that an uniform color remains the same */
204  switch(c->format){
205  case AV_SAMPLE_FMT_S16P:
206  for(i=0;i<tap_count;i++)
207  ((int16_t*)filter)[ph * alloc + i] = av_clip_int16(lrintf(tab[i] * scale / norm));
208  if (phase_count % 2) break;
209  for (i = 0; i < tap_count; i++)
210  ((int16_t*)filter)[(phase_count-ph) * alloc + tap_count-1-i] = ((int16_t*)filter)[ph * alloc + i];
211  break;
212  case AV_SAMPLE_FMT_S32P:
213  for(i=0;i<tap_count;i++)
214  ((int32_t*)filter)[ph * alloc + i] = av_clipl_int32(llrint(tab[i] * scale / norm));
215  if (phase_count % 2) break;
216  for (i = 0; i < tap_count; i++)
217  ((int32_t*)filter)[(phase_count-ph) * alloc + tap_count-1-i] = ((int32_t*)filter)[ph * alloc + i];
218  break;
219  case AV_SAMPLE_FMT_FLTP:
220  for(i=0;i<tap_count;i++)
221  ((float*)filter)[ph * alloc + i] = tab[i] * scale / norm;
222  if (phase_count % 2) break;
223  for (i = 0; i < tap_count; i++)
224  ((float*)filter)[(phase_count-ph) * alloc + tap_count-1-i] = ((float*)filter)[ph * alloc + i];
225  break;
226  case AV_SAMPLE_FMT_DBLP:
227  for(i=0;i<tap_count;i++)
228  ((double*)filter)[ph * alloc + i] = tab[i] * scale / norm;
229  if (phase_count % 2) break;
230  for (i = 0; i < tap_count; i++)
231  ((double*)filter)[(phase_count-ph) * alloc + tap_count-1-i] = ((double*)filter)[ph * alloc + i];
232  break;
233  }
234  }
235 #if 0
236  {
237 #define LEN 1024
238  int j,k;
239  double sine[LEN + tap_count];
240  double filtered[LEN];
241  double maxff=-2, minff=2, maxsf=-2, minsf=2;
242  for(i=0; i<LEN; i++){
243  double ss=0, sf=0, ff=0;
244  for(j=0; j<LEN+tap_count; j++)
245  sine[j]= cos(i*j*M_PI/LEN);
246  for(j=0; j<LEN; j++){
247  double sum=0;
248  ph=0;
249  for(k=0; k<tap_count; k++)
250  sum += filter[ph * tap_count + k] * sine[k+j];
251  filtered[j]= sum / (1<<FILTER_SHIFT);
252  ss+= sine[j + center] * sine[j + center];
253  ff+= filtered[j] * filtered[j];
254  sf+= sine[j + center] * filtered[j];
255  }
256  ss= sqrt(2*ss/LEN);
257  ff= sqrt(2*ff/LEN);
258  sf= 2*sf/LEN;
259  maxff= FFMAX(maxff, ff);
260  minff= FFMIN(minff, ff);
261  maxsf= FFMAX(maxsf, sf);
262  minsf= FFMIN(minsf, sf);
263  if(i%11==0){
264  av_log(NULL, AV_LOG_ERROR, "i:%4d ss:%f ff:%13.6e-%13.6e sf:%13.6e-%13.6e\n", i, ss, maxff, minff, maxsf, minsf);
265  minff=minsf= 2;
266  maxff=maxsf= -2;
267  }
268  }
269  }
270 #endif
271 
272  ret = 0;
273 fail:
274  av_free(tab);
275  av_free(sin_lut);
276  return ret;
277 }
278 
279 static void resample_free(ResampleContext **cc){
280  ResampleContext *c = *cc;
281  if(!c)
282  return;
283  av_freep(&c->filter_bank);
284  av_freep(cc);
285 }
286 
287 static ResampleContext *resample_init(ResampleContext *c, int out_rate, int in_rate, int filter_size, int phase_shift, int linear,
288  double cutoff0, enum AVSampleFormat format, enum SwrFilterType filter_type, double kaiser_beta,
289  double precision, int cheby, int exact_rational)
290 {
291  double cutoff = cutoff0? cutoff0 : 0.97;
292  double factor= FFMIN(out_rate * cutoff / in_rate, 1.0);
293  int phase_count= 1<<phase_shift;
294  int phase_count_compensation = phase_count;
295  int filter_length = FFMAX((int)ceil(filter_size/factor), 1);
296 
297  if (filter_length > 1)
298  filter_length = FFALIGN(filter_length, 2);
299 
300  if (exact_rational) {
301  int phase_count_exact, phase_count_exact_den;
302 
303  av_reduce(&phase_count_exact, &phase_count_exact_den, out_rate, in_rate, INT_MAX);
304  if (phase_count_exact <= phase_count) {
