FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
aacenc.c
Go to the documentation of this file.
1 /*
2  * AAC encoder
3  * Copyright (C) 2008 Konstantin Shishkov
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  * AAC encoder
25  */
26 
27 /***********************************
28  * TODOs:
29  * add sane pulse detection
30  ***********************************/
31 
32 #include "libavutil/libm.h"
33 #include "libavutil/thread.h"
34 #include "libavutil/float_dsp.h"
35 #include "libavutil/opt.h"
36 #include "avcodec.h"
37 #include "put_bits.h"
38 #include "internal.h"
39 #include "mpeg4audio.h"
40 #include "kbdwin.h"
41 #include "sinewin.h"
42 
43 #include "aac.h"
44 #include "aactab.h"
45 #include "aacenc.h"
46 #include "aacenctab.h"
47 #include "aacenc_utils.h"
48 
49 #include "psymodel.h"
50 
52 
53 /**
54  * Make AAC audio config object.
55  * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
56  */
58 {
59  PutBitContext pb;
60  AACEncContext *s = avctx->priv_data;
61  int channels = s->channels - (s->channels == 8 ? 1 : 0);
62 
63  init_put_bits(&pb, avctx->extradata, avctx->extradata_size);
64  put_bits(&pb, 5, s->profile+1); //profile
65  put_bits(&pb, 4, s->samplerate_index); //sample rate index
66  put_bits(&pb, 4, channels);
67  //GASpecificConfig
68  put_bits(&pb, 1, 0); //frame length - 1024 samples
69  put_bits(&pb, 1, 0); //does not depend on core coder
70  put_bits(&pb, 1, 0); //is not extension
71 
72  //Explicitly Mark SBR absent
73  put_bits(&pb, 11, 0x2b7); //sync extension
74  put_bits(&pb, 5, AOT_SBR);
75  put_bits(&pb, 1, 0);
76  flush_put_bits(&pb);
77 }
78 
80 {
83  memset(s->quantize_band_cost_cache, 0, sizeof(s->quantize_band_cost_cache));
85  }
86 }
87 
88 #define WINDOW_FUNC(type) \
89 static void apply_ ##type ##_window(AVFloatDSPContext *fdsp, \
90  SingleChannelElement *sce, \
91  const float *audio)
92 
93 WINDOW_FUNC(only_long)
94 {
95  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
96  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
97  float *out = sce->ret_buf;
98 
99  fdsp->vector_fmul (out, audio, lwindow, 1024);
100  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, pwindow, 1024);
101 }
102 
103 WINDOW_FUNC(long_start)
104 {
105  const float *lwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
106  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
107  float *out = sce->ret_buf;
108 
109  fdsp->vector_fmul(out, audio, lwindow, 1024);
110  memcpy(out + 1024, audio + 1024, sizeof(out[0]) * 448);
111  fdsp->vector_fmul_reverse(out + 1024 + 448, audio + 1024 + 448, swindow, 128);
112  memset(out + 1024 + 576, 0, sizeof(out[0]) * 448);
113 }
114 
115 WINDOW_FUNC(long_stop)
116 {
117  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
118  const float *swindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
119  float *out = sce->ret_buf;
120 
121  memset(out, 0, sizeof(out[0]) * 448);
122  fdsp->vector_fmul(out + 448, audio + 448, swindow, 128);
123  memcpy(out + 576, audio + 576, sizeof(out[0]) * 448);
124  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, lwindow, 1024);
125 }
126 
127 WINDOW_FUNC(eight_short)
128 {
129  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
130  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
131  const float *in = audio + 448;
132  float *out = sce->ret_buf;
133  int w;
134 
135  for (w = 0; w < 8; w++) {
136  fdsp->vector_fmul (out, in, w ? pwindow : swindow, 128);
137  out += 128;
138  in += 128;
139  fdsp->vector_fmul_reverse(out, in, swindow, 128);
140  out += 128;
141  }
142 }
143 
144 static void (*const apply_window[4])(AVFloatDSPContext *fdsp,
146  const float *audio) = {
147  [ONLY_LONG_SEQUENCE] = apply_only_long_window,
148  [LONG_START_SEQUENCE] = apply_long_start_window,
149  [EIGHT_SHORT_SEQUENCE] = apply_eight_short_window,
150  [LONG_STOP_SEQUENCE] = apply_long_stop_window
151 };
152 
154  float *audio)
155 {
156  int i;
157  const float *output = sce->ret_buf;
158 
159  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, audio);
160 
162  s->mdct1024.mdct_calc(&s->mdct1024, sce->coeffs, output);
163  else
164  for (i = 0; i < 1024; i += 128)
165  s->mdct128.mdct_calc(&s->mdct128, &sce->coeffs[i], output + i*2);
166  memcpy(audio, audio + 1024, sizeof(audio[0]) * 1024);
167  memcpy(sce->pcoeffs, sce->coeffs, sizeof(sce->pcoeffs));
168 }
169 
170 /**
171  * Encode ics_info element.
172  * @see Table 4.6 (syntax of ics_info)
173  */
175 {
176  int w;
177 
178  put_bits(&s->pb, 1, 0); // ics_reserved bit
179  put_bits(&s->pb, 2, info->window_sequence[0]);
180  put_bits(&s->pb, 1, info->use_kb_window[0]);
181  if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
182  put_bits(&s->pb, 6, info->max_sfb);
183  put_bits(&s->pb, 1, !!info->predictor_present);
184  } else {
185  put_bits(&s->pb, 4, info->max_sfb);
186  for (w = 1; w < 8; w++)
187  put_bits(&s->pb, 1, !info->group_len[w]);
188  }
189 }
190 
191 /**
192  * Encode MS data.
193  * @see 4.6.8.1 "Joint Coding - M/S Stereo"
194  */
196 {
197  int i, w;
198 
199  put_bits(pb, 2, cpe->ms_mode);
200  if (cpe->ms_mode == 1)
201  for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
202  for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
203  put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
204 }
205 
206 /**
207  * Produce integer coefficients from scalefactors provided by the model.
208  */
209 static void adjust_frame_information(ChannelElement *cpe, int chans)
210 {
211  int i, w, w2, g, ch;
212  int maxsfb, cmaxsfb;
213 
214  for (ch = 0; ch < chans; ch++) {
215  IndividualChannelStream *ics = &cpe->ch[ch].ics;
216  maxsfb = 0;
217  cpe->ch[ch].pulse.num_pulse = 0;
218  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
219  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
220  for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w*16+cmaxsfb-1]; cmaxsfb--)
221  ;
222  maxsfb = FFMAX(maxsfb, cmaxsfb);
223  }
224  }
225  ics->max_sfb = maxsfb;
226 
227  //adjust zero bands for window groups
228  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
229  for (g = 0; g < ics->max_sfb; g++) {
230  i = 1;
231  for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
232  if (!cpe->ch[ch].zeroes[w2*16 + g]) {
233  i = 0;
234  break;
235  }
236  }
237  cpe->ch[ch].zeroes[w*16 + g] = i;
238  }
239  }
240  }
241 
242  if (chans > 1 && cpe->common_window) {
243  IndividualChannelStream *ics0 = &cpe->ch[0].ics;
244  IndividualChannelStream *ics1 = &cpe->ch[1].ics;
245  int msc = 0;
246  ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
247  ics1->max_sfb = ics0->max_sfb;
248  for (w = 0; w < ics0->num_windows*16; w += 16)
249  for (i = 0; i < ics0->max_sfb; i++)
250  if (cpe->ms_mask[w+i])
251  msc++;
252  if (msc == 0 || ics0->max_sfb == 0)
253  cpe->ms_mode = 0;
254  else
255  cpe->ms_mode = msc < ics0->max_sfb * ics0->num_windows ? 1 : 2;
256  }
257 }
258 
260 {
261  int w, w2, g, i;
262  IndividualChannelStream *ics = &cpe->ch[0].ics;
263  if (!cpe->common_window)
264  return;
265  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
266  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
267  int start = (w+w2) * 128;
268  for (g = 0; g < ics->num_swb; g++) {
269  int p = -1 + 2 * (cpe->ch[1].band_type[w*16+g] - 14);
270  float scale = cpe->ch[0].is_ener[w*16+g];
271  if (!cpe->is_mask[w*16 + g]) {
272  start += ics->swb_sizes[g];
273  continue;
274  }
275  if (cpe->ms_mask[w*16 + g])
276  p *= -1;
277  for (i = 0; i < ics->swb_sizes[g]; i++) {
278  float sum = (cpe->ch[0].coeffs[start+i] + p*cpe->ch[1].coeffs[start+i])*scale;
279  cpe->ch[0].coeffs[start+i] = sum;
280  cpe->ch[1].coeffs[start+i] = 0.0f;
281  }
282  start += ics->swb_sizes[g];
283  }
284  }
285  }
286 }
287 
289 {
290  int w, w2, g, i;
291  IndividualChannelStream *ics = &cpe->ch[0].ics;
292  if (!cpe->common_window)
293  return;
294  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
295  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
296  int start = (w+w2) * 128;
297  for (g = 0; g < ics->num_swb; g++) {
298  /* ms_mask can be used for other purposes in PNS and I/S,
299  * so must not apply M/S if any band uses either, even if
300  * ms_mask is set.
