FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
aacdec_template.c
Go to the documentation of this file.
1 /*
2  * AAC decoder
3  * Copyright (c) 2005-2006 Oded Shimon ( ods15 ods15 dyndns org )
4  * Copyright (c) 2006-2007 Maxim Gavrilov ( maxim.gavrilov gmail com )
5  * Copyright (c) 2008-2013 Alex Converse <alex.converse@gmail.com>
6  *
7  * AAC LATM decoder
8  * Copyright (c) 2008-2010 Paul Kendall <paul@kcbbs.gen.nz>
9  * Copyright (c) 2010 Janne Grunau <janne-libav@jannau.net>
10  *
11  * AAC decoder fixed-point implementation
12  * Copyright (c) 2013
13  * MIPS Technologies, Inc., California.
14  *
15  * This file is part of FFmpeg.
16  *
17  * FFmpeg is free software; you can redistribute it and/or
18  * modify it under the terms of the GNU Lesser General Public
19  * License as published by the Free Software Foundation; either
20  * version 2.1 of the License, or (at your option) any later version.
21  *
22  * FFmpeg is distributed in the hope that it will be useful,
23  * but WITHOUT ANY WARRANTY; without even the implied warranty of
24  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
25  * Lesser General Public License for more details.
26  *
27  * You should have received a copy of the GNU Lesser General Public
28  * License along with FFmpeg; if not, write to the Free Software
29  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30  */
31 
32 /**
33  * @file
34  * AAC decoder
35  * @author Oded Shimon ( ods15 ods15 dyndns org )
36  * @author Maxim Gavrilov ( maxim.gavrilov gmail com )
37  *
38  * AAC decoder fixed-point implementation
39  * @author Stanislav Ocovaj ( stanislav.ocovaj imgtec com )
40  * @author Nedeljko Babic ( nedeljko.babic imgtec com )
41  */
42 
43 /*
44  * supported tools
45  *
46  * Support? Name
47  * N (code in SoC repo) gain control
48  * Y block switching
49  * Y window shapes - standard
50  * N window shapes - Low Delay
51  * Y filterbank - standard
52  * N (code in SoC repo) filterbank - Scalable Sample Rate
53  * Y Temporal Noise Shaping
54  * Y Long Term Prediction
55  * Y intensity stereo
56  * Y channel coupling
57  * Y frequency domain prediction
58  * Y Perceptual Noise Substitution
59  * Y Mid/Side stereo
60  * N Scalable Inverse AAC Quantization
61  * N Frequency Selective Switch
62  * N upsampling filter
63  * Y quantization & coding - AAC
64  * N quantization & coding - TwinVQ
65  * N quantization & coding - BSAC
66  * N AAC Error Resilience tools
67  * N Error Resilience payload syntax
68  * N Error Protection tool
69  * N CELP
70  * N Silence Compression
71  * N HVXC
72  * N HVXC 4kbits/s VR
73  * N Structured Audio tools
74  * N Structured Audio Sample Bank Format
75  * N MIDI
76  * N Harmonic and Individual Lines plus Noise
77  * N Text-To-Speech Interface
78  * Y Spectral Band Replication
79  * Y (not in this code) Layer-1
80  * Y (not in this code) Layer-2
81  * Y (not in this code) Layer-3
82  * N SinuSoidal Coding (Transient, Sinusoid, Noise)
83  * Y Parametric Stereo
84  * N Direct Stream Transfer
85  * Y (not in fixed point code) Enhanced AAC Low Delay (ER AAC ELD)
86  *
87  * Note: - HE AAC v1 comprises LC AAC with Spectral Band Replication.
88  * - HE AAC v2 comprises LC AAC with Spectral Band Replication and
89  Parametric Stereo.
90  */
91 
92 #include "libavutil/thread.h"
93 
95 static VLC vlc_spectral[11];
96 
97 static int output_configure(AACContext *ac,
98  uint8_t layout_map[MAX_ELEM_ID*4][3], int tags,
99  enum OCStatus oc_type, int get_new_frame);
100 
101 #define overread_err "Input buffer exhausted before END element found\n"
102 
103 static int count_channels(uint8_t (*layout)[3], int tags)
104 {
105  int i, sum = 0;
106  for (i = 0; i < tags; i++) {
107  int syn_ele = layout[i][0];
108  int pos = layout[i][2];
109  sum += (1 + (syn_ele == TYPE_CPE)) *
110  (pos != AAC_CHANNEL_OFF && pos != AAC_CHANNEL_CC);
111  }
112  return sum;
113 }
114 
115 /**
116  * Check for the channel element in the current channel position configuration.
117  * If it exists, make sure the appropriate element is allocated and map the
118  * channel order to match the internal FFmpeg channel layout.
119  *
120  * @param che_pos current channel position configuration
121  * @param type channel element type
122  * @param id channel element id
123  * @param channels count of the number of channels in the configuration
124  *
125  * @return Returns error status. 0 - OK, !0 - error
126  */
128  enum ChannelPosition che_pos,
129  int type, int id, int *channels)
130 {
131  if (*channels >= MAX_CHANNELS)
132  return AVERROR_INVALIDDATA;
133  if (che_pos) {
134  if (!ac->che[type][id]) {
135  if (!(ac->che[type][id] = av_mallocz(sizeof(ChannelElement))))
136  return AVERROR(ENOMEM);
138  }
139  if (type != TYPE_CCE) {
140  if (*channels >= MAX_CHANNELS - (type == TYPE_CPE || (type == TYPE_SCE && ac->oc[1].m4ac.ps == 1))) {
141  av_log(ac->avctx, AV_LOG_ERROR, "Too many channels\n");
142  return AVERROR_INVALIDDATA;
143  }
144  ac->output_element[(*channels)++] = &ac->che[type][id]->ch[0];
145  if (type == TYPE_CPE ||
146  (type == TYPE_SCE && ac->oc[1].m4ac.ps == 1)) {
147  ac->output_element[(*channels)++] = &ac->che[type][id]->ch[1];
148  }
149  }
150  } else {
151  if (ac->che[type][id])
153  av_freep(&ac->che[type][id]);
154  }
155  return 0;
156 }
157 
159 {
160  AACContext *ac = avctx->priv_data;
161  int type, id, ch, ret;
162 
163  /* set channel pointers to internal buffers by default */
164  for (type = 0; type < 4; type++) {
165  for (id = 0; id < MAX_ELEM_ID; id++) {
166  ChannelElement *che = ac->che[type][id];
167  if (che) {
168  che->ch[0].ret = che->ch[0].ret_buf;
169  che->ch[1].ret = che->ch[1].ret_buf;
170  }
171  }
172  }
173 
174  /* get output buffer */
175  av_frame_unref(ac->frame);
176  if (!avctx->channels)
177  return 1;
178 
179  ac->frame->nb_samples = 2048;
180  if ((ret = ff_get_buffer(avctx, ac->frame, 0)) < 0)
181  return ret;
182 
183  /* map output channel pointers to AVFrame data */
184  for (ch = 0; ch < avctx->channels; ch++) {
185  if (ac->output_element[ch])
186  ac->output_element[ch]->ret = (INTFLOAT *)ac->frame->extended_data[ch];
187  }
188 
189  return 0;
190 }
191 
193  uint64_t av_position;
197 };
198 
199 static int assign_pair(struct elem_to_channel e2c_vec[MAX_ELEM_ID],
200  uint8_t (*layout_map)[3], int offset, uint64_t left,
201  uint64_t right, int pos)
202 {
203  if (layout_map[offset][0] == TYPE_CPE) {
204  e2c_vec[offset] = (struct elem_to_channel) {
205  .av_position = left | right,
206  .syn_ele = TYPE_CPE,
207  .elem_id = layout_map[offset][1],
208  .aac_position = pos
209  };
210  return 1;
211  } else {
212  e2c_vec[offset] = (struct elem_to_channel) {
213  .av_position = left,
214  .syn_ele = TYPE_SCE,
215  .elem_id = layout_map[offset][1],
216  .aac_position = pos
217  };
218  e2c_vec[offset + 1] = (struct elem_to_channel) {
219  .av_position = right,
220  .syn_ele = TYPE_SCE,
221  .elem_id = layout_map[offset + 1][1],
222  .aac_position = pos
223  };
224  return 2;
225  }
226 }
227 
228 static int count_paired_channels(uint8_t (*layout_map)[3], int tags, int pos,
229  int *current)
230 {
231  int num_pos_channels = 0;
232  int first_cpe = 0;
233  int sce_parity = 0;
234  int i;
235  for (i = *current; i < tags; i++) {
236  if (layout_map[i][2] != pos)
237  break;
238  if (layout_map[i][0] == TYPE_CPE) {
239  if (sce_parity) {
240  if (pos == AAC_CHANNEL_FRONT && !first_cpe) {
241  sce_parity = 0;
242  } else {
243  return -1;
244  }
245  }
246  num_pos_channels += 2;
247  first_cpe = 1;
248  } else {
249  num_pos_channels++;
250  sce_parity ^= 1;
251  }
252  }
253  if (sce_parity &&
254  ((pos == AAC_CHANNEL_FRONT && first_cpe) || pos == AAC_CHANNEL_SIDE))
255  return -1;
256  *current = i;
257  return num_pos_channels;
258 }
259 
260 static uint64_t sniff_channel_order(uint8_t (*layout_map)[3], int tags)
261 {
262  int i, n, total_non_cc_elements;
263  struct elem_to_channel e2c_vec[4 * MAX_ELEM_ID] = { { 0 } };
264  int num_front_channels, num_side_channels, num_back_channels;
265  uint64_t layout;
266 
267  if (FF_ARRAY_ELEMS(e2c_vec) < tags)
268  return 0;
269 
270  i = 0;
271  num_front_channels =
272  count_paired_channels(layout_map, tags, AAC_CHANNEL_FRONT, &i);
273  if (num_front_channels < 0)
274  return 0;
275  num_side_channels =
276  count_paired_channels(layout_map, tags, AAC_CHANNEL_SIDE, &i);
277  if (num_side_channels < 0)
278  return 0;
279  num_back_channels =
280  count_paired_channels(layout_map, tags, AAC_CHANNEL_BACK, &i);
281  if (num_back_channels < 0)
282  return 0;
283 
284  if (num_side_channels == 0 && num_back_channels >= 4) {
285  num_side_channels = 2;
286  num_back_channels -= 2;
287  }
288 
289  i = 0;
290  if (num_front_channels & 1) {
291  e2c_vec[i] = (struct elem_to_channel) {
293  .syn_ele = TYPE_SCE,
294  .elem_id = layout_map[i][1],
295  .aac_position = AAC_CHANNEL_FRONT
296  };
297  i++;
298  num_front_channels--;
299  }
300  if (num_front_channels >= 4) {
301  i += assign_pair(e2c_vec, layout_map, i,
305  num_front_channels -= 2;
306  }
307  if (num_front_channels >= 2) {
308  i += assign_pair(e2c_vec, layout_map, i,
312  num_front_channels -= 2;
313  }
314  while (num_front_channels >= 2) {
315  i += assign_pair(e2c_vec, layout_map, i,
316  UINT64_MAX,
317  UINT64_MAX,
319  num_front_channels -= 2;
320  }
321 
322  if (num_side_channels >= 2) {
323  i += assign_pair(e2c_vec, layout_map, i,
327  num_side_channels -= 2;
328  }
329  while (num_side_channels >= 2) {
330  i += assign_pair(e2c_vec, layout_map, i,
331  UINT64_MAX,
332  UINT64_MAX,
334  num_side_channels -= 2;
335  }
336 
337  while (num_back_channels >= 4) {
338  i += assign_pair(e2c_vec, layout_map, i,
339  UINT64_MAX,
340  UINT64_MAX,
342  num_back_channels -= 2;
343  }
344  if (num_back_channels >= 2) {
345  i += assign_pair(e2c_vec, layout_map, i,
349  num_back_channels -= 2;
350  }
351  if (num_back_channels) {
352  e2c_vec[i] = (struct elem_to_channel) {
354  .syn_ele = TYPE_SCE,
355  .elem_id = layout_map[i][1],
356  .aac_position = AAC_CHANNEL_BACK
357  };
358  i++;
359  num_back_channels--;
360  }
361 
362  if (i < tags && layout_map[i][2] == AAC_CHANNEL_LFE) {
363  e2c_vec[i] = (struct elem_to_channel) {
365  .syn_ele = TYPE_LFE,
366  .elem_id = layout_map[i][1],
367  .aac_position = AAC_CHANNEL_LFE
368  };
369  i++;
370  }
371  while (i < tags && layout_map[i][2] == AAC_CHANNEL_LFE) {
372  e2c_vec[i] = (struct elem_to_channel) {
373  .av_position = UINT64_MAX,
374  .syn_ele = TYPE_LFE,
375  .elem_id = layout_map[i][1],
376  .aac_position = AAC_CHANNEL_LFE
377  };
378  i++;
379  }
380 
381  // Must choose a stable sort
382  total_non_cc_elements = n = i;
383  do {
384  int next_n = 0;
385  for (i = 1; i < n; i++)
386  if (e2c_vec[i - 1].av_position > e2c_vec[i].av_position) {
387  FFSWAP(struct elem_to_channel, e2c_vec[i - 1], e2c_vec[i]);
388  next_n = i;
389  }
390  n = next_n;
391  } while (n > 0);
392 
393  layout = 0;
394  for (i = 0; i < total_non_cc_elements; i++) {
395  layout_map[i][0] = e2c_vec[i].syn_ele;
396  layout_map[i][1] = e2c_vec[i].elem_id;
397  layout_map[i][2] = e2c_vec[i].aac_position;
398  if (e2c_vec[i].av_position != UINT64_MAX) {
399  layout |= e2c_vec[i].av_position;
400  }
401  }
402 
403  return layout;
404 }
405 
406 /**
407  * Save current output configuration if and only if it has been locked.