305  phase_count_compensation = phase_count_exact * (phase_count / phase_count_exact);
306  phase_count = phase_count_exact;
307  }
308  }
309 
310  if (!c || c->phase_count != phase_count || c->linear!=linear || c->factor != factor
311  || c->filter_length != filter_length || c->format != format
312  || c->filter_type != filter_type || c->kaiser_beta != kaiser_beta) {
313  resample_free(&c);
314  c = av_mallocz(sizeof(*c));
315  if (!c)
316  return NULL;
317 
318  c->format= format;
319 
320  c->felem_size= av_get_bytes_per_sample(c->format);
321 
322  switch(c->format){
323  case AV_SAMPLE_FMT_S16P:
324  c->filter_shift = 15;
325  break;
326  case AV_SAMPLE_FMT_S32P:
327  c->filter_shift = 30;
328  break;
329  case AV_SAMPLE_FMT_FLTP:
330  case AV_SAMPLE_FMT_DBLP:
331  c->filter_shift = 0;
332  break;
333  default:
334  av_log(NULL, AV_LOG_ERROR, "Unsupported sample format\n");
335  av_assert0(0);
336  }
337 
338  if (filter_size/factor > INT32_MAX/256) {
339  av_log(NULL, AV_LOG_ERROR, "Filter length too large\n");
340  goto error;
341  }
342 
343  c->phase_count = phase_count;
344  c->linear = linear;
345  c->factor = factor;
346  c->filter_length = filter_length;
347  c->filter_alloc = FFALIGN(c->filter_length, 8);
348  c->filter_bank = av_calloc(c->filter_alloc, (phase_count+1)*c->felem_size);
349  c->filter_type = filter_type;
350  c->kaiser_beta = kaiser_beta;
351  c->phase_count_compensation = phase_count_compensation;
352  if (!c->filter_bank)
353  goto error;
354  if (build_filter(c, (void*)c->filter_bank, factor, c->filter_length, c->filter_alloc, phase_count, 1<<c->filter_shift, filter_type, kaiser_beta))
355  goto error;
356  memcpy(c->filter_bank + (c->filter_alloc*phase_count+1)*c->felem_size, c->filter_bank, (c->filter_alloc-1)*c->felem_size);
357  memcpy(c->filter_bank + (c->filter_alloc*phase_count )*c->felem_size, c->filter_bank + (c->filter_alloc - 1)*c->felem_size, c->felem_size);
358  }
359 
360  c->compensation_distance= 0;
361  if(!av_reduce(&c->src_incr, &c->dst_incr, out_rate, in_rate * (int64_t)phase_count, INT32_MAX/2))
362  goto error;
363  while (c->dst_incr < (1<<20) && c->src_incr < (1<<20)) {
364  c->dst_incr *= 2;
365  c->src_incr *= 2;
366  }
367  c->ideal_dst_incr = c->dst_incr;
368  c->dst_incr_div = c->dst_incr / c->src_incr;
369  c->dst_incr_mod = c->dst_incr % c->src_incr;
370 
371  c->index= -phase_count*((c->filter_length-1)/2);
372  c->frac= 0;
373 
374  swri_resample_dsp_init(c);
375 
376  return c;
377 error:
378  av_freep(&c->filter_bank);
379  av_free(c);
380  return NULL;
381 }
382 
383 static int rebuild_filter_bank_with_compensation(ResampleContext *c)
384 {
385  uint8_t *new_filter_bank;
386  int new_src_incr, new_dst_incr;
387  int phase_count = c->phase_count_compensation;
388  int ret;
389 
390  if (phase_count == c->phase_count)
391  return 0;
392 
393  av_assert0(!c->frac && !c->dst_incr_mod);
394 
395  new_filter_bank = av_calloc(c->filter_alloc, (phase_count + 1) * c->felem_size);
396  if (!new_filter_bank)
397  return AVERROR(ENOMEM);
398 
399  ret = build_filter(c, new_filter_bank, c->factor, c->filter_length, c->filter_alloc,
400  phase_count, 1 << c->filter_shift, c->filter_type, c->kaiser_beta);
401  if (ret < 0) {
402  av_freep(&new_filter_bank);
403  return ret;
404  }
405  memcpy(new_filter_bank + (c->filter_alloc*phase_count+1)*c->felem_size, new_filter_bank, (c->filter_alloc-1)*c->felem_size);