301  */
302  if (!cpe->ms_mask[w*16 + g] || cpe->is_mask[w*16 + g]
303  || cpe->ch[0].band_type[w*16 + g] >= NOISE_BT
304  || cpe->ch[1].band_type[w*16 + g] >= NOISE_BT) {
305  start += ics->swb_sizes[g];
306  continue;
307  }
308  for (i = 0; i < ics->swb_sizes[g]; i++) {
309  float L = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) * 0.5f;
310  float R = L - cpe->ch[1].coeffs[start+i];
311  cpe->ch[0].coeffs[start+i] = L;
312  cpe->ch[1].coeffs[start+i] = R;
313  }
314  start += ics->swb_sizes[g];
315  }
316  }
317  }
318 }
319 
320 /**
321  * Encode scalefactor band coding type.
322  */
324 {
325  int w;
326 
329 
330  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
331  s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
332 }
333 
334 /**
335  * Encode scalefactors.
336  */
339 {
340  int diff, off_sf = sce->sf_idx[0], off_pns = sce->sf_idx[0] - NOISE_OFFSET;
341  int off_is = 0, noise_flag = 1;
342  int i, w;
343 
344  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
345  for (i = 0; i < sce->ics.max_sfb; i++) {
346  if (!sce->zeroes[w*16 + i]) {
347  if (sce->band_type[w*16 + i] == NOISE_BT) {
348  diff = sce->sf_idx[w*16 + i] - off_pns;
349  off_pns = sce->sf_idx[w*16 + i];
350  if (noise_flag-- > 0) {
351  put_bits(&s->pb, NOISE_PRE_BITS, diff + NOISE_PRE);
352  continue;
353  }
354  } else if (sce->band_type[w*16 + i] == INTENSITY_BT ||
355  sce->band_type[w*16 + i] == INTENSITY_BT2) {
356  diff = sce->sf_idx[w*16 + i] - off_is;
357  off_is = sce->sf_idx[w*16 + i];
358  } else {
359  diff = sce->sf_idx[w*16 + i] - off_sf;
360  off_sf = sce->sf_idx[w*16 + i];
361  }
362  diff += SCALE_DIFF_ZERO;
363  av_assert0(diff >= 0 && diff <= 120);
365  }
366  }
367  }
368 }
369 
370 /**
371  * Encode pulse data.
372  */
373 static void encode_pulses(AACEncContext *s, Pulse *pulse)
374 {
375  int i;
376 
377  put_bits(&s->pb, 1, !!pulse->num_pulse);
378  if (!pulse->num_pulse)
379  return;
380 
381  put_bits(&s->pb, 2, pulse->num_pulse - 1);
382  put_bits(&s->pb, 6, pulse->start);
383  for (i = 0; i < pulse->num_pulse; i++) {
384  put_bits(&s->pb, 5, pulse->pos[i]);
385  put_bits(&s->pb, 4, pulse->amp[i]);
386  }
387 }
388 
389 /**
390  * Encode spectral coefficients processed by psychoacoustic model.
391  */
393 {
394  int start, i, w, w2;
395 
396  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
397  start = 0;
398  for (i = 0; i < sce->ics.max_sfb; i++) {
399  if (sce->zeroes[w*16 + i]) {
400  start += sce->ics.swb_sizes[i];
401  continue;
402  }
403  for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
404  s->coder->quantize_and_encode_band(s, &s->pb,
405  &sce->coeffs[start + w2*128],
406  NULL, sce->ics.swb_sizes[i],
407  sce->sf_idx[w*16 + i],
408  sce->band_type[w*16 + i],
409  s->lambda,
410  sce->ics.window_clipping[w]);
411  }
412  start += sce->ics.swb_sizes[i];
413  }
414  }
415 }
416 
417 /**
418  * Downscale spectral coefficients for near-clipping windows to avoid artifacts
419  */
421 {
422  int start, i, j, w;
423 
424  if (sce->ics.clip_avoidance_factor < 1.0f) {
425  for (w = 0; w < sce->ics.num_windows; w++) {
426  start = 0;
427  for (i = 0; i < sce->ics.max_sfb; i++) {
428  float *swb_coeffs = &sce->coeffs[start + w*128];
429  for (j = 0; j < sce->ics.swb_sizes[i]; j++)
430  swb_coeffs[j] *= sce->ics.clip_avoidance_factor;
431  start += sce->ics.swb_sizes[i];
432  }
433  }
434  }
435 }
436 
437 /**
438  * Encode one channel of audio data.
439  */
442  int common_window)
443 {
444  put_bits(&s->pb, 8, sce->sf_idx[0]);
445  if (!common_window) {
446  put_ics_info(s, &sce->ics);
447  if (s->coder->encode_main_pred)
448  s->coder->encode_main_pred(s, sce);
449  if (s->coder->encode_ltp_info)
450  s->coder->encode_ltp_info(s, sce, 0);
451  }
452  encode_band_info(s, sce);
453  encode_scale_factors(avctx, s, sce);
454  encode_pulses(s, &sce->pulse);
455  put_bits(&s->pb, 1, !!sce->tns.present);
456  if (s->coder->encode_tns_info)
457  s->coder->encode_tns_info(s, sce);
458  put_bits(&s->pb, 1, 0); //ssr
459  encode_spectral_coeffs(s, sce);
460  return 0;
461 }
462 
463 /**
464  * Write some auxiliary information about the created AAC file.
465  */
466 static void put_bitstream_info(AACEncContext *s, const char *name)
467 {
468  int i, namelen, padbits;
469 
470  namelen = strlen(name) + 2;
471  put_bits(&s->pb, 3, TYPE_FIL);
472  put_bits(&s->pb, 4, FFMIN(namelen, 15));
473  if (namelen >= 15)
474  put_bits(&s->pb, 8, namelen - 14);
475  put_bits(&s->pb, 4, 0); //extension type - filler
476  padbits = -put_bits_count(&s->pb) & 7;
478  for (i = 0; i < namelen - 2; i++)
479  put_bits(&s->pb, 8, name[i]);
480  put_bits(&s->pb, 12 - padbits, 0);
481 }
482 
483 /*
484  * Copy input samples.
485  * Channels are reordered from libavcodec's default order to AAC order.