408  */
410  int pushed = 0;
411 
412  if (ac->oc[1].status == OC_LOCKED || ac->oc[0].status == OC_NONE) {
413  ac->oc[0] = ac->oc[1];
414  pushed = 1;
415  }
416  ac->oc[1].status = OC_NONE;
417  return pushed;
418 }
419 
420 /**
421  * Restore the previous output configuration if and only if the current
422  * configuration is unlocked.
423  */
425  if (ac->oc[1].status != OC_LOCKED && ac->oc[0].status != OC_NONE) {
426  ac->oc[1] = ac->oc[0];
427  ac->avctx->channels = ac->oc[1].channels;
428  ac->avctx->channel_layout = ac->oc[1].channel_layout;
429  output_configure(ac, ac->oc[1].layout_map, ac->oc[1].layout_map_tags,
430  ac->oc[1].status, 0);
431  }
432 }
433 
434 /**
435  * Configure output channel order based on the current program
436  * configuration element.
437  *
438  * @return Returns error status. 0 - OK, !0 - error
439  */
441  uint8_t layout_map[MAX_ELEM_ID * 4][3], int tags,
442  enum OCStatus oc_type, int get_new_frame)
443 {
444  AVCodecContext *avctx = ac->avctx;
445  int i, channels = 0, ret;
446  uint64_t layout = 0;
447  uint8_t id_map[TYPE_END][MAX_ELEM_ID] = {{ 0 }};
448  uint8_t type_counts[TYPE_END] = { 0 };
449 
450  if (ac->oc[1].layout_map != layout_map) {
451  memcpy(ac->oc[1].layout_map, layout_map, tags * sizeof(layout_map[0]));
452  ac->oc[1].layout_map_tags = tags;
453  }
454  for (i = 0; i < tags; i++) {
455  int type = layout_map[i][0];
456  int id = layout_map[i][1];
457  id_map[type][id] = type_counts[type]++;
458  if (id_map[type][id] >= MAX_ELEM_ID) {
459  avpriv_request_sample(ac->avctx, "Too large remapped id");
460  return AVERROR_PATCHWELCOME;
461  }
462  }
463  // Try to sniff a reasonable channel order, otherwise output the
464  // channels in the order the PCE declared them.
466  layout = sniff_channel_order(layout_map, tags);
467  for (i = 0; i < tags; i++) {
468  int type = layout_map[i][0];
469  int id = layout_map[i][1];
470  int iid = id_map[type][id];
471  int position = layout_map[i][2];
472  // Allocate or free elements depending on if they are in the
473  // current program configuration.
474  ret = che_configure(ac, position, type, iid, &channels);
475  if (ret < 0)
476  return ret;
477  ac->tag_che_map[type][id] = ac->che[type][iid];
478  }
479  if (ac->oc[1].m4ac.ps == 1 && channels == 2) {
480  if (layout == AV_CH_FRONT_CENTER) {
482  } else {
483  layout = 0;
484  }
485  }
486 
487  if (layout) avctx->channel_layout = layout;
488  ac->oc[1].channel_layout = layout;
489  avctx->channels = ac->oc[1].channels = channels;
490  ac->oc[1].status = oc_type;
491 
492  if (get_new_frame) {
493  if ((ret = frame_configure_elements(ac->avctx)) < 0)
494  return ret;
495  }
496 
497  return 0;
498 }
499 
500 static void flush(AVCodecContext *avctx)
501 {
502  AACContext *ac= avctx->priv_data;
503  int type, i, j;
504 
505  for (type = 3; type >= 0; type--) {
506  for (i = 0; i < MAX_ELEM_ID; i++) {
507  ChannelElement *che = ac->che[type][i];
508  if (che) {
509  for (j = 0; j <= 1; j++) {
510  memset(che->ch[j].saved, 0, sizeof(che->ch[j].saved));
511  }
512  }
513  }
514  }
515 }
516 
517 /**
518  * Set up channel positions based on a default channel configuration
519  * as specified in table 1.17.
520  *
521  * @return Returns error status. 0 - OK, !0 - error
522  */
524  uint8_t (*layout_map)[3],
525  int *tags,
526  int channel_config)
527 {
528  if (channel_config < 1 || (channel_config > 7 && channel_config < 11) ||
529  channel_config > 12) {
530  av_log(avctx, AV_LOG_ERROR,
531  "invalid default channel configuration (%d)\n",
532  channel_config);
533  return AVERROR_INVALIDDATA;
534  }
535  *tags = tags_per_config[channel_config];
536  memcpy(layout_map, aac_channel_layout_map[channel_config - 1],
537  *tags * sizeof(*layout_map));
538 
539  /*
540  * AAC specification has 7.1(wide) as a default layout for 8-channel streams.
541  * However, at least Nero AAC encoder encodes 7.1 streams using the default
542  * channel config 7, mapping the side channels of the original audio stream
543  * to the second AAC_CHANNEL_FRONT pair in the AAC stream. Similarly, e.g. FAAD
544  * decodes the second AAC_CHANNEL_FRONT pair as side channels, therefore decoding
545  * the incorrect streams as if they were correct (and as the encoder intended).
546  *
547  * As actual intended 7.1(wide) streams are very rare, default to assuming a
548  * 7.1 layout was intended.
549  */
550  if (channel_config == 7 && avctx->strict_std_compliance < FF_COMPLIANCE_STRICT) {
551  av_log(avctx, AV_LOG_INFO, "Assuming an incorrectly encoded 7.1 channel layout"
552  " instead of a spec-compliant 7.1(wide) layout, use -strict %d to decode"
553  " according to the specification instead.\n", FF_COMPLIANCE_STRICT);
554  layout_map[2][2] = AAC_CHANNEL_SIDE;
555  }
556 
557  return 0;
558 }
559 
560 static ChannelElement *get_che(AACContext *ac, int type, int elem_id)
561 {
562  /* For PCE based channel configurations map the channels solely based
563  * on tags. */
564  if (!ac->oc[1].m4ac.chan_config) {
565  return ac->tag_che_map[type][elem_id];
566  }
567  // Allow single CPE stereo files to be signalled with mono configuration.
568  if (!ac->tags_mapped && type == TYPE_CPE &&
569  ac->oc[1].m4ac.chan_config == 1) {
570  uint8_t layout_map[MAX_ELEM_ID*4][3];
571  int layout_map_tags;
573 
574  av_log(ac->avctx, AV_LOG_DEBUG, "mono with CPE\n");
575 
576  if (set_default_channel_config(ac->avctx, layout_map,
577  &layout_map_tags, 2) < 0)
578  return NULL;
579  if (output_configure(ac, layout_map, layout_map_tags,
580  OC_TRIAL_FRAME, 1) < 0)
581  return NULL;
582 
583  ac->oc[1].m4ac.chan_config = 2;
584  ac->oc[1].m4ac.ps = 0;
585  }
586  // And vice-versa
587  if (!ac->tags_mapped && type == TYPE_SCE &&
588  ac->oc[1].m4ac.chan_config == 2) {
589  uint8_t layout_map[MAX_ELEM_ID * 4][3];
590  int layout_map_tags;
592 
593  av_log(ac->avctx, AV_LOG_DEBUG, "stereo with SCE\n");
594 
595  if (set_default_channel_config(ac->avctx, layout_map,
596  &layout_map_tags, 1) < 0)
597  return NULL;
598  if (output_configure(ac, layout_map, layout_map_tags,
599  OC_TRIAL_FRAME, 1) < 0)
600  return NULL;
601 
602  ac->oc[1].m4ac.chan_config = 1;
603  if (ac->oc[1].m4ac.sbr)
604  ac->oc[1].m4ac.ps = -1;
605  }
606  /* For indexed channel configurations map the channels solely based
607  * on position. */
608  switch (ac->oc[1].m4ac.chan_config) {
609  case 12:
610  case 7:
611  if (ac->tags_mapped == 3 && type == TYPE_CPE) {
612  ac->tags_mapped++;
613  return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][2];
614  }
615  case 11:
616  if (ac->tags_mapped == 2 &&
617  ac->oc[1].m4ac.chan_config == 11 &&
618  type == TYPE_SCE) {
619  ac->tags_mapped++;
620  return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][1];
621  }
622  case 6:
623  /* Some streams incorrectly code 5.1 audio as
624  * SCE[0] CPE[0] CPE[1] SCE[1]
625  * instead of
626  * SCE[0] CPE[0] CPE[1] LFE[0].
627  * If we seem to have encountered such a stream, transfer
628  * the LFE[0] element to the SCE[1]'s mapping */
629  if (ac->tags_mapped == tags_per_config[ac->oc[1].m4ac.chan_config] - 1 && (type == TYPE_LFE || type == TYPE_SCE)) {
630  if (!ac->warned_remapping_once && (type != TYPE_LFE || elem_id != 0)) {
632  "This stream seems to incorrectly report its last channel as %s[%d], mapping to LFE[0]\n",
633  type == TYPE_SCE ? "SCE" : "LFE", elem_id);
634  ac->warned_remapping_once++;
635  }
636  ac->tags_mapped++;
637  return ac->tag_che_map[type][elem_id] = ac->che[TYPE_LFE][0];
638  }
639  case 5:
640  if (ac->tags_mapped == 2 && type == TYPE_CPE) {
641  ac->tags_mapped++;
642  return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][1];
643  }
644  case 4:
645  /* Some streams incorrectly code 4.0 audio as
646  * SCE[0] CPE[0] LFE[0]
647  * instead of
648  * SCE[0] CPE[0] SCE[1].