406  memcpy(new_filter_bank + (c->filter_alloc*phase_count )*c->felem_size, new_filter_bank + (c->filter_alloc - 1)*c->felem_size, c->felem_size);
407 
408  if (!av_reduce(&new_src_incr, &new_dst_incr, c->src_incr,
409  c->dst_incr * (int64_t)(phase_count/c->phase_count), INT32_MAX/2))
410  {
411  av_freep(&new_filter_bank);
412  return AVERROR(EINVAL);
413  }
414 
415  c->src_incr = new_src_incr;
416  c->dst_incr = new_dst_incr;
417  while (c->dst_incr < (1<<20) && c->src_incr < (1<<20)) {
418  c->dst_incr *= 2;
419  c->src_incr *= 2;
420  }
421  c->ideal_dst_incr = c->dst_incr;
422  c->dst_incr_div = c->dst_incr / c->src_incr;
423  c->dst_incr_mod = c->dst_incr % c->src_incr;
424  c->index *= phase_count / c->phase_count;
425  c->phase_count = phase_count;
426  av_freep(&c->filter_bank);
427  c->filter_bank = new_filter_bank;
428  return 0;
429 }
430 
431 static int set_compensation(ResampleContext *c, int sample_delta, int compensation_distance){
432  int ret;
433 
434  if (compensation_distance && sample_delta) {
435  ret = rebuild_filter_bank_with_compensation(c);
436  if (ret < 0)
437  return ret;
438  }
439 
440  c->compensation_distance= compensation_distance;
441  if (compensation_distance)
442  c->dst_incr = c->ideal_dst_incr - c->ideal_dst_incr * (int64_t)sample_delta / compensation_distance;
443  else
444  c->dst_incr = c->ideal_dst_incr;
445 
446  c->dst_incr_div = c->dst_incr / c->src_incr;
447  c->dst_incr_mod = c->dst_incr % c->src_incr;
448 
449  return 0;
450 }
451 
452 static int multiple_resample(ResampleContext *c, AudioData *dst, int dst_size, AudioData *src, int src_size, int *consumed){
453  int i;
454  int av_unused mm_flags = av_get_cpu_flags();
455  int need_emms = c->format == AV_SAMPLE_FMT_S16P && ARCH_X86_32 &&
456  (mm_flags & (AV_CPU_FLAG_MMX2 | AV_CPU_FLAG_SSE2)) == AV_CPU_FLAG_MMX2;
457  int64_t max_src_size = (INT64_MAX/2 / c->phase_count) / c->src_incr;
458 
459  if (c->compensation_distance)
460  dst_size = FFMIN(dst_size, c->compensation_distance);
461  src_size = FFMIN(src_size, max_src_size);
462 
463  *consumed = 0;
464 
465  if (c->filter_length == 1 && c->phase_count == 1) {
466  int64_t index2= (1LL<<32)*c->frac/c->src_incr + (1LL<<32)*c->index;
467  int64_t incr= (1LL<<32) * c->dst_incr / c->src_incr;
468  int new_size = (src_size * (int64_t)c->src_incr - c->frac + c->dst_incr - 1) / c->dst_incr;
469 
470  dst_size = FFMAX(FFMIN(dst_size, new_size), 0);
471  if (dst_size > 0) {
472  for (i = 0; i < dst->ch_count; i++) {
473  c->dsp.resample_one(dst->ch[i], src->ch[i], dst_size, index2, incr);
474  if (i+1 == dst->ch_count) {
475  c->index += dst_size * c->dst_incr_div;
476  c->index += (c->frac + dst_size * (int64_t)c->dst_incr_mod) / c->src_incr;
477  av_assert2(c->index >= 0);
478  *consumed = c->index;
479  c->frac = (c->frac + dst_size * (int64_t)c->dst_incr_mod) % c->src_incr;
480  c->index = 0;
481  }
482  }
483  }
484  } else {
485  int64_t end_index = (1LL + src_size - c->filter_length) * c->phase_count;
486  int64_t delta_frac = (end_index - c->index) * c->src_incr - c->frac;
487  int delta_n = (delta_frac + c->dst_incr - 1) / c->dst_incr;
488  int (*resample_func)(struct ResampleContext *c, void *dst,
489  const void *src, int n, int update_ctx);
490 
491  dst_size = FFMAX(FFMIN(dst_size, delta_n), 0);
492  if (dst_size > 0) {
493  /* resample_linear and resample_common should have same behavior
494  * when frac and dst_incr_mod are zero */
495  resample_func = (c->linear && (c->frac || c->dst_incr_mod)) ?