486  */
488 {
489  int ch;
490  int end = 2048 + (frame ? frame->nb_samples : 0);
491  const uint8_t *channel_map = aac_chan_maps[s->channels - 1];
492 
493  /* copy and remap input samples */
494  for (ch = 0; ch < s->channels; ch++) {
495  /* copy last 1024 samples of previous frame to the start of the current frame */
496  memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0]));
497 
498  /* copy new samples and zero any remaining samples */
499  if (frame) {
500  memcpy(&s->planar_samples[ch][2048],
501  frame->extended_data[channel_map[ch]],
502  frame->nb_samples * sizeof(s->planar_samples[0][0]));
503  }
504  memset(&s->planar_samples[ch][end], 0,
505  (3072 - end) * sizeof(s->planar_samples[0][0]));
506  }
507 }
508 
509 static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
510  const AVFrame *frame, int *got_packet_ptr)
511 {
512  AACEncContext *s = avctx->priv_data;
513  float **samples = s->planar_samples, *samples2, *la, *overlap;
514  ChannelElement *cpe;
517  int i, its, ch, w, chans, tag, start_ch, ret, frame_bits;
518  int target_bits, rate_bits, too_many_bits, too_few_bits;
519  int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
520  int chan_el_counter[4];
522 
523  if (s->last_frame == 2)
524  return 0;
525 
526  /* add current frame to queue */
527  if (frame) {
528  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
529  return ret;
530  }
531 
532  copy_input_samples(s, frame);
533  if (s->psypp)
535 
536  if (!avctx->frame_number)
537  return 0;
538 
539  start_ch = 0;
540  for (i = 0; i < s->chan_map[0]; i++) {
541  FFPsyWindowInfo* wi = windows + start_ch;
542  tag = s->chan_map[i+1];
543  chans = tag == TYPE_CPE ? 2 : 1;
544  cpe = &s->cpe[i];
545  for (ch = 0; ch < chans; ch++) {
546  int k;
547  float clip_avoidance_factor;
548  sce = &cpe->ch[ch];
549  ics = &sce->ics;
550  s->cur_channel = start_ch + ch;
551  overlap = &samples[s->cur_channel][0];
552  samples2 = overlap + 1024;
553  la = samples2 + (448+64);
554  if (!frame)
555  la = NULL;
556  if (tag == TYPE_LFE) {
557  wi[ch].window_type[0] = wi[ch].window_type[1] = ONLY_LONG_SEQUENCE;
558  wi[ch].window_shape = 0;
559  wi[ch].num_windows = 1;
560  wi[ch].grouping[0] = 1;
561  wi[ch].clipping[0] = 0;
562 
563  /* Only the lowest 12 coefficients are used in a LFE channel.
564  * The expression below results in only the bottom 8 coefficients
565  * being used for 11.025kHz to 16kHz sample rates.
566  */
567  ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
568  } else {
569  wi[ch] = s->psy.model->window(&s->psy, samples2, la, s->cur_channel,
570  ics->window_sequence[0]);
571  }
572  ics->window_sequence[1] = ics->window_sequence[0];
573  ics->window_sequence[0] = wi[ch].window_type[0];
574  ics->use_kb_window[1] = ics->use_kb_window[0];
575  ics->use_kb_window[0] = wi[ch].window_shape;
576  ics->num_windows = wi[ch].num_windows;
577  ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
578  ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
579  ics->max_sfb = FFMIN(ics->max_sfb, ics->num_swb);
580  ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
586 
587  for (w = 0; w < ics->num_windows; w++)
588  ics->group_len[w] = wi[ch].grouping[w];
589 
590  /* Calculate input sample maximums and evaluate clipping risk */
591  clip_avoidance_factor = 0.0f;
592  for (w = 0; w < ics->num_windows; w++) {
593  const float *wbuf = overlap + w * 128;
594  const int wlen = 2048 / ics->num_windows;
595  float max = 0;
596  int j;
597  /* mdct input is 2 * output */
598  for (j = 0; j < wlen; j++)
599  max = FFMAX(max, fabsf(wbuf[j]));
600  wi[ch].clipping[w] = max;
601  }
602  for (w = 0; w < ics->num_windows; w++) {
603  if (wi[ch].clipping[w] > CLIP_AVOIDANCE_FACTOR) {
604  ics->window_clipping[w] = 1;
605  clip_avoidance_factor = FFMAX(clip_avoidance_factor, wi[ch].clipping[w]);
606  } else {
607  ics->window_clipping[w] = 0;
608  }
609  }
610  if (clip_avoidance_factor > CLIP_AVOIDANCE_FACTOR) {
611  ics->clip_avoidance_factor = CLIP_AVOIDANCE_FACTOR / clip_avoidance_factor;
612  } else {
613  ics->clip_avoidance_factor = 1.0f;
614  }
615 
616  apply_window_and_mdct(s, sce, overlap);
617 
618  if (s->options.ltp && s->coder->update_ltp) {
619  s->coder->update_ltp(s, sce);
620  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
621  s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
622  }
623 
624  for (k = 0; k < 1024; k++) {
625  if (!(fabs(cpe->ch[ch].coeffs[k]) < 1E16)) { // Ensure headroom for energy calculation
626  av_log(avctx, AV_LOG_ERROR, "Input contains (near) NaN/+-Inf\n");
627  return AVERROR(EINVAL);
628  }
629  }
630  avoid_clipping(s, sce);
631  }
632  start_ch += chans;
633  }
634  if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels, 0)) < 0)
635  return ret;
636  frame_bits = its = 0;
637  do {
638  init_put_bits(&s->pb, avpkt->data, avpkt->size);
639 
640  if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & AV_CODEC_FLAG_BITEXACT))
642  start_ch = 0;
643  target_bits = 0;
644  memset(chan_el_counter, 0, sizeof(chan_el_counter));
645  for (i = 0; i < s->chan_map[0]; i++) {
646  FFPsyWindowInfo* wi = windows + start_ch;
647  const float *coeffs[2];
648  tag = s->chan_map[i+1];
649  chans = tag == TYPE_CPE ? 2 : 1;
650  cpe = &s->cpe[i];
651  cpe->common_window = 0;
652  memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
653  memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
654  put_bits(&s->pb, 3, tag);
655  put_bits(&s->pb, 4, chan_el_counter[tag]++);
656  for (ch = 0; ch < chans; ch++) {
657  sce = &cpe->ch[ch];
658  coeffs[ch] = sce->coeffs;
659  sce->ics.predictor_present = 0;
660  sce->ics.ltp.present = 0;
661  memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
662  memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
663  memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
664  for (w = 0; w < 128; w++)
665  if (sce->band_type[w] > RESERVED_BT)
666  sce->band_type[w] = 0;
667  }
668  s->psy.bitres.alloc = -1;
670  s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
671  if (s->psy.bitres.alloc > 0) {
672  /* Lambda unused here on purpose, we need to take psy's unscaled allocation */
673  target_bits += s->psy.bitres.alloc
674  * (s->lambda / (avctx->global_quality ? avctx->global_quality : 120));
675  s->psy.bitres.alloc /= chans;
676  }
677  s->cur_type = tag;
678  for (ch = 0; ch < chans; ch++) {
679  s->cur_channel = start_ch + ch;
680  if (s->options.pns && s->coder->mark_pns)
681  s->coder->mark_pns(s, avctx, &cpe->ch[ch]);
682  s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
683  }
684  if (chans > 1
685  && wi[0].window_type[0] == wi[1].window_type[0]
686  && wi[0].window_shape == wi[1].window_shape) {
687 
688  cpe->common_window = 1;
689  for (w = 0; w < wi[0].num_windows; w++) {
690  if (wi[0].grouping[w] != wi[1].grouping[w]) {
691  cpe->common_window = 0;