649  * If we seem to have encountered such a stream, transfer
650  * the SCE[1] element to the LFE[0]'s mapping */
651  if (ac->tags_mapped == tags_per_config[ac->oc[1].m4ac.chan_config] - 1 && (type == TYPE_LFE || type == TYPE_SCE)) {
652  if (!ac->warned_remapping_once && (type != TYPE_SCE || elem_id != 1)) {
654  "This stream seems to incorrectly report its last channel as %s[%d], mapping to SCE[1]\n",
655  type == TYPE_SCE ? "SCE" : "LFE", elem_id);
656  ac->warned_remapping_once++;
657  }
658  ac->tags_mapped++;
659  return ac->tag_che_map[type][elem_id] = ac->che[TYPE_SCE][1];
660  }
661  if (ac->tags_mapped == 2 &&
662  ac->oc[1].m4ac.chan_config == 4 &&
663  type == TYPE_SCE) {
664  ac->tags_mapped++;
665  return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][1];
666  }
667  case 3:
668  case 2:
669  if (ac->tags_mapped == (ac->oc[1].m4ac.chan_config != 2) &&
670  type == TYPE_CPE) {
671  ac->tags_mapped++;
672  return ac->tag_che_map[TYPE_CPE][elem_id] = ac->che[TYPE_CPE][0];
673  } else if (ac->oc[1].m4ac.chan_config == 2) {
674  return NULL;
675  }
676  case 1:
677  if (!ac->tags_mapped && type == TYPE_SCE) {
678  ac->tags_mapped++;
679  return ac->tag_che_map[TYPE_SCE][elem_id] = ac->che[TYPE_SCE][0];
680  }
681  default:
682  return NULL;
683  }
684 }
685 
686 /**
687  * Decode an array of 4 bit element IDs, optionally interleaved with a
688  * stereo/mono switching bit.
689  *
690  * @param type speaker type/position for these channels
691  */
692 static void decode_channel_map(uint8_t layout_map[][3],
693  enum ChannelPosition type,
694  GetBitContext *gb, int n)
695 {
696  while (n--) {
697  enum RawDataBlockType syn_ele;
698  switch (type) {
699  case AAC_CHANNEL_FRONT:
700  case AAC_CHANNEL_BACK:
701  case AAC_CHANNEL_SIDE:
702  syn_ele = get_bits1(gb);
703  break;
704  case AAC_CHANNEL_CC:
705  skip_bits1(gb);
706  syn_ele = TYPE_CCE;
707  break;
708  case AAC_CHANNEL_LFE:
709  syn_ele = TYPE_LFE;
710  break;
711  default:
712  // AAC_CHANNEL_OFF has no channel map
713  av_assert0(0);
714  }
715  layout_map[0][0] = syn_ele;
716  layout_map[0][1] = get_bits(gb, 4);
717  layout_map[0][2] = type;
718  layout_map++;
719  }
720 }
721 
722 static inline void relative_align_get_bits(GetBitContext *gb,
723  int reference_position) {
724  int n = (reference_position - get_bits_count(gb) & 7);
725  if (n)
726  skip_bits(gb, n);
727 }
728 
729 /**
730  * Decode program configuration element; reference: table 4.2.
731  *
732  * @return Returns error status. 0 - OK, !0 - error
733  */
734 static int decode_pce(AVCodecContext *avctx, MPEG4AudioConfig *m4ac,
735  uint8_t (*layout_map)[3],
736  GetBitContext *gb, int byte_align_ref)
737 {
738  int num_front, num_side, num_back, num_lfe, num_assoc_data, num_cc;
739  int sampling_index;
740  int comment_len;
741  int tags;
742 
743  skip_bits(gb, 2); // object_type
744 
745  sampling_index = get_bits(gb, 4);
746  if (m4ac->sampling_index != sampling_index)
747  av_log(avctx, AV_LOG_WARNING,
748  "Sample rate index in program config element does not "
749  "match the sample rate index configured by the container.\n");
750 
751  num_front = get_bits(gb, 4);
752  num_side = get_bits(gb, 4);
753  num_back = get_bits(gb, 4);
754  num_lfe = get_bits(gb, 2);
755  num_assoc_data = get_bits(gb, 3);
756  num_cc = get_bits(gb, 4);
757 
758  if (get_bits1(gb))
759  skip_bits(gb, 4); // mono_mixdown_tag
760  if (get_bits1(gb))
761  skip_bits(gb, 4); // stereo_mixdown_tag
762 
763  if (get_bits1(gb))
764  skip_bits(gb, 3); // mixdown_coeff_index and pseudo_surround
765 
766  if (get_bits_left(gb) < 5 * (num_front + num_side + num_back + num_cc) + 4 *(num_lfe + num_assoc_data + num_cc)) {
767  av_log(avctx, AV_LOG_ERROR, "decode_pce: " overread_err);
768  return -1;
769  }
770  decode_channel_map(layout_map , AAC_CHANNEL_FRONT, gb, num_front);
771  tags = num_front;
772  decode_channel_map(layout_map + tags, AAC_CHANNEL_SIDE, gb, num_side);
773  tags += num_side;
774  decode_channel_map(layout_map + tags, AAC_CHANNEL_BACK, gb, num_back);
775  tags += num_back;
776  decode_channel_map(layout_map + tags, AAC_CHANNEL_LFE, gb, num_lfe);
777  tags += num_lfe;
778 
779  skip_bits_long(gb, 4 * num_assoc_data);
780 
781  decode_channel_map(layout_map + tags, AAC_CHANNEL_CC, gb, num_cc);
782  tags += num_cc;
783 
784  relative_align_get_bits(gb, byte_align_ref);
785 
786  /* comment field, first byte is length */
787  comment_len = get_bits(gb, 8) * 8;
788  if (get_bits_left(gb) < comment_len) {
789  av_log(avctx, AV_LOG_ERROR, "decode_pce: " overread_err);
790  return AVERROR_INVALIDDATA;
791  }
792  skip_bits_long(gb, comment_len);
793  return tags;
794 }
795 
796 /**
797  * Decode GA "General Audio" specific configuration; reference: table 4.1.
798  *
799  * @param ac pointer to AACContext, may be null
800  * @param avctx pointer to AVCCodecContext, used for logging
801  *
802  * @return Returns error status. 0 - OK, !0 - error
803  */
805  GetBitContext *gb,
806  int get_bit_alignment,
807  MPEG4AudioConfig *m4ac,
808  int channel_config)
809 {
810  int extension_flag, ret, ep_config, res_flags;
811  uint8_t layout_map[MAX_ELEM_ID*4][3];
812  int tags = 0;
813 
814 #if USE_FIXED
815  if (get_bits1(gb)) { // frameLengthFlag
816  avpriv_report_missing_feature(avctx, "Fixed point 960/120 MDCT window");
817  return AVERROR_PATCHWELCOME;
818  }
819  m4ac->frame_length_short = 0;
820 #else
821  m4ac->frame_length_short = get_bits1(gb);
822  if (m4ac->frame_length_short && m4ac->sbr == 1) {
823  avpriv_report_missing_feature(avctx, "SBR with 960 frame length");
824  if (ac) ac->warned_960_sbr = 1;
825  m4ac->sbr = 0;
826  m4ac->ps = 0;
827  }
828 #endif
829 
830  if (get_bits1(gb)) // dependsOnCoreCoder
831  skip_bits(gb, 14); // coreCoderDelay
832  extension_flag = get_bits1(gb);
833 
834  if (m4ac->object_type == AOT_AAC_SCALABLE ||
836  skip_bits(gb, 3); // layerNr
837 
838  if (channel_config == 0) {
839  skip_bits(gb, 4); // element_instance_tag
840  tags = decode_pce(avctx, m4ac, layout_map, gb, get_bit_alignment);
841  if (tags < 0)
842  return tags;
843  } else {
844  if ((ret = set_default_channel_config(avctx, layout_map,
845  &tags, channel_config)))
846  return ret;
847  }
848 
849  if (count_channels(layout_map, tags) > 1) {
850  m4ac->ps = 0;
851  } else if (m4ac->sbr == 1 && m4ac->ps == -1)
852  m4ac->ps = 1;
853 
854  if (ac && (ret = output_configure(ac, layout_map, tags, OC_GLOBAL_HDR, 0)))
855  return ret;
856 
857  if (extension_flag) {
858  switch (m4ac->object_type) {
859  case AOT_ER_BSAC:
860  skip_bits(gb, 5); // numOfSubFrame
861  skip_bits(gb, 11); // layer_length
862  break;
863  case AOT_ER_AAC_LC:
864  case AOT_ER_AAC_LTP:
865  case AOT_ER_AAC_SCALABLE:
866  case AOT_ER_AAC_LD:
867  res_flags = get_bits(gb, 3);
868  if (res_flags) {
870  "AAC data resilience (flags %x)",
871  res_flags);
872  return AVERROR_PATCHWELCOME;
873  }
874  break;
875  }
876  skip_bits1(gb); // extensionFlag3 (TBD in version 3)
877  }
878  switch (m4ac->object_type) {
879  case AOT_ER_AAC_LC:
880  case AOT_ER_AAC_LTP:
881  case AOT_ER_AAC_SCALABLE:
882  case AOT_ER_AAC_LD:
883  ep_config = get_bits(gb, 2);
884  if (ep_config) {
886  "epConfig %d", ep_config);
887  return AVERROR_PATCHWELCOME;
888  }
889  }
890  return 0;
891 }
892 
894  GetBitContext *gb,
895  MPEG4AudioConfig *m4ac,
896  int channel_config)
897 {
898  int ret, ep_config, res_flags;
899  uint8_t layout_map[MAX_ELEM_ID*4][3];
900  int tags = 0;
901  const int ELDEXT_TERM = 0;
902 
903  m4ac->ps = 0;
904  m4ac->sbr = 0;
905 #if USE_FIXED
906  if (get_bits1(gb)) { // frameLengthFlag
907  avpriv_request_sample(avctx, "960/120 MDCT window");
908  return AVERROR_PATCHWELCOME;
909  }
910 #else
911  m4ac->frame_length_short = get_bits1(gb);
912 #endif
913  res_flags = get_bits(gb, 3);
914  if (res_flags) {
916  "AAC data resilience (flags %x)",
917  res_flags);
918  return AVERROR_PATCHWELCOME;
919  }
920 
921  if (get_bits1(gb)) { // ldSbrPresentFlag
923  "Low Delay SBR");
924  return AVERROR_PATCHWELCOME;
925  }
926 
927  while (get_bits(gb, 4) != ELDEXT_TERM) {
928  int len = get_bits(gb, 4);
929  if (len == 15)
930  len += get_bits(gb, 8);
931  if (len == 15 + 255)
932  len += get_bits(gb, 16);
933  if (get_bits_left(gb) < len * 8 + 4) {
935  return AVERROR_INVALIDDATA;
936  }
937  skip_bits_long(gb, 8 * len);
938  }
939 
940  if ((ret = set_default_channel_config(avctx, layout_map,
941  &tags, channel_config)))
942  return ret;
943 
944  if (ac && (ret = output_configure(ac, layout_map, tags, OC_GLOBAL_HDR, 0)))
945  return ret;
946 
947  ep_config = get_bits(gb, 2);
948  if (ep_config) {
950  "epConfig %d", ep_config);
951  return AVERROR_PATCHWELCOME;
952  }
953  return 0;
954 }
955 
956 /**
957  * Decode audio specific configuration; reference: table 1.13.