496  c->dsp.resample_linear : c->dsp.resample_common;
497  for (i = 0; i < dst->ch_count; i++)
498  *consumed = resample_func(c, dst->ch[i], src->ch[i], dst_size, i+1 == dst->ch_count);
499  }
500  }
501 
502  if(need_emms)
503  emms_c();
504 
505  if (c->compensation_distance) {
506  c->compensation_distance -= dst_size;
507  if (!c->compensation_distance) {
508  c->dst_incr = c->ideal_dst_incr;
509  c->dst_incr_div = c->dst_incr / c->src_incr;
510  c->dst_incr_mod = c->dst_incr % c->src_incr;
511  }
512  }
513 
514  return dst_size;
515 }
516 
517 static int64_t get_delay(struct SwrContext *s, int64_t base){
518  ResampleContext *c = s->resample;
519  int64_t num = s->in_buffer_count - (c->filter_length-1)/2;
520  num *= c->phase_count;
521  num -= c->index;
522  num *= c->src_incr;
523  num -= c->frac;
524  return av_rescale(num, base, s->in_sample_rate*(int64_t)c->src_incr * c->phase_count);
525 }
526 
527 static int64_t get_out_samples(struct SwrContext *s, int in_samples) {
528  ResampleContext *c = s->resample;
529  // The + 2 are added to allow implementations to be slightly inaccurate, they should not be needed currently.
530  // They also make it easier to proof that changes and optimizations do not
531  // break the upper bound.
532  int64_t num = s->in_buffer_count + 2LL + in_samples;
533  num *= c->phase_count;
534  num -= c->index;
535  num = av_rescale_rnd(num, s->out_sample_rate, ((int64_t)s->in_sample_rate) * c->phase_count, AV_ROUND_UP) + 2;
536 
537  if (c->compensation_distance) {
538  if (num > INT_MAX)
539  return AVERROR(EINVAL);
540 
541  num = FFMAX(num, (num * c->ideal_dst_incr - 1) / c->dst_incr + 1);
542  }
543  return num;
544 }
545 
546 static int resample_flush(struct SwrContext *s) {
547  ResampleContext *c = s->resample;
548  AudioData *a= &s->in_buffer;
549  int i, j, ret;
550  int reflection = (FFMIN(s->in_buffer_count, c->filter_length) + 1) / 2;
551 
552  if((ret = swri_realloc_audio(a, s->in_buffer_index + s->in_buffer_count + reflection)) < 0)
553  return ret;
554  av_assert0(a->planar);
555  for(i=0; i<a->ch_count; i++){
556  for(j=0; j<reflection; j++){
557  memcpy(a->ch[i] + (s->in_buffer_index+s->in_buffer_count+j )*a->bps,
558  a->ch[i] + (s->in_buffer_index+s->in_buffer_count-j-1)*a->bps, a->bps);
559  }
560  }
561  s->in_buffer_count += reflection;
562  return 0;
563 }
564 
565 // in fact the whole handle multiple ridiculously small buffers might need more thinking...