692  break;
693  }
694  }
695  }
696  for (ch = 0; ch < chans; ch++) { /* TNS and PNS */
697  sce = &cpe->ch[ch];
698  s->cur_channel = start_ch + ch;
699  if (s->options.tns && s->coder->search_for_tns)
700  s->coder->search_for_tns(s, sce);
701  if (s->options.tns && s->coder->apply_tns_filt)
702  s->coder->apply_tns_filt(s, sce);
703  if (sce->tns.present)
704  tns_mode = 1;
705  if (s->options.pns && s->coder->search_for_pns)
706  s->coder->search_for_pns(s, avctx, sce);
707  }
708  s->cur_channel = start_ch;
709  if (s->options.intensity_stereo) { /* Intensity Stereo */
710  if (s->coder->search_for_is)
711  s->coder->search_for_is(s, avctx, cpe);
712  if (cpe->is_mode) is_mode = 1;
714  }
715  if (s->options.pred) { /* Prediction */
716  for (ch = 0; ch < chans; ch++) {
717  sce = &cpe->ch[ch];
718  s->cur_channel = start_ch + ch;
719  if (s->options.pred && s->coder->search_for_pred)
720  s->coder->search_for_pred(s, sce);
721  if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
722  }
723  if (s->coder->adjust_common_pred)
724  s->coder->adjust_common_pred(s, cpe);
725  for (ch = 0; ch < chans; ch++) {
726  sce = &cpe->ch[ch];
727  s->cur_channel = start_ch + ch;
728  if (s->options.pred && s->coder->apply_main_pred)
729  s->coder->apply_main_pred(s, sce);
730  }
731  s->cur_channel = start_ch;
732  }
733  if (s->options.mid_side) { /* Mid/Side stereo */
734  if (s->options.mid_side == -1 && s->coder->search_for_ms)
735  s->coder->search_for_ms(s, cpe);
736  else if (cpe->common_window)
737  memset(cpe->ms_mask, 1, sizeof(cpe->ms_mask));
739  }
740  adjust_frame_information(cpe, chans);
741  if (s->options.ltp) { /* LTP */
742  for (ch = 0; ch < chans; ch++) {
743  sce = &cpe->ch[ch];
744  s->cur_channel = start_ch + ch;
745  if (s->coder->search_for_ltp)
746  s->coder->search_for_ltp(s, sce, cpe->common_window);
747  if (sce->ics.ltp.present) pred_mode = 1;
748  }
749  s->cur_channel = start_ch;
750  if (s->coder->adjust_common_ltp)
751  s->coder->adjust_common_ltp(s, cpe);
752  }
753  if (chans == 2) {
754  put_bits(&s->pb, 1, cpe->common_window);
755  if (cpe->common_window) {
756  put_ics_info(s, &cpe->ch[0].ics);
757  if (s->coder->encode_main_pred)
758  s->coder->encode_main_pred(s, &cpe->ch[0]);
759  if (s->coder->encode_ltp_info)
760  s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
761  encode_ms_info(&s->pb, cpe);
762  if (cpe->ms_mode) ms_mode = 1;
763  }
764  }
765  for (ch = 0; ch < chans; ch++) {
766  s->cur_channel = start_ch + ch;
767  encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
768  }
769  start_ch += chans;
770  }
771 
772  if (avctx->flags & CODEC_FLAG_QSCALE) {
773  /* When using a constant Q-scale, don't mess with lambda */
774  break;
775  }
776 
777  /* rate control stuff
778  * allow between the nominal bitrate, and what psy's bit reservoir says to target
779  * but drift towards the nominal bitrate always
780  */
781  frame_bits = put_bits_count(&s->pb);
782  rate_bits = avctx->bit_rate * 1024 / avctx->sample_rate;
783  rate_bits = FFMIN(rate_bits, 6144 * s->channels - 3);
784  too_many_bits = FFMAX(target_bits, rate_bits);
785  too_many_bits = FFMIN(too_many_bits, 6144 * s->channels - 3);
786  too_few_bits = FFMIN(FFMAX(rate_bits - rate_bits/4, target_bits), too_many_bits);
787 
788  /* When using ABR, be strict (but only for increasing) */
789  too_few_bits = too_few_bits - too_few_bits/8;
790  too_many_bits = too_many_bits + too_many_bits/2;
791 
792  if ( its == 0 /* for steady-state Q-scale tracking */
793  || (its < 5 && (frame_bits < too_few_bits || frame_bits > too_many_bits))
794  || frame_bits >= 6144 * s->channels - 3 )
795  {
796  float ratio = ((float)rate_bits) / frame_bits;
797 
798  if (frame_bits >= too_few_bits && frame_bits <= too_many_bits) {
799  /*
800  * This path is for steady-state Q-scale tracking
801  * When frame bits fall within the stable range, we still need to adjust
802  * lambda to maintain it like so in a stable fashion (large jumps in lambda
803  * create artifacts and should be avoided), but slowly
804  */
805  ratio = sqrtf(sqrtf(ratio));
806  ratio = av_clipf(ratio, 0.9f, 1.1f);
807  } else {
808  /* Not so fast though */
809  ratio = sqrtf(ratio);
810  }
811  s->lambda = FFMIN(s->lambda * ratio, 65536.f);
812 
813  /* Keep iterating if we must reduce and lambda is in the sky */
814  if (ratio > 0.9f && ratio < 1.1f) {
815  break;
816  } else {
817  if (is_mode || ms_mode || tns_mode || pred_mode) {
818  for (i = 0; i < s->chan_map[0]; i++) {
819  // Must restore coeffs
820  chans = tag == TYPE_CPE ? 2 : 1;
821  cpe = &s->cpe[i];
822  for (ch = 0; ch < chans; ch++)
823  memcpy(cpe->ch[ch].coeffs, cpe->ch[ch].pcoeffs, sizeof(cpe->ch[ch].coeffs));
824  }
825  }
826  its++;
827  }
828  } else {
829  break;
830  }
831  } while (1);
832 
833  if (s->options.ltp && s->coder->ltp_insert_new_frame)
835 
836  put_bits(&s->pb, 3, TYPE_END);
837  flush_put_bits(&s->pb);
838 
840 
841  s->lambda_sum += s->lambda;
842  s->lambda_count++;
843 
844  if (!frame)
845  s->last_frame++;
846 
847  ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
848  &avpkt->duration);
849 
850  avpkt->size = put_bits_count(&s->pb) >> 3;
851  *got_packet_ptr = 1;
852  return 0;
853 }
854 
856 {
857  AACEncContext *s = avctx->priv_data;
858 
859  av_log(avctx, AV_LOG_INFO, "Qavg: %.3f\n", s->lambda_sum / s->lambda_count);
860 
861  ff_mdct_end(&s->mdct1024);
862  ff_mdct_end(&s->mdct128);
863  ff_psy_end(&s->psy);
864  ff_lpc_end(&s->lpc);
865  if (s->psypp)
867  av_freep(&s->buffer.samples);
868  av_freep(&s->cpe);
869  av_freep(&s->fdsp);
870  ff_af_queue_close(&s->afq);
871  return 0;
872 }
873 
875 {
876  int ret = 0;
877 
879  if (!s->fdsp)
880  return AVERROR(ENOMEM);
881 
882  // window init
887 
888  if ((ret = ff_mdct_init(&s->mdct1024, 11, 0, 32768.0)) < 0)
889  return ret;
890  if ((ret = ff_mdct_init(&s->mdct128, 8, 0, 32768.0)) < 0)
891  return ret;
892 
893  return 0;
894 }
895 
897 {
898  int ch;
899  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->buffer.samples, s->channels, 3 * 1024 * sizeof(s->buffer.samples[0]), alloc_fail);
900  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->cpe, s->chan_map[0], sizeof(ChannelElement), alloc_fail);
901  FF_ALLOCZ_OR_GOTO(avctx, avctx->extradata, 5 + AV_INPUT_BUFFER_PADDING_SIZE, alloc_fail);
902 
903  for(ch = 0; ch < s->channels; ch++)
904  s->planar_samples[ch] = s->buffer.samples + 3 * 1024 * ch;
905 
906  return 0;
907 alloc_fail:
908  return AVERROR(ENOMEM);
909 }
910 
912 {
914 }
915 
917 {
918  AACEncContext *s = avctx->priv_data;
919  int i, ret = 0;
920  const uint8_t *sizes[2];
921  uint8_t grouping[AAC_MAX_CHANNELS];
922  int lengths[2];
923 
924  /* Constants */
925  s->last_frame_pb_count = 0;
926  avctx->extradata_size = 5;
927  avctx->frame_size = 1024;
928  avctx->initial_padding = 1024;