958  *
959  * @param ac pointer to AACContext, may be null
960  * @param avctx pointer to AVCCodecContext, used for logging
961  * @param m4ac pointer to MPEG4AudioConfig, used for parsing
962  * @param gb buffer holding an audio specific config
963  * @param get_bit_alignment relative alignment for byte align operations
964  * @param sync_extension look for an appended sync extension
965  *
966  * @return Returns error status or number of consumed bits. <0 - error
967  */
969  AVCodecContext *avctx,
970  MPEG4AudioConfig *m4ac,
971  GetBitContext *gb,
972  int get_bit_alignment,
973  int sync_extension)
974 {
975  int i, ret;
976  GetBitContext gbc = *gb;
977 
978  if ((i = ff_mpeg4audio_get_config_gb(m4ac, &gbc, sync_extension)) < 0)
979  return AVERROR_INVALIDDATA;
980 
981  if (m4ac->sampling_index > 12) {
982  av_log(avctx, AV_LOG_ERROR,
983  "invalid sampling rate index %d\n",
984  m4ac->sampling_index);
985  return AVERROR_INVALIDDATA;
986  }
987  if (m4ac->object_type == AOT_ER_AAC_LD &&
988  (m4ac->sampling_index < 3 || m4ac->sampling_index > 7)) {
989  av_log(avctx, AV_LOG_ERROR,
990  "invalid low delay sampling rate index %d\n",
991  m4ac->sampling_index);
992  return AVERROR_INVALIDDATA;
993  }
994 
995  skip_bits_long(gb, i);
996 
997  switch (m4ac->object_type) {
998  case AOT_AAC_MAIN:
999  case AOT_AAC_LC:
1000  case AOT_AAC_LTP:
1001  case AOT_ER_AAC_LC:
1002  case AOT_ER_AAC_LD:
1003  if ((ret = decode_ga_specific_config(ac, avctx, gb, get_bit_alignment,
1004  m4ac, m4ac->chan_config)) < 0)
1005  return ret;
1006  break;
1007  case AOT_ER_AAC_ELD:
1008  if ((ret = decode_eld_specific_config(ac, avctx, gb,
1009  m4ac, m4ac->chan_config)) < 0)
1010  return ret;
1011  break;
1012  default:
1014  "Audio object type %s%d",
1015  m4ac->sbr == 1 ? "SBR+" : "",
1016  m4ac->object_type);
1017  return AVERROR(ENOSYS);
1018  }
1019 
1020  ff_dlog(avctx,
1021  "AOT %d chan config %d sampling index %d (%d) SBR %d PS %d\n",
1022  m4ac->object_type, m4ac->chan_config, m4ac->sampling_index,
1023  m4ac->sample_rate, m4ac->sbr,
1024  m4ac->ps);
1025 
1026  return get_bits_count(gb);
1027 }
1028 
1030  AVCodecContext *avctx,
1031  MPEG4AudioConfig *m4ac,
1032  const uint8_t *data, int64_t bit_size,
1033  int sync_extension)
1034 {
1035  int i, ret;
1036  GetBitContext gb;
1037 
1038  if (bit_size < 0 || bit_size > INT_MAX) {
1039  av_log(avctx, AV_LOG_ERROR, "Audio specific config size is invalid\n");
1040  return AVERROR_INVALIDDATA;
1041  }
1042 
1043  ff_dlog(avctx, "audio specific config size %d\n", (int)bit_size >> 3);
1044  for (i = 0; i < bit_size >> 3; i++)
1045  ff_dlog(avctx, "%02x ", data[i]);
1046  ff_dlog(avctx, "\n");
1047 
1048  if ((ret = init_get_bits(&gb, data, bit_size)) < 0)
1049  return ret;
1050 
1051  return decode_audio_specific_config_gb(ac, avctx, m4ac, &gb, 0,
1052  sync_extension);
1053 }
1054 
1055 /**
1056  * linear congruential pseudorandom number generator
1057  *
1058  * @param previous_val pointer to the current state of the generator
1059  *
1060  * @return Returns a 32-bit pseudorandom integer
1061  */
1062 static av_always_inline int lcg_random(unsigned previous_val)
1063 {
1064  union { unsigned u; int s; } v = { previous_val * 1664525u + 1013904223 };
1065  return v.s;
1066 }
1067 
1069 {
1070  int i;
1071  for (i = 0; i < MAX_PREDICTORS; i++)
1072  reset_predict_state(&ps[i]);
1073 }
1074 
1075 static int sample_rate_idx (int rate)
1076 {
1077  if (92017 <= rate) return 0;
1078  else if (75132 <= rate) return 1;
1079  else if (55426 <= rate) return 2;
1080  else if (46009 <= rate) return 3;
1081  else if (37566 <= rate) return 4;
1082  else if (27713 <= rate) return 5;
1083  else if (23004 <= rate) return 6;
1084  else if (18783 <= rate) return 7;
1085  else if (13856 <= rate) return 8;
1086  else if (11502 <= rate) return 9;
1087  else if (9391 <= rate) return 10;
1088  else return 11;
1089 }
1090 
1091 static void reset_predictor_group(PredictorState *ps, int group_num)
1092 {
1093  int i;
1094  for (i = group_num - 1; i < MAX_PREDICTORS; i += 30)
1095  reset_predict_state(&ps[i]);
1096 }
1097 
1098 #define AAC_INIT_VLC_STATIC(num, size) \
1099  INIT_VLC_STATIC(&vlc_spectral[num], 8, ff_aac_spectral_sizes[num], \
1100  ff_aac_spectral_bits[num], sizeof(ff_aac_spectral_bits[num][0]), \
1101  sizeof(ff_aac_spectral_bits[num][0]), \
1102  ff_aac_spectral_codes[num], sizeof(ff_aac_spectral_codes[num][0]), \
1103  sizeof(ff_aac_spectral_codes[num][0]), \
1104  size);
1105 
1106 static void aacdec_init(AACContext *ac);
1107 
1109 {
1110  AAC_INIT_VLC_STATIC( 0, 304);
1111  AAC_INIT_VLC_STATIC( 1, 270);
1112  AAC_INIT_VLC_STATIC( 2, 550);
1113  AAC_INIT_VLC_STATIC( 3, 300);
1114  AAC_INIT_VLC_STATIC( 4, 328);
1115  AAC_INIT_VLC_STATIC( 5, 294);
1116  AAC_INIT_VLC_STATIC( 6, 306);
1117  AAC_INIT_VLC_STATIC( 7, 268);
1118  AAC_INIT_VLC_STATIC( 8, 510);
1119  AAC_INIT_VLC_STATIC( 9, 366);
1120  AAC_INIT_VLC_STATIC(10, 462);
1121 
1123 
1124  ff_aac_tableinit();
1125 
1126  INIT_VLC_STATIC(&vlc_scalefactors, 7,
1129  sizeof(ff_aac_scalefactor_bits[0]),
1130  sizeof(ff_aac_scalefactor_bits[0]),
1132  sizeof(ff_aac_scalefactor_code[0]),
1133  sizeof(ff_aac_scalefactor_code[0]),
1134  352);
1135 
1136  // window initialization
1139 #if !USE_FIXED
1142  AAC_RENAME(ff_sine_window_init)(AAC_RENAME(ff_sine_960), 960);
1143  AAC_RENAME(ff_sine_window_init)(AAC_RENAME(ff_sine_120), 120);
1144 #endif
1148 
1150 }
1151 
1153 
1155 {
1156  AACContext *ac = avctx->priv_data;
1157  int ret;
1158 
1159  ret = ff_thread_once(&aac_table_init, &aac_static_table_init);
1160  if (ret != 0)
1161  return AVERROR_UNKNOWN;
1162 
1163  ac->avctx = avctx;
1164  ac->oc[1].m4ac.sample_rate = avctx->sample_rate;
1165 
1166  aacdec_init(ac);
1167 #if USE_FIXED
1168  avctx->sample_fmt = AV_SAMPLE_FMT_S32P;
1169 #else
1170  avctx->sample_fmt = AV_SAMPLE_FMT_FLTP;
1171 #endif /* USE_FIXED */
1172 
1173  if (avctx->extradata_size > 0) {
1174  if ((ret = decode_audio_specific_config(ac, ac->avctx, &ac->oc[1].m4ac,
1175  avctx->extradata,
1176  avctx->extradata_size * 8LL,
1177  1)) < 0)
1178  return ret;
1179  } else {
1180  int sr, i;
1181  uint8_t layout_map[MAX_ELEM_ID*4][3];
1182  int layout_map_tags;
1183 
1184  sr = sample_rate_idx(avctx->sample_rate);
1185  ac->oc[1].m4ac.sampling_index = sr;
1186  ac->oc[1].m4ac.channels = avctx->channels;
1187  ac->oc[1].m4ac.sbr = -1;
1188  ac->oc[1].m4ac.ps = -1;
1189 
1190  for (i = 0; i < FF_ARRAY_ELEMS(ff_mpeg4audio_channels); i++)
1191  if (ff_mpeg4audio_channels[i] == avctx->channels)
1192  break;
1194  i = 0;
1195  }
1196  ac->oc[1].m4ac.chan_config = i;
1197 
1198  if (ac->oc[1].m4ac.chan_config) {
1199  int ret = set_default_channel_config(avctx, layout_map,
1200  &layout_map_tags, ac->oc[1].m4ac.chan_config);
1201  if (!ret)
1202  output_configure(ac, layout_map, layout_map_tags,
1203  OC_GLOBAL_HDR, 0);
1204  else if (avctx->err_recognition & AV_EF_EXPLODE)
1205  return AVERROR_INVALIDDATA;
1206  }
1207  }
1208 
1209  if (avctx->channels > MAX_CHANNELS) {
1210  av_log(avctx, AV_LOG_ERROR, "Too many channels\n");
1211  return AVERROR_INVALIDDATA;
1212  }
1213 
1214 #if USE_FIXED
1216 #else
1218 #endif /* USE_FIXED */
1219  if (!ac->fdsp) {
1220  return AVERROR(ENOMEM);
1221  }
1222 
1223  ac->random_state = 0x1f2e3d4c;
1224 
1225  AAC_RENAME_32(ff_mdct_init)(&ac->mdct, 11, 1, 1.0 / RANGE15(1024.0));
1226  AAC_RENAME_32(ff_mdct_init)(&ac->mdct_ld, 10, 1, 1.0 / RANGE15(512.0));
1227  AAC_RENAME_32(ff_mdct_init)(&ac->mdct_small, 8, 1, 1.0 / RANGE15(128.0));
1228  AAC_RENAME_32(ff_mdct_init)(&ac->mdct_ltp, 11, 0, RANGE15(-2.0));
1229 #if !USE_FIXED
1230  ret = ff_mdct15_init(&ac->mdct120, 1, 3, 1.0f/(16*1024*120*2));
1231  if (ret < 0)
1232  return ret;
1233  ret = ff_mdct15_init(&ac->mdct480, 1, 5, 1.0f/(16*1024*960));
1234  if (ret < 0)
1235  return ret;
1236  ret = ff_mdct15_init(&ac->mdct960, 1, 6, 1.0f/(16*1024*960*2));
1237  if (ret < 0)
1238  return ret;
1239 #endif
1240 
1241  return 0;
1242 }
1243 
1244 /**
1245  * Skip data_stream_element; reference: table 4.10.
1246  */
1248 {
1249  int byte_align = get_bits1(gb);
1250  int count = get_bits(gb, 8);
1251  if (count == 255)
1252  count += get_bits(gb, 8);
1253  if (byte_align)
1254  align_get_bits(gb);
1255 
1256  if (get_bits_left(gb) < 8 * count) {
1257  av_log(ac->avctx, AV_LOG_ERROR, "skip_data_stream_element: "overread_err);
1258  return AVERROR_INVALIDDATA;
1259  }
1260  skip_bits_long(gb, 8 * count);
1261  return 0;
1262 }
1263 
1265  GetBitContext *gb)
1266 {
1267  int sfb;
1268  if (get_bits1(gb)) {
1269  ics->predictor_reset_group = get_bits(gb, 5);
1270  if (ics->predictor_reset_group == 0 ||
1271  ics->predictor_reset_group > 30) {
1272  av_log(ac->avctx, AV_LOG_ERROR,
1273  "Invalid Predictor Reset Group.\n");
1274  return AVERROR_INVALIDDATA;
1275  }
1276  }
1277  for (sfb = 0; sfb < FFMIN(ics->max_sfb, ff_aac_pred_sfb_max[ac->oc[1].m4ac.sampling_index]); sfb++) {
1278  ics->prediction_used[sfb] = get_bits1(gb);
1279  }
1280  return 0;
1281 }
1282 
1283 /**
1284  * Decode Long Term Prediction data; reference: table 4.xx.
1285  */
1287  GetBitContext *gb, uint8_t max_sfb)
1288 {
1289  int sfb;
1290 
1291  ltp->lag = get_bits(gb, 11);
1292  ltp->coef = ltp_coef[get_bits(gb, 3)];
1293  for (sfb = 0; sfb < FFMIN(max_sfb, MAX_LTP_LONG_SFB); sfb++)
1294  ltp->used[sfb] = get_bits1(gb);
1295 }
1296 
1297 /**
1298  * Decode Individual Channel Stream info; reference: table 4.6.