566 static int invert_initial_buffer(ResampleContext *c, AudioData *dst, const AudioData *src,
567  int in_count, int *out_idx, int *out_sz)
568 {
569  int n, ch, num = FFMIN(in_count + *out_sz, c->filter_length + 1), res;
570 
571  if (c->index >= 0)
572  return 0;
573 
574  if ((res = swri_realloc_audio(dst, c->filter_length * 2 + 1)) < 0)
575  return res;
576 
577  // copy
578  for (n = *out_sz; n < num; n++) {
579  for (ch = 0; ch < src->ch_count; ch++) {
580  memcpy(dst->ch[ch] + ((c->filter_length + n) * c->felem_size),
581  src->ch[ch] + ((n - *out_sz) * c->felem_size), c->felem_size);
582  }
583  }
584 
585  // if not enough data is in, return and wait for more
586  if (num < c->filter_length + 1) {
587  *out_sz = num;
588  *out_idx = c->filter_length;
589  return INT_MAX;
590  }
591 
592  // else invert
593  for (n = 1; n <= c->filter_length; n++) {
594  for (ch = 0; ch < src->ch_count; ch++) {
595  memcpy(dst->ch[ch] + ((c->filter_length - n) * c->felem_size),
596  dst->ch[ch] + ((c->filter_length + n) * c->felem_size),
597  c->felem_size);
598  }
599  }
600 
601  res = num - *out_sz;
602  *out_idx = c->filter_length;
603  while (c->index < 0) {
604  --*out_idx;
605  c->index += c->phase_count;
606  }
607  *out_sz = FFMAX(*out_sz + c->filter_length,
608  1 + c->filter_length * 2) - *out_idx;
609 
610  return FFMAX(res, 0);
611 }
612 
613 struct Resampler const swri_resampler={
614  resample_init,
615  resample_free,
616  multiple_resample,
617  resample_flush,
618  set_compensation,
619  get_delay,
620  invert_initial_buffer,
621  get_out_samples,
622 };
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:69
q1
static const uint8_t q1[256]
Definition: twofish.c:96
r
const char * r
Definition: vf_curves.c:114
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
n
int n
Definition: avisynth_c.h:760
ch
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(const uint8_t *) pi - 0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(const int16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(const int16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(const int32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(const int32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(const int64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0f/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float *) pi *(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double *) pi *(INT64_C(1)<< 63))) #define FMT_PAIR_FUNC(out, in) static conv_func_type *const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB *AV_SAMPLE_FMT_NB]={ FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64), };static void cpy1(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, len);} static void cpy2(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 2 *len);} static void cpy4(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 4 *len);} static void cpy8(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 8 *len);} AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, const int *ch_map, int flags) { AudioConvert *ctx;conv_func_type *f=fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt)+AV_SAMPLE_FMT_NB *av_get_packed_sample_fmt(in_fmt)];if(!f) return NULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) return NULL;if(channels==1){ in_fmt=av_get_planar_sample_fmt(in_fmt);out_fmt=av_get_planar_sample_fmt(out_fmt);} ctx->channels=channels;ctx->conv_f=f;ctx->ch_map=ch_map;if(in_fmt==AV_SAMPLE_FMT_U8||in_fmt==AV_SAMPLE_FMT_U8P) memset(ctx->silence, 0x80, sizeof(ctx->silence));if(out_fmt==in_fmt &&!ch_map) { switch(av_get_bytes_per_sample(in_fmt)){ case 1:ctx->simd_f=cpy1;break;case 2:ctx->simd_f=cpy2;break;case 4:ctx->simd_f=cpy4;break;case 8:ctx->simd_f=cpy8;break;} } if(HAVE_X86ASM &&1) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels);if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels);if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels);return ctx;} void swri_audio_convert_free(AudioConvert **ctx) { av_freep(ctx);} int swri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, int len) { int ch;int off=0;const int os=(out->planar ? 