929  s->lambda = avctx->global_quality > 0 ? avctx->global_quality : 120;
930 
931  /* Channel map and unspecified bitrate guessing */
932  s->channels = avctx->channels;
933  ERROR_IF(s->channels > AAC_MAX_CHANNELS || s->channels == 7,
934  "Unsupported number of channels: %d\n", s->channels);
936  if (!avctx->bit_rate) {
937  for (i = 1; i <= s->chan_map[0]; i++) {
938  avctx->bit_rate += s->chan_map[i] == TYPE_CPE ? 128000 : /* Pair */
939  s->chan_map[i] == TYPE_LFE ? 16000 : /* LFE */
940  69000 ; /* SCE */
941  }
942  }
943 
944  /* Samplerate */
945  for (i = 0; i < 16; i++)
947  break;
948  s->samplerate_index = i;
949  ERROR_IF(s->samplerate_index == 16 ||
952  "Unsupported sample rate %d\n", avctx->sample_rate);
953 
954  /* Bitrate limiting */
955  WARN_IF(1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * s->channels,
956  "Too many bits %f > %d per frame requested, clamping to max\n",
957  1024.0 * avctx->bit_rate / avctx->sample_rate,
958  6144 * s->channels);
959  avctx->bit_rate = (int64_t)FFMIN(6144 * s->channels / 1024.0 * avctx->sample_rate,
960  avctx->bit_rate);
961 
962  /* Profile and option setting */
964  avctx->profile;
965  for (i = 0; i < FF_ARRAY_ELEMS(aacenc_profiles); i++)
966  if (avctx->profile == aacenc_profiles[i])
967  break;
968  if (avctx->profile == FF_PROFILE_MPEG2_AAC_LOW) {
969  avctx->profile = FF_PROFILE_AAC_LOW;
970  ERROR_IF(s->options.pred,
971  "Main prediction unavailable in the \"mpeg2_aac_low\" profile\n");
972  ERROR_IF(s->options.ltp,
973  "LTP prediction unavailable in the \"mpeg2_aac_low\" profile\n");
974  WARN_IF(s->options.pns,
975  "PNS unavailable in the \"mpeg2_aac_low\" profile, turning off\n");
976  s->options.pns = 0;
977  } else if (avctx->profile == FF_PROFILE_AAC_LTP) {
978  s->options.ltp = 1;
979  ERROR_IF(s->options.pred,
980  "Main prediction unavailable in the \"aac_ltp\" profile\n");
981  } else if (avctx->profile == FF_PROFILE_AAC_MAIN) {
982  s->options.pred = 1;
983  ERROR_IF(s->options.ltp,
984  "LTP prediction unavailable in the \"aac_main\" profile\n");
985  } else if (s->options.ltp) {
986  avctx->profile = FF_PROFILE_AAC_LTP;
987  WARN_IF(1,
988  "Chainging profile to \"aac_ltp\"\n");
989  ERROR_IF(s->options.pred,
990  "Main prediction unavailable in the \"aac_ltp\" profile\n");
991  } else if (s->options.pred) {
992  avctx->profile = FF_PROFILE_AAC_MAIN;
993  WARN_IF(1,
994  "Chainging profile to \"aac_main\"\n");
995  ERROR_IF(s->options.ltp,
996  "LTP prediction unavailable in the \"aac_main\" profile\n");
997  }
998  s->profile = avctx->profile;
999 
1000  /* Coder limitations */
1001  s->coder = &ff_aac_coders[s->options.coder];
1002  if (s->options.coder == AAC_CODER_ANMR) {
1004  "The ANMR coder is considered experimental, add -strict -2 to enable!\n");
1005  s->options.intensity_stereo = 0;
1006  s->options.pns = 0;
1007  }
1009  "The LPT profile requires experimental compliance, add -strict -2 to enable!\n");
1010 
1011  /* M/S introduces horrible artifacts with multichannel files, this is temporary */
1012  if (s->channels > 3)
1013  s->options.mid_side = 0;
1014 
1015  if ((ret = dsp_init(avctx, s)) < 0)
1016  goto fail;
1017 
1018  if ((ret = alloc_buffers(avctx, s)) < 0)
1019  goto fail;
1020 
1022 
1023  sizes[0] = ff_aac_swb_size_1024[s->samplerate_index];
1024  sizes[1] = ff_aac_swb_size_128[s->samplerate_index];
1025  lengths[0] = ff_aac_num_swb_1024[s->samplerate_index];
1026  lengths[1] = ff_aac_num_swb_128[s->samplerate_index];
1027  for (i = 0; i < s->chan_map[0]; i++)
1028  grouping[i] = s->chan_map[i + 1] == TYPE_CPE;
1029  if ((ret = ff_psy_init(&s->psy, avctx, 2, sizes, lengths,
1030  s->chan_map[0], grouping)) < 0)
1031  goto fail;
1032  s->psypp = ff_psy_preprocess_init(avctx);
1034  s->random_state = 0x1f2e3d4c;
1035 
1036  s->abs_pow34 = abs_pow34_v;
1038 
1039  if (ARCH_X86)
1041 
1042  if (HAVE_MIPSDSP)
1044 
1046  return AVERROR_UNKNOWN;
1047 
1048  ff_af_queue_init(avctx, &s->afq);
1049 
1050  return 0;
1051 fail:
1052  aac_encode_end(avctx);
1053  return ret;
1054 }
1055 
1056 #define AACENC_FLAGS AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
1057 static const AVOption aacenc_options[] = {
1058  {"aac_coder", "Coding algorithm", offsetof(AACEncContext, options.coder), AV_OPT_TYPE_INT, {.i64 = AAC_CODER_TWOLOOP}, 0, AAC_CODER_NB-1, AACENC_FLAGS, "coder"},
1059  {"anmr", "ANMR method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_ANMR}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1060  {"twoloop", "Two loop searching method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_TWOLOOP}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1061  {"fast", "Constant quantizer", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_FAST}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1062  {"aac_ms", "Force M/S stereo coding", offsetof(AACEncContext, options.mid_side), AV_OPT_TYPE_BOOL, {.i64 = -1}, -1, 1, AACENC_FLAGS},
1063  {"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1064  {"aac_pns", "Perceptual noise substitution", offsetof(AACEncContext, options.pns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1065  {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1066  {"aac_ltp", "Long term prediction", offsetof(AACEncContext, options.ltp), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1067  {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1068  {NULL}
1069 };
1070 
1071 static const AVClass aacenc_class = {
1072  "AAC encoder",
1076 };
1077 
1079  { "b", "0" },
1080  { NULL }
1081 };
1082 
1084  .name = "aac",
1085  .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
1086  .type = AVMEDIA_TYPE_AUDIO,
1087  .id = AV_CODEC_ID_AAC,
1088  .priv_data_size = sizeof(AACEncContext),
1089  .init = aac_encode_init,
1090  .encode2 = aac_encode_frame,
1091  .close = aac_encode_end,
1092  .defaults = aac_encode_defaults,
1094  .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
1096  .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
1098  .priv_class = &aacenc_class,
1099 };
#define FF_COMPLIANCE_EXPERIMENTAL
Allow nonstandardized experimental things.
Definition: avcodec.h:2899
float, planar
Definition: samplefmt.h:69
void ff_af_queue_remove(AudioFrameQueue *afq, int nb_samples, int64_t *pts, int64_t *duration)
Remove frame(s) from the queue.
void ff_quantize_band_cost_cache_init(struct AACEncContext *s)
Definition: aacenc.c:79
#define NULL
Definition: coverity.c:32
const char * s
Definition: avisynth_c.h:768
void(* search_for_pns)(struct AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
Definition: aacenc.h:72
Band types following are encoded differently from others.
Definition: aac.h:86
static const uint8_t aac_chan_configs[AAC_MAX_CHANNELS][6]
default channel configurations
Definition: aacenctab.h:47
uint8_t use_kb_window[2]
If set, use Kaiser-Bessel window, otherwise use a sine window.