1299  */
1301  GetBitContext *gb)
1302 {
1303  const MPEG4AudioConfig *const m4ac = &ac->oc[1].m4ac;
1304  const int aot = m4ac->object_type;
1305  const int sampling_index = m4ac->sampling_index;
1306  int ret_fail = AVERROR_INVALIDDATA;
1307 
1308  if (aot != AOT_ER_AAC_ELD) {
1309  if (get_bits1(gb)) {
1310  av_log(ac->avctx, AV_LOG_ERROR, "Reserved bit set.\n");
1312  return AVERROR_INVALIDDATA;
1313  }
1314  ics->window_sequence[1] = ics->window_sequence[0];
1315  ics->window_sequence[0] = get_bits(gb, 2);
1316  if (aot == AOT_ER_AAC_LD &&
1317  ics->window_sequence[0] != ONLY_LONG_SEQUENCE) {
1318  av_log(ac->avctx, AV_LOG_ERROR,
1319  "AAC LD is only defined for ONLY_LONG_SEQUENCE but "
1320  "window sequence %d found.\n", ics->window_sequence[0]);
1322  return AVERROR_INVALIDDATA;
1323  }
1324  ics->use_kb_window[1] = ics->use_kb_window[0];
1325  ics->use_kb_window[0] = get_bits1(gb);
1326  }
1327  ics->num_window_groups = 1;
1328  ics->group_len[0] = 1;
1329  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
1330  int i;
1331  ics->max_sfb = get_bits(gb, 4);
1332  for (i = 0; i < 7; i++) {
1333  if (get_bits1(gb)) {
1334  ics->group_len[ics->num_window_groups - 1]++;
1335  } else {
1336  ics->num_window_groups++;
1337  ics->group_len[ics->num_window_groups - 1] = 1;
1338  }
1339  }
1340  ics->num_windows = 8;
1341  if (m4ac->frame_length_short) {
1342  ics->swb_offset = ff_swb_offset_120[sampling_index];
1343  ics->num_swb = ff_aac_num_swb_120[sampling_index];
1344  } else {
1345  ics->swb_offset = ff_swb_offset_128[sampling_index];
1346  ics->num_swb = ff_aac_num_swb_128[sampling_index];
1347  }
1348  ics->tns_max_bands = ff_tns_max_bands_128[sampling_index];
1349  ics->predictor_present = 0;
1350  } else {
1351  ics->max_sfb = get_bits(gb, 6);
1352  ics->num_windows = 1;
1353  if (aot == AOT_ER_AAC_LD || aot == AOT_ER_AAC_ELD) {
1354  if (m4ac->frame_length_short) {
1355  ics->swb_offset = ff_swb_offset_480[sampling_index];
1356  ics->num_swb = ff_aac_num_swb_480[sampling_index];
1357  ics->tns_max_bands = ff_tns_max_bands_480[sampling_index];
1358  } else {
1359  ics->swb_offset = ff_swb_offset_512[sampling_index];
1360  ics->num_swb = ff_aac_num_swb_512[sampling_index];
1361  ics->tns_max_bands = ff_tns_max_bands_512[sampling_index];
1362  }
1363  if (!ics->num_swb || !ics->swb_offset) {
1364  ret_fail = AVERROR_BUG;
1365  goto fail;
1366  }
1367  } else {
1368  if (m4ac->frame_length_short) {
1369  ics->num_swb = ff_aac_num_swb_960[sampling_index];
1370  ics->swb_offset = ff_swb_offset_960[sampling_index];
1371  } else {
1372  ics->num_swb = ff_aac_num_swb_1024[sampling_index];
1373  ics->swb_offset = ff_swb_offset_1024[sampling_index];
1374  }
1375  ics->tns_max_bands = ff_tns_max_bands_1024[sampling_index];
1376  }
1377  if (aot != AOT_ER_AAC_ELD) {
1378  ics->predictor_present = get_bits1(gb);
1379  ics->predictor_reset_group = 0;
1380  }
1381  if (ics->predictor_present) {
1382  if (aot == AOT_AAC_MAIN) {
1383  if (decode_prediction(ac, ics, gb)) {
1384  goto fail;
1385  }
1386  } else if (aot == AOT_AAC_LC ||
1387  aot == AOT_ER_AAC_LC) {
1388  av_log(ac->avctx, AV_LOG_ERROR,
1389  "Prediction is not allowed in AAC-LC.\n");
1390  goto fail;
1391  } else {
1392  if (aot == AOT_ER_AAC_LD) {
1393  av_log(ac->avctx, AV_LOG_ERROR,
1394  "LTP in ER AAC LD not yet implemented.\n");
1395  ret_fail = AVERROR_PATCHWELCOME;
1396  goto fail;
1397  }
1398  if ((ics->ltp.present = get_bits(gb, 1)))
1399  decode_ltp(&ics->ltp, gb, ics->max_sfb);
1400  }
1401  }
1402  }
1403 
1404  if (ics->max_sfb > ics->num_swb) {
1405  av_log(ac->avctx, AV_LOG_ERROR,
1406  "Number of scalefactor bands in group (%d) "
1407  "exceeds limit (%d).\n",
1408  ics->max_sfb, ics->num_swb);
1409  goto fail;
1410  }
1411 
1412  return 0;
1413 fail:
1414  ics->max_sfb = 0;
1415  return ret_fail;
1416 }
1417 
1418 /**
1419  * Decode band types (section_data payload); reference: table 4.46.
1420  *
1421  * @param band_type array of the used band type
1422  * @param band_type_run_end array of the last scalefactor band of a band type run
1423  *
1424  * @return Returns error status. 0 - OK, !0 - error
1425  */
1426 static int decode_band_types(AACContext *ac, enum BandType band_type[120],
1427  int band_type_run_end[120], GetBitContext *gb,
1429 {
1430  int g, idx = 0;
1431  const int bits = (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) ? 3 : 5;
1432  for (g = 0; g < ics->num_window_groups; g++) {
1433  int k = 0;
1434  while (k < ics->max_sfb) {
1435  uint8_t sect_end = k;
1436  int sect_len_incr;
1437  int sect_band_type = get_bits(gb, 4);
1438  if (sect_band_type == 12) {
1439  av_log(ac->avctx, AV_LOG_ERROR, "invalid band type\n");
1440  return AVERROR_INVALIDDATA;
1441  }
1442  do {
1443  sect_len_incr = get_bits(gb, bits);
1444  sect_end += sect_len_incr;
1445  if (get_bits_left(gb) < 0) {
1446  av_log(ac->avctx, AV_LOG_ERROR, "decode_band_types: "overread_err);
1447  return AVERROR_INVALIDDATA;
1448  }
1449  if (sect_end > ics->max_sfb) {
1450  av_log(ac->avctx, AV_LOG_ERROR,
1451  "Number of bands (%d) exceeds limit (%d).\n",
1452  sect_end, ics->max_sfb);
1453  return AVERROR_INVALIDDATA;
1454  }
1455  } while (sect_len_incr == (1 << bits) - 1);
1456  for (; k < sect_end; k++) {
1457  band_type [idx] = sect_band_type;
1458  band_type_run_end[idx++] = sect_end;
1459  }
1460  }
1461  }
1462  return 0;
1463 }
1464 
1465 /**
1466  * Decode scalefactors; reference: table 4.47.
1467  *
1468  * @param global_gain first scalefactor value as scalefactors are differentially coded
1469  * @param band_type array of the used band type
1470  * @param band_type_run_end array of the last scalefactor band of a band type run
1471  * @param sf array of scalefactors or intensity stereo positions
1472  *
1473  * @return Returns error status. 0 - OK, !0 - error
1474  */
1476  unsigned int global_gain,
1478  enum BandType band_type[120],
1479  int band_type_run_end[120])
1480 {
1481  int g, i, idx = 0;
1482  int offset[3] = { global_gain, global_gain - NOISE_OFFSET, 0 };
1483  int clipped_offset;
1484  int noise_flag = 1;
1485  for (g = 0; g < ics->num_window_groups; g++) {
1486  for (i = 0; i < ics->max_sfb;) {
1487  int run_end = band_type_run_end[idx];
1488  if (band_type[idx] == ZERO_BT) {
1489  for (; i < run_end; i++, idx++)
1490  sf[idx] = FIXR(0.);
1491  } else if ((band_type[idx] == INTENSITY_BT) ||
1492  (band_type[idx] == INTENSITY_BT2)) {
1493  for (; i < run_end; i++, idx++) {
1494  offset[2] += get_vlc2(gb, vlc_scalefactors.table, 7, 3) - SCALE_DIFF_ZERO;
1495  clipped_offset = av_clip(offset[2], -155, 100);
1496  if (offset[2] != clipped_offset) {
1498  "If you heard an audible artifact, there may be a bug in the decoder. "
1499  "Clipped intensity stereo position (%d -> %d)",
1500  offset[2], clipped_offset);
1501  }
1502 #if USE_FIXED
1503  sf[idx] = 100 - clipped_offset;
1504 #else
1505  sf[idx] = ff_aac_pow2sf_tab[-clipped_offset + POW_SF2_ZERO];
1506 #endif /* USE_FIXED */
1507  }
1508  } else if (band_type[idx] == NOISE_BT) {
1509  for (; i < run_end; i++, idx++) {
1510  if (noise_flag-- > 0)
1511  offset[1] += get_bits(gb, NOISE_PRE_BITS) - NOISE_PRE;
1512  else
1513  offset[1] += get_vlc2(gb, vlc_scalefactors.table, 7, 3) - SCALE_DIFF_ZERO;
1514  clipped_offset = av_clip(offset[1], -100, 155);
1515  if (offset[1] != clipped_offset) {
1517  "If you heard an audible artifact, there may be a bug in the decoder. "
1518  "Clipped noise gain (%d -> %d)",
1519  offset[1], clipped_offset);
1520  }
1521 #if USE_FIXED
1522  sf[idx] = -(100 + clipped_offset);
1523 #else
1524  sf[idx] = -ff_aac_pow2sf_tab[clipped_offset + POW_SF2_ZERO];
1525 #endif /* USE_FIXED */
1526  }
1527  } else {
1528  for (; i < run_end; i++, idx++) {
1529  offset[0] += get_vlc2(gb, vlc_scalefactors.table, 7, 3) - SCALE_DIFF_ZERO;
1530  if (offset[0] > 255U) {
1531  av_log(ac->avctx, AV_LOG_ERROR,
1532  "Scalefactor (%d) out of range.\n", offset[0]);
1533  return AVERROR_INVALIDDATA;
1534  }
1535 #if USE_FIXED
1536  sf[idx] = -offset[0];
1537 #else
1538  sf[idx] = -ff_aac_pow2sf_tab[offset[0] - 100 + POW_SF2_ZERO];
1539 #endif /* USE_FIXED */
1540  }
1541  }
1542  }
1543  }
1544  return 0;
1545 }
1546 
1547 /**
1548  * Decode pulse data; reference: table 4.7.
1549  */
1550 static int decode_pulses(Pulse *pulse, GetBitContext *gb,
1551  const uint16_t *swb_offset, int num_swb)
1552 {
1553  int i, pulse_swb;
1554  pulse->num_pulse = get_bits(gb, 2) + 1;
1555  pulse_swb = get_bits(gb, 6);
1556  if (pulse_swb >= num_swb)
1557  return -1;
1558  pulse->pos[0] = swb_offset[pulse_swb];
1559  pulse->pos[0] += get_bits(gb, 5);
1560  if (pulse->pos[0] >= swb_offset[num_swb])
1561  return -1;
1562  pulse->amp[0] = get_bits(gb, 4);
1563  for (i = 1; i < pulse->num_pulse; i++) {
1564  pulse->pos[i] = get_bits(gb, 5) + pulse->pos[i - 1];
1565  if (pulse->pos[i] >= swb_offset[num_swb])
1566  return -1;
1567  pulse->amp[i] = get_bits(gb, 4);
1568  }
1569  return 0;
1570 }
1571 
1572 /**
1573  * Decode Temporal Noise Shaping data; reference: table 4.48.