1 :out->ch_count) *out->bps;unsigned misaligned=0;av_assert0(ctx->channels==out->ch_count);if(ctx->in_simd_align_mask) { int planes=in->planar ? in->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) in->ch[ch];misaligned|=m &ctx->in_simd_align_mask;} if(ctx->out_simd_align_mask) { int planes=out->planar ? out->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) out->ch[ch];misaligned|=m &ctx->out_simd_align_mask;} if(ctx->simd_f &&!ctx->ch_map &&!misaligned){ off=len &~15;av_assert1(off >=0);av_assert1(off<=len);av_assert2(ctx->channels==SWR_CH_MAX||!in->ch[ctx->channels]);if(off >0){ if(out->planar==in->planar){ int planes=out->planar ? out->ch_count :1;for(ch=0;ch< planes;ch++){ ctx->simd_f(out-> ch ch
Definition: audioconvert.c:56
av_unused
#define av_unused
Definition: attributes.h:125
get_out_samples
static int64_t get_out_samples(struct SwrContext *s, int in_samples)
Definition: resample.c:527
w
uint8_t w
Definition: llviddspenc.c:38
kaiser_beta
static float kaiser_beta(float att, float tr_bw)
Definition: asrc_sinc.c:140
set_compensation
static int set_compensation(ResampleContext *c, int sample_delta, int compensation_distance)
Definition: resample.c:431
AV_SAMPLE_FMT_S32P
@ AV_SAMPLE_FMT_S32P
signed 32 bits, planar
Definition: samplefmt.h:68
linear
static int linear(InterplayACMContext *s, unsigned ind, unsigned col)
Definition: interplayacm.c:121
LEN
#define LEN
swri_resampler
struct Resampler const swri_resampler
Definition: resample.c:613
base
uint8_t base
Definition: vp3data.h:202
filter
filter_frame For filters that do not use the this method is called when a frame is pushed to the filter s input It can be called at any time except in a reentrant way If the input frame is enough to produce then the filter should push the output frames on the output link immediately As an exception to the previous rule if the input frame is enough to produce several output frames then the filter needs output only at least one per link The additional frames can be left buffered in the filter
Definition: filter_design.txt:228
av_get_cpu_flags
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:93
swri_resample_dsp_init
void swri_resample_dsp_init(ResampleContext *c)
Definition: resample_dsp.c:46
resample_free
static void resample_free(ResampleContext **cc)
Definition: resample.c:279
AudioData
Audio buffer used for intermediate storage between conversion phases.
Definition: audio_data.h:37
SWR_FILTER_TYPE_BLACKMAN_NUTTALL
@ SWR_FILTER_TYPE_BLACKMAN_NUTTALL
Blackman Nuttall windowed sinc.
Definition: swresample.h:168
ResampleContext::filter_length
int filter_length
Definition: resample.h:32
fail
#define fail()
Definition: checkasm.h:120
swri_realloc_audio
int swri_realloc_audio(AudioData *a, int count)
Definition: swresample.c:400
tab
static const struct twinvq_data tab
Definition: twinvq_data.h:11135
AV_ROUND_UP
@ AV_ROUND_UP
Round toward +infinity.
Definition: mathematics.h:83
SWR_FILTER_TYPE_KAISER
@ SWR_FILTER_TYPE_KAISER
Kaiser windowed sinc.
Definition: swresample.h:169
ss
#define ss(width, name, subs,...)
Definition: cbs_vp9.c:261
av_reduce
int av_reduce(int *dst_num, int *dst_den, int64_t num, int64_t den, int64_t max)
Reduce a fraction.
Definition: rational.c:35
src
#define src
Definition: vp8dsp.c:254
avassert.h
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
s
#define s(width, name)
Definition: cbs_vp9.c:257
ResampleContext
Definition: af_resample.c:38
format
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 format(the sample packing is implied by the sample format) and sample rate. The lists are not just lists
av_assert0
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
rebuild_filter_bank_with_compensation
static int rebuild_filter_bank_with_compensation(ResampleContext *c)
Definition: resample.c:383
SwrContext
The libswresample context.
Definition: swresample_internal.h:95
AudioData::ch
uint8_t * ch[SWR_CH_MAX]
samples buffer per channel
Definition: swresample_internal.h:46
int32_t
int32_t
Definition: audio_convert.c:194
if
if(ret)
Definition: filter_design.txt:179
NULL
#define NULL
Definition: coverity.c:32
SwrFilterType
SwrFilterType
Resampling Filter Types.