Definition: aac.h:177
int coder
Definition: aacenc.h:44
AACCoefficientsEncoder ff_aac_coders[AAC_CODER_NB]
Definition: aaccoder.c:897
Definition: vf_geq.c:46
This structure describes decoded (raw) audio or video data.
Definition: frame.h:184
#define FF_ALLOCZ_ARRAY_OR_GOTO(ctx, p, nelem, elsize, label)
Definition: internal.h:157
int grouping[8]
window grouping (for e.g. AAC)
Definition: psymodel.h:81
AVOption.
Definition: opt.h:245
enum RawDataBlockType cur_type
channel group type cur_channel belongs to
Definition: aacenc.h:121
uint8_t ** bands
scalefactor band sizes for possible frame sizes
Definition: psymodel.h:98
Definition: aac.h:224
void(* mdct_calc)(struct FFTContext *s, FFTSample *output, const FFTSample *input)
Definition: fft.h:109
AACQuantizeBandCostCacheEntry quantize_band_cost_cache[256][128]
memoization area for quantize_band_cost
Definition: aacenc.h:128
static void abs_pow34_v(float *out, const float *in, const int size)
Definition: aacenc_utils.h:40
static const AVClass aacenc_class
Definition: aacenc.c:1071
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:206
av_cold void ff_kbd_window_init(float *window, float alpha, int n)
Generate a Kaiser-Bessel Derived Window.
Definition: kbdwin.c:26
int64_t bit_rate
the average bitrate
Definition: avcodec.h:1741
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
#define SCALE_DIFF_ZERO
codebook index corresponding to zero scalefactor indices difference
Definition: aac.h:152
uint8_t window_clipping[8]
set if a certain window is near clipping
Definition: aac.h:191
Definition: aac.h:63
const char * g
Definition: vf_curves.c:112
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
void(* encode_ltp_info)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:63
Definition: aac.h:57
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx)
Cleanup audio preprocessing module.
Definition: psymodel.c:152
#define WARN_IF(cond,...)
Definition: aacenc_utils.h:274
int size
Definition: avcodec.h:1602
const int ff_aac_swb_size_1024_len
Definition: aacenctab.c:108
AACCoefficientsEncoder * coder
Definition: aacenc.h:113
void avpriv_align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: bitstream.c:49
static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
Encode ics_info element.
Definition: aacenc.c:174
int common_window
Set if channels share a common 'IndividualChannelStream' in bitstream.
Definition: aac.h:278
int alloc
number of bits allocated by the psy, or -1 if no allocation was done
Definition: psymodel.h:105
const uint8_t * ff_aac_swb_size_1024[]
Definition: aacenctab.c:99
#define FF_PROFILE_AAC_MAIN
Definition: avcodec.h:3185
int lambda_count
count(lambda), for Qvg reporting
Definition: aacenc.h:120
av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, int num_lens, const uint8_t **bands, const int *num_bands, int num_groups, const uint8_t *group_map)
Initialize psychoacoustic model.
Definition: psymodel.c:31
uint8_t ms_mask[128]
Set if mid/side stereo is used for each scalefactor window band.
Definition: aac.h:281
float lambda
Definition: aacenc.h:117
#define NOISE_PRE
preamble for NOISE_BT, put in bitstream with the first noise band
Definition: aac.h:156
int profile
profile
Definition: avcodec.h:3181
AVCodec.
Definition: avcodec.h:3600
void(* search_for_tns)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:74
static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
Encode spectral coefficients processed by psychoacoustic model.
Definition: aacenc.c:392
Spectral data are scaled white noise not coded in the bitstream.
Definition: aac.h:87
int * num_bands
number of scalefactor bands for possible frame sizes
Definition: psymodel.h:99
static AVOnce aac_table_init
Definition: aacenc.c:51
const uint8_t ff_aac_num_swb_128[]
Definition: aactab.c:55
INTFLOAT pcoeffs[1024]
coefficients for IMDCT, pristine
Definition: aac.h:261
const uint16_t * swb_offset
table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular wind...
Definition: aac.h:181
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: avcodec.h:984
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
supported_samplerates
AACEncOptions options
encoding options
Definition: aacenc.h:97
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:40
AAC encoder context.
Definition: aacenc.h:95
uint8_t
#define av_cold
Definition: attributes.h:82
void(* quant_bands)(int *out, const float *in, const float *scaled, int size, int is_signed, int maxval, const float Q34, const float rounding)
Definition: aacenc.h:131
AVOptions.
int intensity_stereo
Definition: aacenc.h:50
#define WINDOW_FUNC(type)
Definition: aacenc.c:88
LPCContext lpc
used by TNS
Definition: aacenc.h:105
void ff_aac_coder_init_mips(AACEncContext *c)
SingleChannelElement ch[2]
Definition: aac.h:284
int samplerate_index
MPEG-4 samplerate index.
Definition: aacenc.h:106
Definition: aac.h:59
av_cold void ff_af_queue_init(AVCodecContext *avctx, AudioFrameQueue *afq)
Initialize AudioFrameQueue.
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: avcodec.h:1619
const uint8_t * chan_map
channel configuration map
Definition: aacenc.h:108
TemporalNoiseShaping tns
Definition: aac.h:250
const uint8_t ff_aac_scalefactor_bits[121]
Definition: aactab.c:82
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1791
AudioFrameQueue afq
Definition: aacenc.h:123
static AVFrame * frame
static const uint8_t aac_chan_maps[AAC_MAX_CHANNELS][AAC_MAX_CHANNELS]
Table to remap channels from libavcodec's default order to AAC order.
Definition: aacenctab.h:61
#define FF_PROFILE_AAC_LTP
Definition: avcodec.h:3188
uint8_t * data
Definition: avcodec.h:1601
const uint8_t * ff_aac_swb_size_128[]
Definition: aacenctab.c:91
uint32_t tag
Definition: movenc.c:1382
Scalefactor data are intensity stereo positions (in phase).
Definition: aac.h:89
int profile
copied from avctx
Definition: aacenc.h:104
#define AVOnce
Definition: thread.h:158
const OptionDef options[]
Definition: ffserver.c:3969
static void adjust_frame_information(ChannelElement *cpe, int chans)
Produce integer coefficients from scalefactors provided by the model.
Definition: aacenc.c:209
#define av_log(a,...)
float * planar_samples[8]
saved preprocessed input
Definition: aacenc.h:102
static const AVOption aacenc_options[]
Definition: aacenc.c:1057
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
av_default_item_name
static const int sizes[][2]
Definition: img2dec.c:50
#define AVERROR(e)
Definition: error.h:43
const uint8_t ff_aac_num_swb_1024[]
Definition: aactab.c:43
int last_frame
Definition: aacenc.h:115
#define FF_PROFILE_MPEG2_AAC_LOW
Definition: avcodec.h:3193
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:176
float is_ener[128]
Intensity stereo pos (used by encoder)
Definition: aac.h:259
int initial_padding
Audio only.
Definition: avcodec.h:3366
float ff_aac_kbd_long_1024[1024]
Definition: aactab.c:38
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1771
static const int mpeg4audio_sample_rates[16]
Definition: aacenctab.h:74
int amp[4]
Definition: aac.h:228
#define CODEC_FLAG_QSCALE
Definition: avcodec.h:1065
const char * name
Name of the codec implementation.
Definition: avcodec.h:3607
int num_windows
number of windows in a frame
Definition: psymodel.h:80
static void copy_input_samples(AACEncContext *s, const AVFrame *frame)
Definition: aacenc.c:487
uint8_t max_sfb
number of scalefactor bands per group
Definition: aac.h:175
#define ff_mdct_init
Definition: fft.h:169
Definition: aac.h:62
void(* set_special_band_scalefactors)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:71
int num_swb
number of scalefactor window bands
Definition: aac.h:183
int ff_af_queue_add(AudioFrameQueue *afq, const AVFrame *f)
Add a frame to the queue.