1574  *
1575  * @return Returns error status. 0 - OK, !0 - error
1576  */
1578  GetBitContext *gb, const IndividualChannelStream *ics)
1579 {
1580  int w, filt, i, coef_len, coef_res, coef_compress;
1581  const int is8 = ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE;
1582  const int tns_max_order = is8 ? 7 : ac->oc[1].m4ac.object_type == AOT_AAC_MAIN ? 20 : 12;
1583  for (w = 0; w < ics->num_windows; w++) {
1584  if ((tns->n_filt[w] = get_bits(gb, 2 - is8))) {
1585  coef_res = get_bits1(gb);
1586 
1587  for (filt = 0; filt < tns->n_filt[w]; filt++) {
1588  int tmp2_idx;
1589  tns->length[w][filt] = get_bits(gb, 6 - 2 * is8);
1590 
1591  if ((tns->order[w][filt] = get_bits(gb, 5 - 2 * is8)) > tns_max_order) {
1592  av_log(ac->avctx, AV_LOG_ERROR,
1593  "TNS filter order %d is greater than maximum %d.\n",
1594  tns->order[w][filt], tns_max_order);
1595  tns->order[w][filt] = 0;
1596  return AVERROR_INVALIDDATA;
1597  }
1598  if (tns->order[w][filt]) {
1599  tns->direction[w][filt] = get_bits1(gb);
1600  coef_compress = get_bits1(gb);
1601  coef_len = coef_res + 3 - coef_compress;
1602  tmp2_idx = 2 * coef_compress + coef_res;
1603 
1604  for (i = 0; i < tns->order[w][filt]; i++)
1605  tns->coef[w][filt][i] = tns_tmp2_map[tmp2_idx][get_bits(gb, coef_len)];
1606  }
1607  }
1608  }
1609  }
1610  return 0;
1611 }
1612 
1613 /**
1614  * Decode Mid/Side data; reference: table 4.54.
1615  *
1616  * @param ms_present Indicates mid/side stereo presence. [0] mask is all 0s;
1617  * [1] mask is decoded from bitstream; [2] mask is all 1s;
1618  * [3] reserved for scalable AAC
1619  */
1621  int ms_present)
1622 {
1623  int idx;
1624  int max_idx = cpe->ch[0].ics.num_window_groups * cpe->ch[0].ics.max_sfb;
1625  if (ms_present == 1) {
1626  for (idx = 0; idx < max_idx; idx++)
1627  cpe->ms_mask[idx] = get_bits1(gb);
1628  } else if (ms_present == 2) {
1629  memset(cpe->ms_mask, 1, max_idx * sizeof(cpe->ms_mask[0]));
1630  }
1631 }
1632 
1633 /**
1634  * Decode spectral data; reference: table 4.50.
1635  * Dequantize and scale spectral data; reference: 4.6.3.3.
1636  *
1637  * @param coef array of dequantized, scaled spectral data
1638  * @param sf array of scalefactors or intensity stereo positions
1639  * @param pulse_present set if pulses are present
1640  * @param pulse pointer to pulse data struct
1641  * @param band_type array of the used band type
1642  *
1643  * @return Returns error status. 0 - OK, !0 - error
1644  */
1646  GetBitContext *gb, const INTFLOAT sf[120],
1647  int pulse_present, const Pulse *pulse,
1648  const IndividualChannelStream *ics,
1649  enum BandType band_type[120])
1650 {
1651  int i, k, g, idx = 0;
1652  const int c = 1024 / ics->num_windows;
1653  const uint16_t *offsets = ics->swb_offset;
1654  INTFLOAT *coef_base = coef;
1655 
1656  for (g = 0; g < ics->num_windows; g++)
1657  memset(coef + g * 128 + offsets[ics->max_sfb], 0,
1658  sizeof(INTFLOAT) * (c - offsets[ics->max_sfb]));
1659 
1660  for (g = 0; g < ics->num_window_groups; g++) {
1661  unsigned g_len = ics->group_len[g];
1662 
1663  for (i = 0; i < ics->max_sfb; i++, idx++) {
1664  const unsigned cbt_m1 = band_type[idx] - 1;
1665  INTFLOAT *cfo = coef + offsets[i];
1666  int off_len = offsets[i + 1] - offsets[i];
1667  int group;
1668 
1669  if (cbt_m1 >= INTENSITY_BT2 - 1) {
1670  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1671  memset(cfo, 0, off_len * sizeof(*cfo));
1672  }
1673  } else if (cbt_m1 == NOISE_BT - 1) {
1674  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1675 #if !USE_FIXED
1676  float scale;
1677 #endif /* !USE_FIXED */
1678  INTFLOAT band_energy;
1679 
1680  for (k = 0; k < off_len; k++) {
1682 #if USE_FIXED
1683  cfo[k] = ac->random_state >> 3;
1684 #else
1685  cfo[k] = ac->random_state;
1686 #endif /* USE_FIXED */
1687  }
1688 
1689 #if USE_FIXED
1690  band_energy = ac->fdsp->scalarproduct_fixed(cfo, cfo, off_len);
1691  band_energy = fixed_sqrt(band_energy, 31);
1692  noise_scale(cfo, sf[idx], band_energy, off_len);
1693 #else
1694  band_energy = ac->fdsp->scalarproduct_float(cfo, cfo, off_len);
1695  scale = sf[idx] / sqrtf(band_energy);
1696  ac->fdsp->vector_fmul_scalar(cfo, cfo, scale, off_len);
1697 #endif /* USE_FIXED */
1698  }
1699  } else {
1700 #if !USE_FIXED
1701  const float *vq = ff_aac_codebook_vector_vals[cbt_m1];
1702 #endif /* !USE_FIXED */
1703  const uint16_t *cb_vector_idx = ff_aac_codebook_vector_idx[cbt_m1];
1704  VLC_TYPE (*vlc_tab)[2] = vlc_spectral[cbt_m1].table;
1705  OPEN_READER(re, gb);
1706 
1707  switch (cbt_m1 >> 1) {
1708  case 0:
1709  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1710  INTFLOAT *cf = cfo;
1711  int len = off_len;
1712 
1713  do {
1714  int code;
1715  unsigned cb_idx;
1716 
1717  UPDATE_CACHE(re, gb);
1718  GET_VLC(code, re, gb, vlc_tab, 8, 2);
1719  cb_idx = cb_vector_idx[code];
1720 #if USE_FIXED
1721  cf = DEC_SQUAD(cf, cb_idx);
1722 #else
1723  cf = VMUL4(cf, vq, cb_idx, sf + idx);
1724 #endif /* USE_FIXED */
1725  } while (len -= 4);
1726  }
1727  break;
1728 
1729  case 1:
1730  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1731  INTFLOAT *cf = cfo;
1732  int len = off_len;
1733 
1734  do {
1735  int code;
1736  unsigned nnz;
1737  unsigned cb_idx;
1738  uint32_t bits;
1739 
1740  UPDATE_CACHE(re, gb);
1741  GET_VLC(code, re, gb, vlc_tab, 8, 2);
1742  cb_idx = cb_vector_idx[code];
1743  nnz = cb_idx >> 8 & 15;
1744  bits = nnz ? GET_CACHE(re, gb) : 0;
1745  LAST_SKIP_BITS(re, gb, nnz);
1746 #if USE_FIXED
1747  cf = DEC_UQUAD(cf, cb_idx, bits);
1748 #else
1749  cf = VMUL4S(cf, vq, cb_idx, bits, sf + idx);
1750 #endif /* USE_FIXED */
1751  } while (len -= 4);
1752  }
1753  break;
1754 
1755  case 2:
1756  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1757  INTFLOAT *cf = cfo;
1758  int len = off_len;
1759 
1760  do {
1761  int code;
1762  unsigned cb_idx;
1763 
1764  UPDATE_CACHE(re, gb);
1765  GET_VLC(code, re, gb, vlc_tab, 8, 2);
1766  cb_idx = cb_vector_idx[code];
1767 #if USE_FIXED
1768  cf = DEC_SPAIR(cf, cb_idx);
1769 #else
1770  cf = VMUL2(cf, vq, cb_idx, sf + idx);
1771 #endif /* USE_FIXED */
1772  } while (len -= 2);
1773  }
1774  break;
1775 
1776  case 3:
1777  case 4:
1778  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1779  INTFLOAT *cf = cfo;
1780  int len = off_len;
1781 
1782  do {
1783  int code;
1784  unsigned nnz;
1785  unsigned cb_idx;
1786  unsigned sign;
1787 
1788  UPDATE_CACHE(re, gb);
1789  GET_VLC(code, re, gb, vlc_tab, 8, 2);
1790  cb_idx = cb_vector_idx[code];
1791  nnz = cb_idx >> 8 & 15;
1792  sign = nnz ? SHOW_UBITS(re, gb, nnz) << (cb_idx >> 12) : 0;
1793  LAST_SKIP_BITS(re, gb, nnz);
1794 #if USE_FIXED
1795  cf = DEC_UPAIR(cf, cb_idx, sign);
1796 #else
1797  cf = VMUL2S(cf, vq, cb_idx, sign, sf + idx);
1798 #endif /* USE_FIXED */
1799  } while (len -= 2);
1800  }
1801  break;
1802 
1803  default:
1804  for (group = 0; group < (AAC_SIGNE)g_len; group++, cfo+=128) {
1805 #if USE_FIXED
1806  int *icf = cfo;
1807  int v;
1808 #else
1809  float *cf = cfo;
1810  uint32_t *icf = (uint32_t *) cf;
1811 #endif /* USE_FIXED */
1812  int len = off_len;
1813 
1814  do {
1815  int code;
1816  unsigned nzt, nnz;
1817  unsigned cb_idx;
1818  uint32_t bits;
1819  int j;
1820 
1821  UPDATE_CACHE(re, gb);
1822  GET_VLC(code, re, gb, vlc_tab, 8, 2);
1823 
1824  if (!code) {
1825  *icf++ = 0;
1826  *icf++ = 0;
1827  continue;
1828  }
1829 
1830  cb_idx = cb_vector_idx[code];
1831  nnz = cb_idx >> 12;
1832  nzt = cb_idx >> 8;
1833  bits = SHOW_UBITS(re, gb, nnz) << (32-nnz);
1834  LAST_SKIP_BITS(re, gb, nnz);
1835 
1836  for (j = 0; j < 2; j++) {
1837  if (nzt & 1<<j) {
1838  uint32_t b;
1839  int n;
1840  /* The total length of escape_sequence must be < 22 bits according
1841  to the specification (i.e. max is 111111110xxxxxxxxxxxx). */
1842  UPDATE_CACHE(re, gb);
1843  b = GET_CACHE(re, gb);
1844  b = 31 - av_log2(~b);
1845 
1846  if (b > 8) {
1847  av_log(ac->avctx, AV_LOG_ERROR, "error in spectral data, ESC overflow\n");
1848  return AVERROR_INVALIDDATA;
1849  }
1850 
1851  SKIP_BITS(re, gb, b + 1);
1852  b += 4;
1853  n = (1 << b) + SHOW_UBITS(re, gb, b);
1854  LAST_SKIP_BITS(re, gb, b);
1855 #if USE_FIXED
1856  v = n;
1857  if (bits & 1U<<31)
1858  v = -v;
1859  *icf++ = v;
1860 #else
1861  *icf++ = ff_cbrt_tab[n] | (bits & 1U<<31);
1862 #endif /* USE_FIXED */
1863  bits <<= 1;
1864  } else {
1865 #if USE_FIXED
1866  v = cb_idx & 15;
1867  if (bits & 1U<<31)
1868  v = -v;
1869  *icf++ = v;
1870 #else
1871  unsigned v = ((const uint32_t*)vq)[cb_idx & 15];
1872  *icf++ = (bits & 1U<<31) | v;
1873 #endif /* USE_FIXED */
1874  bits <<= !!v;
1875  }
1876  cb_idx >>= 4;
1877  }
1878  } while (len -= 2);
1879 #if !USE_FIXED
1880  ac->fdsp->vector_fmul_scalar(cfo, cfo, sf[idx], off_len);
1881 #endif /* !USE_FIXED */
1882  }
1883  }
1884 
1885  CLOSE_READER(re, gb);
1886  }
1887  }
1888  coef += g_len << 7;
1889  }
1890 
1891  if (pulse_present) {
1892  idx = 0;
1893  for (i = 0; i < pulse->num_pulse; i++) {
1894  INTFLOAT co = coef_base[ pulse->pos[i] ];
1895  while (offsets[idx + 1] <= pulse->pos[i])
1896  idx++;
1897  if (band_type[idx] != NOISE_BT && sf[idx]) {
1898  INTFLOAT ico = -pulse->amp[i];
1899 #if USE_FIXED
1900  if (co) {
1901  ico = co + (co > 0 ? -ico : ico);
1902  }
1903  coef_base[ pulse->pos[i] ] = ico;
1904 #else
1905  if (co) {
1906  co /= sf[idx];
1907  ico = co / sqrtf(sqrtf(fabsf(co))) + (co > 0 ? -ico : ico);
1908  }
1909  coef_base[ pulse->pos[i] ] = cbrtf(fabsf(ico)) * ico * sf[idx];
1910 #endif /* USE_FIXED */
1911  }
1912  }
1913  }
1914 #if USE_FIXED
1915  coef = coef_base;
1916  idx = 0;
1917  for (g = 0; g < ics->num_window_groups; g++) {
1918  unsigned g_len = ics->group_len[g];
1919 
1920  for (i = 0; i < ics->max_sfb; i++, idx++) {
1921  const unsigned cbt_m1 = band_type[idx] - 1;
1922  int *cfo = coef + offsets[i];
1923  int off_len = offsets[i + 1] - offsets[i];
1924  int group;
1925 
1926  if (cbt_m1 < NOISE_BT - 1) {
1927  for (group = 0; group < (int)g_len; group++, cfo+=128) {
1928  ac->vector_pow43(cfo, off_len);
1929  ac->subband_scale(cfo, cfo, sf[idx], 34, off_len);
1930  }
1931  }
1932  }
1933  coef += g_len << 7;
1934  }
1935 #endif /* USE_FIXED */
1936  return 0;
1937 }
1938 
1939 /**
1940  * Apply AAC-Main style frequency domain prediction.