Definition: swresample.h:166
eval_poly
static double eval_poly(const double *coeff, int size, double x)
Definition: resample.c:32
exp
int8_t exp
Definition: eval.c:72
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
error
static void error(const char *err)
Definition: target_dec_fuzzer.c:61
AudioData::ch_count
int ch_count
number of channels
Definition: swresample_internal.h:48
AV_CPU_FLAG_SSE2
#define AV_CPU_FLAG_SSE2
PIV SSE2 functions.
Definition: cpu.h:36
av_rescale_rnd
int64_t av_rescale_rnd(int64_t a, int64_t b, int64_t c, enum AVRounding rnd)
Rescale a 64-bit integer with specified rounding.
Definition: mathematics.c:58
FFMAX
#define FFMAX(a, b)
Definition: common.h:94
size
int size
Definition: twinvq_data.h:11134
SWR_FILTER_TYPE_CUBIC
@ SWR_FILTER_TYPE_CUBIC
Cubic.
Definition: swresample.h:167
a
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
Definition: undefined.txt:41
FFMIN
#define FFMIN(a, b)
Definition: common.h:96
get_delay
static int64_t get_delay(struct SwrContext *s, int64_t base)
Definition: resample.c:517
Resampler
Definition: swresample_internal.h:81
bessel
static double bessel(double x)
0th order modified bessel function of the first kind.
Definition: resample.c:72
M_PI
#define M_PI
Definition: mathematics.h:52
AV_SAMPLE_FMT_S16P
@ AV_SAMPLE_FMT_S16P
signed 16 bits, planar
Definition: samplefmt.h:67
multiple_resample
static int multiple_resample(ResampleContext *c, AudioData *dst, int dst_size, AudioData *src, int src_size, int *consumed)
Definition: resample.c:452
av_assert2
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
Definition: avassert.h:64
AV_CPU_FLAG_MMX2
#define AV_CPU_FLAG_MMX2
SSE integer functions or AMD MMX ext.
Definition: cpu.h:33
lrintf
#define lrintf(x)
Definition: libm_mips.h:70
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
av_get_bytes_per_sample
int av_get_bytes_per_sample(enum AVSampleFormat sample_fmt)
Return number of bytes per sample.
Definition: samplefmt.c:106
av_malloc_array
#define av_malloc_array(a, b)
Definition: tableprint_vlc.h:32
AVSampleFormat
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
uint8_t
uint8_t
Definition: audio_convert.c:194
resample_init
static ResampleContext * resample_init(ResampleContext *c, int out_rate, int in_rate, int filter_size, int phase_shift, int linear, double cutoff0, enum AVSampleFormat format, enum SwrFilterType filter_type, double kaiser_beta, double precision, int cheby, int exact_rational)
Definition: resample.c:287
av_mallocz
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:236
av_rescale
int64_t av_rescale(int64_t a, int64_t b, int64_t c)
Rescale a 64-bit integer with rounding to nearest.
Definition: mathematics.c:129
build_filter
static int build_filter(ResampleContext *c, void *filter, double factor, int tap_count, int alloc, int phase_count, int scale, int filter_type, double kaiser_beta)
builds a polyphase filterbank.
Definition: resample.c:144
ret
ret
Definition: filter_design.txt:187
FF_ARRAY_ELEMS
#define FF_ARRAY_ELEMS(a)
Definition: sinewin_tablegen_template.c:38
av_calloc
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
Definition: mem.c:244
AV_SAMPLE_FMT_DBLP
@ AV_SAMPLE_FMT_DBLP
double, planar
Definition: samplefmt.h:70
resample.h
factor
static const int factor[16]
Definition: vf_pp7.c:75
resample_flush
static int resample_flush(struct SwrContext *s)
Definition: resample.c:546
llrint
#define llrint(x)
Definition: libm.h:394
av_free
#define av_free(p)
Definition: tableprint_vlc.h:34
FFALIGN
#define FFALIGN(x, a)
Definition: macros.h:48
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:35
invert_initial_buffer
static int invert_initial_buffer(ResampleContext *c, AudioData *dst, const AudioData *src, int in_count, int *out_idx, int *out_sz)
Definition: resample.c:566
coeff
static const double coeff[2][5]
Definition: vf_owdenoise.c:72
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:28
int
int
Definition: ffmpeg_filter.c:191
Generated on Wed Aug 24 2022 21:27:26 for FFmpeg by   doxygen 1.8.17