#define FFMAX(a, b)
Definition: common.h:94
#define fail()
Definition: checkasm.h:83
void(* search_for_quantizers)(AVCodecContext *avctx, struct AACEncContext *s, SingleChannelElement *sce, const float lambda)
Definition: aacenc.h:56
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:85
#define AACENC_FLAGS
Definition: aacenc.c:1056
INTFLOAT ret_buf[2048]
PCM output buffer.
Definition: aac.h:264
struct AACEncContext::@43 buffer
void(* search_for_is)(struct AACEncContext *s, AVCodecContext *avctx, ChannelElement *cpe)
Definition: aacenc.h:77
void(* abs_pow34)(float *out, const float *in, const int size)
Definition: aacenc.h:130
enum WindowSequence window_sequence[2]
Definition: aac.h:176
INTFLOAT ltp_state[3072]
time signal for LTP
Definition: aac.h:265
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:886
av_cold void ff_lpc_end(LPCContext *s)
Uninitialize LPCContext.
Definition: lpc.c:322
#define AV_CODEC_CAP_SMALL_LAST_FRAME
Codec can be fed a final frame with a smaller size.
Definition: avcodec.h:989
int cur_channel
current channel for coder context
Definition: aacenc.h:114
int last_frame_pb_count
number of bits for the previous frame
Definition: aacenc.h:118
#define FFMIN(a, b)
Definition: common.h:96
static void apply_intensity_stereo(ChannelElement *cpe)
Definition: aacenc.c:259
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: aacenc.c:509
void(* apply_tns_filt)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:68
void(* analyze)(FFPsyContext *ctx, int channel, const float **coeffs, const FFPsyWindowInfo *wi)
Perform psychoacoustic analysis and set band info (threshold, energy) for a group of channels...
Definition: psymodel.h:139
typedef void(APIENTRY *FF_PFNGLACTIVETEXTUREPROC)(GLenum texture)
#define FF_PROFILE_AAC_LOW
Definition: avcodec.h:3186
static const AVCodecDefault aac_encode_defaults[]
Definition: aacenc.c:1078
#define FF_PROFILE_UNKNOWN
Definition: avcodec.h:3182
int pos[4]
Definition: aac.h:227
void(* quantize_and_encode_band)(struct AACEncContext *s, PutBitContext *pb, const float *in, float *out, int size, int scale_idx, int cb, const float lambda, int rtz)
Definition: aacenc.h:60
int channels
channel count
Definition: aacenc.h:107
void(* encode_main_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:64
AAC definitions and structures.
void(* search_for_ltp)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:75
const uint8_t ff_tns_max_bands_1024[]
Definition: aactab.c:1274
static void quantize_bands(int *out, const float *in, const float *scaled, int size, int is_signed, int maxval, const float Q34, const float rounding)
Definition: aacenc_utils.h:65
FFTContext mdct128
short (128 samples) frame transform context
Definition: aacenc.h:100
PutBitContext pb
Definition: aacenc.h:98
static void(*const apply_window[4])(AVFloatDSPContext *fdsp, SingleChannelElement *sce, const float *audio)
Definition: aacenc.c:144
void(* search_for_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:78
#define L(x)
Definition: vp56_arith.h:36
AVFloatDSPContext * fdsp
Definition: aacenc.h:101
int mid_side
Definition: aacenc.h:49
#define FF_ARRAY_ELEMS(a)
void(* adjust_common_pred)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:65
static av_cold int aac_encode_end(AVCodecContext *avctx)
Definition: aacenc.c:855
void ff_aac_dsp_init_x86(AACEncContext *s)
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:2458
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
#define AV_ONCE_INIT
Definition: thread.h:159
#define CLIP_AVOIDANCE_FACTOR
Definition: aac.h:53
Libavcodec external API header.
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
void(* search_for_ms)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:76
static void put_audio_specific_config(AVCodecContext *avctx)
Make AAC audio config object.
Definition: aacenc.c:57
Temporal Noise Shaping.
Definition: aac.h:198
int sample_rate
samples per second
Definition: avcodec.h:2438
float ff_aac_kbd_short_128[128]
Definition: aactab.c:39
static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
Encode MS data.
Definition: aacenc.c:195
FFPsyWindowInfo(* window)(FFPsyContext *ctx, const float *audio, const float *la, int channel, int prev_type)
Suggest window sequence for channel.
Definition: psymodel.h:129
void(* apply_main_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:67
main external API structure.
Definition: avcodec.h:1676
int bits
number of bits used in the bitresevoir
Definition: psymodel.h:104
#define NOISE_PRE_BITS
length of preamble
Definition: aac.h:157
Levinson-Durbin recursion.
Definition: lpc.h:47
IndividualChannelStream ics
Definition: aac.h:249
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(constint16_t *) pi >>8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(constint32_t *) pi >>24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(constfloat *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(constfloat *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(constfloat *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(constdouble *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(constdouble *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(constdouble *) pi *(1U<< 31))))#defineSET_CONV_FUNC_GROUP(ofmt, ifmt) staticvoidset_generic_function(AudioConvert *ac){}voidff_audio_convert_free(AudioConvert **ac){if(!*ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);}AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enumAVSampleFormatout_fmt, enumAVSampleFormatin_fmt, intchannels, intsample_rate, intapply_map){AudioConvert *ac;intin_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) returnNULL;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);returnNULL;}returnac;}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;}elseif(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;elseac->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);returnac;}intff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in){intuse_generic=1;intlen=in->nb_samples;intp;if(ac->dc){av_log(ac->avr, AV_LOG_TRACE,"%dsamples-audio_convert:%sto%s(dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));returnff_convert_dither(ac-> in
void(* encode_window_bands_info)(struct AACEncContext *s, SingleChannelElement *sce, int win, int group_len, const float lambda)
Definition: aacenc.h:58
void(* mark_pns)(struct AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
Definition: aacenc.h:73
int extradata_size
Definition: avcodec.h:1792
uint8_t group_len[8]
Definition: aac.h:179
Replacements for frequently missing libm functions.
float lambda_sum
sum(lambda), for Qvg reporting
Definition: aacenc.h:119
Describe the class of an AVClass context structure.
Definition: log.h:67
static void put_bitstream_info(AACEncContext *s, const char *name)
Write some auxiliary information about the created AAC file.
Definition: aacenc.c:466
const int ff_aac_swb_size_128_len
Definition: aacenctab.c:107
int window_shape
window shape (sine/KBD/whatever)
Definition: psymodel.h:79
static void encode_pulses(AACEncContext *s, Pulse *pulse)
Encode pulse data.
Definition: aacenc.c:373
uint16_t quantize_band_cost_cache_generation
Definition: aacenc.h:127
static av_cold void aac_encode_init_tables(void)
Definition: aacenc.c:911
const uint8_t * swb_sizes
table of scalefactor band sizes for a particular window
Definition: aac.h:182
#define TNS_MAX_ORDER
Definition: aac.h:50
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:119
int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
Check AVPacket size and/or allocate data.
Definition: utils.c:1722
FFPsyContext psy
Definition: aacenc.h:111
const uint32_t ff_aac_scalefactor_code[121]
Definition: aactab.c:63
LongTermPrediction ltp
Definition: aac.h:180
static av_cold int alloc_buffers(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:896
const struct FFPsyModel * model
encoder-specific model functions
Definition: psymodel.h:91
av_cold int ff_lpc_init(LPCContext *s, int blocksize, int max_order, enum FFLPCType lpc_type)
Initialize LPCContext.
Definition: lpc.c:300
#define AAC_MAX_CHANNELS
Definition: aacenctab.h:39
int ms_mode
Signals mid/side stereo flags coding mode (used by encoder)
Definition: aac.h:279
AAC encoder data.
const uint8_t ff_tns_max_bands_128[]
Definition: aactab.c:1286
struct FFPsyPreprocessContext * psypp
Definition: aacenc.h:112
#define NOISE_OFFSET
subtracted from global gain, used as offset for the preamble
Definition: aac.h:158
void(* encode_tns_info)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:62
int global_quality
Global quality for codecs which cannot change it per frame.