1941  */
1943 {
1944  int sfb, k;
1945 
1946  if (!sce->ics.predictor_initialized) {
1948  sce->ics.predictor_initialized = 1;
1949  }
1950 
1951  if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
1952  for (sfb = 0;
1953  sfb < ff_aac_pred_sfb_max[ac->oc[1].m4ac.sampling_index];
1954  sfb++) {
1955  for (k = sce->ics.swb_offset[sfb];
1956  k < sce->ics.swb_offset[sfb + 1];
1957  k++) {
1958  predict(&sce->predictor_state[k], &sce->coeffs[k],
1959  sce->ics.predictor_present &&
1960  sce->ics.prediction_used[sfb]);
1961  }
1962  }
1963  if (sce->ics.predictor_reset_group)
1965  sce->ics.predictor_reset_group);
1966  } else
1968 }
1969 
1970 /**
1971  * Decode an individual_channel_stream payload; reference: table 4.44.
1972  *
1973  * @param common_window Channels have independent [0], or shared [1], Individual Channel Stream information.
1974  * @param scale_flag scalable [1] or non-scalable [0] AAC (Unused until scalable AAC is implemented.)
1975  *
1976  * @return Returns error status. 0 - OK, !0 - error
1977  */
1979  GetBitContext *gb, int common_window, int scale_flag)
1980 {
1981  Pulse pulse;
1982  TemporalNoiseShaping *tns = &sce->tns;
1983  IndividualChannelStream *ics = &sce->ics;
1984  INTFLOAT *out = sce->coeffs;
1985  int global_gain, eld_syntax, er_syntax, pulse_present = 0;
1986  int ret;
1987 
1988  eld_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
1989  er_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_LC ||
1990  ac->oc[1].m4ac.object_type == AOT_ER_AAC_LTP ||
1991  ac->oc[1].m4ac.object_type == AOT_ER_AAC_LD ||
1992  ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
1993 
1994  /* This assignment is to silence a GCC warning about the variable being used
1995  * uninitialized when in fact it always is.
1996  */
1997  pulse.num_pulse = 0;
1998 
1999  global_gain = get_bits(gb, 8);
2000 
2001  if (!common_window && !scale_flag) {
2002  ret = decode_ics_info(ac, ics, gb);
2003  if (ret < 0)
2004  goto fail;
2005  }
2006 
2007  if ((ret = decode_band_types(ac, sce->band_type,
2008  sce->band_type_run_end, gb, ics)) < 0)
2009  goto fail;
2010  if ((ret = decode_scalefactors(ac, sce->sf, gb, global_gain, ics,
2011  sce->band_type, sce->band_type_run_end)) < 0)
2012  goto fail;
2013 
2014  pulse_present = 0;
2015  if (!scale_flag) {
2016  if (!eld_syntax && (pulse_present = get_bits1(gb))) {
2017  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2018  av_log(ac->avctx, AV_LOG_ERROR,
2019  "Pulse tool not allowed in eight short sequence.\n");
2020  ret = AVERROR_INVALIDDATA;
2021  goto fail;
2022  }
2023  if (decode_pulses(&pulse, gb, ics->swb_offset, ics->num_swb)) {
2024  av_log(ac->avctx, AV_LOG_ERROR,
2025  "Pulse data corrupt or invalid.\n");
2026  ret = AVERROR_INVALIDDATA;
2027  goto fail;
2028  }
2029  }
2030  tns->present = get_bits1(gb);
2031  if (tns->present && !er_syntax) {
2032  ret = decode_tns(ac, tns, gb, ics);
2033  if (ret < 0)
2034  goto fail;
2035  }
2036  if (!eld_syntax && get_bits1(gb)) {
2037  avpriv_request_sample(ac->avctx, "SSR");
2038  ret = AVERROR_PATCHWELCOME;
2039  goto fail;
2040  }
2041  // I see no textual basis in the spec for this occurring after SSR gain
2042  // control, but this is what both reference and real implmentations do
2043  if (tns->present && er_syntax) {
2044  ret = decode_tns(ac, tns, gb, ics);
2045  if (ret < 0)
2046  goto fail;
2047  }
2048  }
2049 
2050  ret = decode_spectrum_and_dequant(ac, out, gb, sce->sf, pulse_present,
2051  &pulse, ics, sce->band_type);
2052  if (ret < 0)
2053  goto fail;
2054 
2055  if (ac->oc[1].m4ac.object_type == AOT_AAC_MAIN && !common_window)
2056  apply_prediction(ac, sce);
2057 
2058  return 0;
2059 fail:
2060  tns->present = 0;
2061  return ret;
2062 }
2063 
2064 /**
2065  * Mid/Side stereo decoding; reference: 4.6.8.1.3.
2066  */
2068 {
2069  const IndividualChannelStream *ics = &cpe->ch[0].ics;
2070  INTFLOAT *ch0 = cpe->ch[0].coeffs;
2071  INTFLOAT *ch1 = cpe->ch[1].coeffs;
2072  int g, i, group, idx = 0;
2073  const uint16_t *offsets = ics->swb_offset;
2074  for (g = 0; g < ics->num_window_groups; g++) {
2075  for (i = 0; i < ics->max_sfb; i++, idx++) {
2076  if (cpe->ms_mask[idx] &&
2077  cpe->ch[0].band_type[idx] < NOISE_BT &&
2078  cpe->ch[1].band_type[idx] < NOISE_BT) {
2079 #if USE_FIXED
2080  for (group = 0; group < ics->group_len[g]; group++) {
2081  ac->fdsp->butterflies_fixed(ch0 + group * 128 + offsets[i],
2082  ch1 + group * 128 + offsets[i],
2083  offsets[i+1] - offsets[i]);
2084 #else
2085  for (group = 0; group < ics->group_len[g]; group++) {
2086  ac->fdsp->butterflies_float(ch0 + group * 128 + offsets[i],
2087  ch1 + group * 128 + offsets[i],
2088  offsets[i+1] - offsets[i]);
2089 #endif /* USE_FIXED */
2090  }
2091  }
2092  }
2093  ch0 += ics->group_len[g] * 128;
2094  ch1 += ics->group_len[g] * 128;
2095  }
2096 }
2097 
2098 /**
2099  * intensity stereo decoding; reference: 4.6.8.2.3
2100  *
2101  * @param ms_present Indicates mid/side stereo presence. [0] mask is all 0s;
2102  * [1] mask is decoded from bitstream; [2] mask is all 1s;
2103  * [3] reserved for scalable AAC
2104  */
2106  ChannelElement *cpe, int ms_present)
2107 {
2108  const IndividualChannelStream *ics = &cpe->ch[1].ics;
2109  SingleChannelElement *sce1 = &cpe->ch[1];
2110  INTFLOAT *coef0 = cpe->ch[0].coeffs, *coef1 = cpe->ch[1].coeffs;
2111  const uint16_t *offsets = ics->swb_offset;
2112  int g, group, i, idx = 0;
2113  int c;
2114  INTFLOAT scale;
2115  for (g = 0; g < ics->num_window_groups; g++) {
2116  for (i = 0; i < ics->max_sfb;) {
2117  if (sce1->band_type[idx] == INTENSITY_BT ||
2118  sce1->band_type[idx] == INTENSITY_BT2) {
2119  const int bt_run_end = sce1->band_type_run_end[idx];
2120  for (; i < bt_run_end; i++, idx++) {
2121  c = -1 + 2 * (sce1->band_type[idx] - 14);
2122  if (ms_present)
2123  c *= 1 - 2 * cpe->ms_mask[idx];
2124  scale = c * sce1->sf[idx];
2125  for (group = 0; group < ics->group_len[g]; group++)
2126 #if USE_FIXED
2127  ac->subband_scale(coef1 + group * 128 + offsets[i],
2128  coef0 + group * 128 + offsets[i],
2129  scale,
2130  23,
2131  offsets[i + 1] - offsets[i]);
2132 #else
2133  ac->fdsp->vector_fmul_scalar(coef1 + group * 128 + offsets[i],
2134  coef0 + group * 128 + offsets[i],
2135  scale,
2136  offsets[i + 1] - offsets[i]);
2137 #endif /* USE_FIXED */
2138  }
2139  } else {
2140  int bt_run_end = sce1->band_type_run_end[idx];
2141  idx += bt_run_end - i;
2142  i = bt_run_end;
2143  }
2144  }
2145  coef0 += ics->group_len[g] * 128;
2146  coef1 += ics->group_len[g] * 128;
2147  }
2148 }
2149 
2150 /**
2151  * Decode a channel_pair_element; reference: table 4.4.
2152  *
2153  * @return Returns error status. 0 - OK, !0 - error
2154  */
2156 {
2157  int i, ret, common_window, ms_present = 0;
2158  int eld_syntax = ac->oc[1].m4ac.object_type == AOT_ER_AAC_ELD;
2159 
2160  common_window = eld_syntax || get_bits1(gb);
2161  if (common_window) {
2162  if (decode_ics_info(ac, &cpe->ch[0].ics, gb))
2163  return AVERROR_INVALIDDATA;
2164  i = cpe->ch[1].ics.use_kb_window[0];
2165  cpe->ch[1].ics = cpe->ch[0].ics;
2166  cpe->ch[1].ics.use_kb_window[1] = i;
2167  if (cpe->ch[1].ics.predictor_present &&
2168  (ac->oc[1].m4ac.object_type != AOT_AAC_MAIN))
2169  if ((cpe->ch[1].ics.ltp.present = get_bits(gb, 1)))
2170  decode_ltp(&cpe->ch[1].ics.ltp, gb, cpe->ch[1].ics.max_sfb);
2171  ms_present = get_bits(gb, 2);
2172  if (ms_present == 3) {
2173  av_log(ac->avctx, AV_LOG_ERROR, "ms_present = 3 is reserved.\n");
2174  return AVERROR_INVALIDDATA;
2175  } else if (ms_present)
2176  decode_mid_side_stereo(cpe, gb, ms_present);
2177  }
2178  if ((ret = decode_ics(ac, &cpe->ch[0], gb, common_window, 0)))
2179  return ret;
2180  if ((ret = decode_ics(ac, &cpe->ch[1], gb, common_window, 0)))
2181  return ret;
2182 
2183  if (common_window) {
2184  if (ms_present)
2185  apply_mid_side_stereo(ac, cpe);
2186  if (ac->oc[1].m4ac.object_type == AOT_AAC_MAIN) {
2187  apply_prediction(ac, &cpe->ch[0]);
2188  apply_prediction(ac, &cpe->ch[1]);
2189  }
2190  }
2191 
2192  apply_intensity_stereo(ac, cpe, ms_present);
2193  return 0;
2194 }
2195 
2196 static const float cce_scale[] = {
2197  1.09050773266525765921, //2^(1/8)
2198  1.18920711500272106672, //2^(1/4)
2199  M_SQRT2,
2200  2,
2201 };
2202 
2203 /**
2204  * Decode coupling_channel_element; reference: table 4.8.