Definition: avcodec.h:1757
uint8_t zeroes[128]
band is not coded (used by encoder)
Definition: aac.h:257
int sf_idx[128]
scalefactor indices (used by encoder)
Definition: aac.h:256
AVCodec ff_aac_encoder
Definition: aacenc.c:1083
struct FFPsyContext::@104 bitres
uint8_t is_mode
Set if any bands have been encoded using intensity stereo (used by encoder)
Definition: aac.h:280
INTFLOAT coeffs[1024]
coefficients for IMDCT, maybe processed
Definition: aac.h:262
const int avpriv_mpeg4audio_sample_rates[16]
Definition: mpeg4audio.c:57
Scalefactor data are intensity stereo positions (out of phase).
Definition: aac.h:88
Y Spectral Band Replication.
Definition: mpeg4audio.h:65
float * samples
Definition: aacenc.h:136
uint8_t prediction_used[41]
Definition: aac.h:190
static av_cold int aac_encode_init(AVCodecContext *avctx)
Definition: aacenc.c:916
common internal api header.
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:101
if(ret< 0)
Definition: vf_mcdeint.c:282
AAC encoder utilities.
Single Channel Element - used for both SCE and LFE elements.
Definition: aac.h:248
windowing related information
Definition: psymodel.h:77
#define ff_mdct_end
Definition: fft.h:170
av_cold struct FFPsyPreprocessContext * ff_psy_preprocess_init(AVCodecContext *avctx)
psychoacoustic model audio preprocessing initialization
Definition: psymodel.c:103
const uint16_t *const ff_swb_offset_1024[]
Definition: aactab.c:1232
void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx, float **audio, int channels)
Preprocess several channel in audio frame in order to compress it better.
Definition: psymodel.c:139
static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce)
Encode scalefactors.
Definition: aacenc.c:337
ChannelElement * cpe
channel elements
Definition: aacenc.h:110
Individual Channel Stream.
Definition: aac.h:174
float clip_avoidance_factor
set if any window is near clipping to the necessary atennuation factor to avoid it ...
Definition: aac.h:192
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:48
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(constuint8_t *) pi-0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(constuint8_t *) pi-0x80)*(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(constuint8_t *) pi-0x80)*(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(constint16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(constint16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16,*(constint16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16,*(constint16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(constint32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(constint32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32,*(constint32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32,*(constint32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(constint64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64,*(constint64_t *) pi *(1.0f/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64,*(constint64_t *) pi *(1.0/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(constfloat *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(constfloat *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(constfloat *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(constfloat *) pi *(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(constdouble *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(constdouble *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(constdouble *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(constdouble *) pi *(INT64_C(1)<< 63)))#defineFMT_PAIR_FUNC(out, in) staticconv_func_type *constfmt_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),};staticvoidcpy1(uint8_t **dst, constuint8_t **src, intlen){memcpy(*dst,*src, len);}staticvoidcpy2(uint8_t **dst, constuint8_t **src, intlen){memcpy(*dst,*src, 2 *len);}staticvoidcpy4(uint8_t **dst, constuint8_t **src, intlen){memcpy(*dst,*src, 4 *len);}staticvoidcpy8(uint8_t **dst, constuint8_t **src, intlen){memcpy(*dst,*src, 8 *len);}AudioConvert *swri_audio_convert_alloc(enumAVSampleFormatout_fmt, enumAVSampleFormatin_fmt, intchannels, constint *ch_map, intflags){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) returnNULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) returnNULL;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)){case1:ctx->simd_f=cpy1;break;case2:ctx->simd_f=cpy2;break;case4:ctx->simd_f=cpy4;break;case8:ctx->simd_f=cpy8;break;}}if(HAVE_YASM &&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);returnctx;}voidswri_audio_convert_free(AudioConvert **ctx){av_freep(ctx);}intswri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, intlen){intch;intoff=0;constintos=(out->planar?1:out->ch_count)*out->bps;unsignedmisaligned=0;av_assert0(ctx->channels==out->ch_count);if(ctx->in_simd_align_mask){intplanes=in->planar?in->ch_count:1;unsignedm=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){intplanes=out->planar?out->ch_count:1;unsignedm=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){intplanes=out->planar?out->ch_count:1;for(ch=0;ch< planes;ch++){ctx->simd_f(out-> ch ch
Definition: audioconvert.c:56
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:71
void(* ltp_insert_new_frame)(struct AACEncContext *s)
Definition: aacenc.h:70
#define ERROR_IF(cond,...)
Definition: aacenc_utils.h:268
static void ff_aac_tableinit(void)
Definition: aactab.h:45
#define AV_INPUT_BUFFER_PADDING_SIZE
Required number of additionally allocated bytes at the end of the input bitstream for decoding...
Definition: avcodec.h:734
channel element - generic struct for SCE/CPE/CCE/LFE
Definition: aac.h:275
void * priv_data
Definition: avcodec.h:1718
int start
Definition: aac.h:226
FFTContext mdct1024
long (1024 samples) frame transform context
Definition: aacenc.h:99
int random_state
Definition: aacenc.h:116
static av_always_inline int diff(const uint32_t a, const uint32_t b)
static const int16_t coeffs[]
int channels
number of audio channels
Definition: avcodec.h:2439
int num_pulse
Definition: aac.h:225
AAC_FLOAT lcoeffs[1024]
MDCT of LTP coefficients (used by encoder)
Definition: aac.h:266
static int ff_thread_once(char *control, void(*routine)(void))
Definition: thread.h:161
static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
Encode scalefactor band coding type.
Definition: aacenc.c:323
static void apply_mid_side_stereo(ChannelElement *cpe)
Definition: aacenc.c:288
void ff_af_queue_close(AudioFrameQueue *afq)
Close AudioFrameQueue.
enum BandType band_type[128]
band types
Definition: aac.h:252
#define LIBAVCODEC_IDENT
Definition: version.h:42
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:701
int frame_number
Frame counter, set by libavcodec.
Definition: avcodec.h:2469
FILE * out
Definition: movenc.c:54
#define av_freep(p)
void INT64 start
Definition: avisynth_c.h:690
static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce, int common_window)
Encode one channel of audio data.
Definition: aacenc.c:440
void(* adjust_common_ltp)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:66
static const AVCodecDefault defaults[]
Definition: dcaenc.c:975
int8_t used[MAX_LTP_LONG_SFB]
Definition: aac.h:168
static void apply_window_and_mdct(AACEncContext *s, SingleChannelElement *sce, float *audio)
Definition: aacenc.c:153
const uint16_t *const ff_swb_offset_128[]
Definition: aactab.c:1256
int8_t present
Definition: aac.h:164
uint8_t is_mask[128]
Set if intensity stereo is used (used by encoder)
Definition: aac.h:282
static const int aacenc_profiles[]
Definition: aacenctab.h:121
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:231
AAC data declarations.
av_cold void ff_psy_end(FFPsyContext *ctx)
Cleanup model context at the end.
Definition: psymodel.c:83
This structure stores compressed data.
Definition: avcodec.h:1578
static void avoid_clipping(AACEncContext *s, SingleChannelElement *sce)
Downscale spectral coefficients for near-clipping windows to avoid artifacts.
Definition: aacenc.c:420
int window_type[3]
window type (short/long/transitional, etc.) - current, previous and next
Definition: psymodel.h:78
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:241
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:2894
static av_cold int dsp_init(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:874
void(* update_ltp)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:69
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: avcodec.h:1594
for(j=16;j >0;--j)
int pred
Definition: aacenc.h:48
#define FF_ALLOCZ_OR_GOTO(ctx, p, size, label)
Definition: internal.h:139
void AAC_RENAME() ff_init_ff_sine_windows(int index)
initialize the specified entry of ff_sine_windows
float clipping[8]
maximum absolute normalized intensity in the given window for clip avoidance
Definition: psymodel.h:82
const char * name
Definition: opengl_enc.c:103
bitstream writer API