2205  *
2206  * @return Returns error status. 0 - OK, !0 - error
2207  */
2209 {
2210  int num_gain = 0;
2211  int c, g, sfb, ret;
2212  int sign;
2213  INTFLOAT scale;
2214  SingleChannelElement *sce = &che->ch[0];
2215  ChannelCoupling *coup = &che->coup;
2216 
2217  coup->coupling_point = 2 * get_bits1(gb);
2218  coup->num_coupled = get_bits(gb, 3);
2219  for (c = 0; c <= coup->num_coupled; c++) {
2220  num_gain++;
2221  coup->type[c] = get_bits1(gb) ? TYPE_CPE : TYPE_SCE;
2222  coup->id_select[c] = get_bits(gb, 4);
2223  if (coup->type[c] == TYPE_CPE) {
2224  coup->ch_select[c] = get_bits(gb, 2);
2225  if (coup->ch_select[c] == 3)
2226  num_gain++;
2227  } else
2228  coup->ch_select[c] = 2;
2229  }
2230  coup->coupling_point += get_bits1(gb) || (coup->coupling_point >> 1);
2231 
2232  sign = get_bits(gb, 1);
2233 #if USE_FIXED
2234  scale = get_bits(gb, 2);
2235 #else
2236  scale = cce_scale[get_bits(gb, 2)];
2237 #endif
2238 
2239  if ((ret = decode_ics(ac, sce, gb, 0, 0)))
2240  return ret;
2241 
2242  for (c = 0; c < num_gain; c++) {
2243  int idx = 0;
2244  int cge = 1;
2245  int gain = 0;
2246  INTFLOAT gain_cache = FIXR10(1.);
2247  if (c) {
2248  cge = coup->coupling_point == AFTER_IMDCT ? 1 : get_bits1(gb);
2249  gain = cge ? get_vlc2(gb, vlc_scalefactors.table, 7, 3) - 60: 0;
2250  gain_cache = GET_GAIN(scale, gain);
2251 #if USE_FIXED
2252  if ((abs(gain_cache)-1024) >> 3 > 30)
2253  return AVERROR(ERANGE);
2254 #endif
2255  }
2256  if (coup->coupling_point == AFTER_IMDCT) {
2257  coup->gain[c][0] = gain_cache;
2258  } else {
2259  for (g = 0; g < sce->ics.num_window_groups; g++) {
2260  for (sfb = 0; sfb < sce->ics.max_sfb; sfb++, idx++) {
2261  if (sce->band_type[idx] != ZERO_BT) {
2262  if (!cge) {
2263  int t = get_vlc2(gb, vlc_scalefactors.table, 7, 3) - 60;
2264  if (t) {
2265  int s = 1;
2266  t = gain += t;
2267  if (sign) {
2268  s -= 2 * (t & 0x1);
2269  t >>= 1;
2270  }
2271  gain_cache = GET_GAIN(scale, t) * s;
2272 #if USE_FIXED
2273  if ((abs(gain_cache)-1024) >> 3 > 30)
2274  return AVERROR(ERANGE);
2275 #endif
2276  }
2277  }
2278  coup->gain[c][idx] = gain_cache;
2279  }
2280  }
2281  }
2282  }
2283  }
2284  return 0;
2285 }
2286 
2287 /**
2288  * Parse whether channels are to be excluded from Dynamic Range Compression; reference: table 4.53.
2289  *
2290  * @return Returns number of bytes consumed.
2291  */
2293  GetBitContext *gb)
2294 {
2295  int i;
2296  int num_excl_chan = 0;
2297 
2298  do {
2299  for (i = 0; i < 7; i++)
2300  che_drc->exclude_mask[num_excl_chan++] = get_bits1(gb);
2301  } while (num_excl_chan < MAX_CHANNELS - 7 && get_bits1(gb));
2302 
2303  return num_excl_chan / 7;
2304 }
2305 
2306 /**
2307  * Decode dynamic range information; reference: table 4.52.
2308  *
2309  * @return Returns number of bytes consumed.
2310  */
2312  GetBitContext *gb)
2313 {
2314  int n = 1;
2315  int drc_num_bands = 1;
2316  int i;
2317 
2318  /* pce_tag_present? */
2319  if (get_bits1(gb)) {
2320  che_drc->pce_instance_tag = get_bits(gb, 4);
2321  skip_bits(gb, 4); // tag_reserved_bits
2322  n++;
2323  }
2324 
2325  /* excluded_chns_present? */
2326  if (get_bits1(gb)) {
2327  n += decode_drc_channel_exclusions(che_drc, gb);
2328  }
2329 
2330  /* drc_bands_present? */
2331  if (get_bits1(gb)) {
2332  che_drc->band_incr = get_bits(gb, 4);
2333  che_drc->interpolation_scheme = get_bits(gb, 4);
2334  n++;
2335  drc_num_bands += che_drc->band_incr;
2336  for (i = 0; i < drc_num_bands; i++) {
2337  che_drc->band_top[i] = get_bits(gb, 8);
2338  n++;
2339  }
2340  }
2341 
2342  /* prog_ref_level_present? */
2343  if (get_bits1(gb)) {
2344  che_drc->prog_ref_level = get_bits(gb, 7);
2345  skip_bits1(gb); // prog_ref_level_reserved_bits
2346  n++;
2347  }
2348 
2349  for (i = 0; i < drc_num_bands; i++) {
2350  che_drc->dyn_rng_sgn[i] = get_bits1(gb);
2351  che_drc->dyn_rng_ctl[i] = get_bits(gb, 7);
2352  n++;
2353  }
2354 
2355  return n;
2356 }
2357 
2358 static int decode_fill(AACContext *ac, GetBitContext *gb, int len) {
2359  uint8_t buf[256];
2360  int i, major, minor;
2361 
2362  if (len < 13+7*8)
2363  goto unknown;
2364 
2365  get_bits(gb, 13); len -= 13;
2366 
2367  for(i=0; i+1<sizeof(buf) && len>=8; i++, len-=8)
2368  buf[i] = get_bits(gb, 8);
2369 
2370  buf[i] = 0;
2371  if (ac->avctx->debug & FF_DEBUG_PICT_INFO)
2372  av_log(ac->avctx, AV_LOG_DEBUG, "FILL:%s\n", buf);
2373 
2374  if (sscanf(buf, "libfaac %d.%d", &major, &minor) == 2){
2375  ac->avctx->internal->skip_samples = 1024;
2376  }
2377 
2378 unknown:
2379  skip_bits_long(gb, len);
2380 
2381  return 0;
2382 }
2383 
2384 /**
2385  * Decode extension data (incomplete); reference: table 4.51.
2386  *
2387  * @param cnt length of TYPE_FIL syntactic element in bytes
2388  *
2389  * @return Returns number of bytes consumed
2390  */
2392  ChannelElement *che, enum RawDataBlockType elem_type)
2393 {
2394  int crc_flag = 0;
2395  int res = cnt;
2396  int type = get_bits(gb, 4);
2397 
2398  if (ac->avctx->debug & FF_DEBUG_STARTCODE)
2399  av_log(ac->avctx, AV_LOG_DEBUG, "extension type: %d len:%d\n", type, cnt);
2400 
2401  switch (type) { // extension type
2402  case EXT_SBR_DATA_CRC:
2403  crc_flag++;
2404  case EXT_SBR_DATA:
2405  if (!che) {
2406  av_log(ac->avctx, AV_LOG_ERROR, "SBR was found before the first channel element.\n");
2407  return res;
2408  } else if (ac->oc[1].m4ac.frame_length_short) {
2409  if (!ac->warned_960_sbr)
2411  "SBR with 960 frame length");
2412  ac->warned_960_sbr = 1;
2413  skip_bits_long(gb, 8 * cnt - 4);
2414  return res;
2415  } else if (!ac->oc[1].m4ac.sbr) {
2416  av_log(ac->avctx, AV_LOG_ERROR, "SBR signaled to be not-present but was found in the bitstream.\n");
2417  skip_bits_long(gb, 8 * cnt - 4);
2418  return res;
2419  } else if (ac->oc[1].m4ac.sbr == -1 && ac->oc[1].status == OC_LOCKED) {
2420  av_log(ac->avctx, AV_LOG_ERROR, "Implicit SBR was found with a first occurrence after the first frame.\n");
2421  skip_bits_long(gb, 8 * cnt - 4);
2422  return res;
2423  } else if (ac->oc[1].m4ac.ps == -1 && ac->oc[1].status < OC_LOCKED && ac->avctx->channels == 1) {
2424  ac->oc[1].m4ac.sbr = 1;
2425  ac->oc[1].m4ac.ps = 1;
2427  output_configure(ac, ac->oc[1].layout_map, ac->oc[1].layout_map_tags,
2428  ac->oc[1].status, 1);
2429  } else {
2430  ac->oc[1].m4ac.sbr = 1;
2432  }
2433  res = AAC_RENAME(ff_decode_sbr_extension)(ac, &che->sbr, gb, crc_flag, cnt, elem_type);
2434  break;
2435  case EXT_DYNAMIC_RANGE:
2436  res = decode_dynamic_range(&ac->che_drc, gb);
2437  break;
2438  case EXT_FILL:
2439  decode_fill(ac, gb, 8 * cnt - 4);
2440  break;
2441  case EXT_FILL_DATA:
2442  case EXT_DATA_ELEMENT:
2443  default:
2444  skip_bits_long(gb, 8 * cnt - 4);
2445  break;
2446  };
2447  return res;
2448 }
2449 
2450 /**
2451  * Decode Temporal Noise Shaping filter coefficients and apply all-pole filters; reference: 4.6.9.3.
2452  *
2453  * @param decode 1 if tool is used normally, 0 if tool is used in LTP.
2454  * @param coef spectral coefficients
2455  */
2456 static void apply_tns(INTFLOAT coef_param[1024], TemporalNoiseShaping *tns,
2457  IndividualChannelStream *ics, int decode)
2458 {
2459  const int mmm = FFMIN(ics->tns_max_bands, ics->max_sfb);
2460  int w, filt, m, i;
2461  int bottom, top, order, start, end, size, inc;
2462  INTFLOAT lpc[TNS_MAX_ORDER];
2464  UINTFLOAT *coef = coef_param;
2465 
2466  for (w = 0; w < ics->num_windows; w++) {
2467  bottom = ics->num_swb;
2468  for (filt = 0; filt < tns->n_filt[w]; filt++) {
2469  top = bottom;
2470  bottom = FFMAX(0, top - tns->length[w][filt]);
2471  order = tns->order[w][filt];
2472  if (order == 0)
2473  continue;
2474 
2475  // tns_decode_coef
2476  AAC_RENAME(compute_lpc_coefs)(tns->coef[w][filt], order, lpc, 0, 0, 0);
2477 
2478  start = ics->swb_offset[FFMIN(bottom, mmm)];
2479  end = ics->swb_offset[FFMIN( top, mmm)];
2480  if ((size = end - start) <= 0)
2481  continue;
2482  if (tns->direction[w][filt]) {
2483  inc = -1;
2484  start = end - 1;
2485  } else {
2486  inc = 1;
2487  }
2488  start += w * 128;
2489 
2490  if (decode) {
2491  // ar filter
2492  for (m = 0; m < size; m++, start += inc)
2493  for (i = 1; i <= FFMIN(m, order); i++)
2494  coef[start] -= AAC_MUL26((INTFLOAT)coef[start - i * inc], lpc[i - 1]);
2495  } else {
2496  // ma filter
2497  for (m = 0; m < size; m++, start += inc) {
2498  tmp[0] = coef[start];
2499  for (i = 1; i <= FFMIN(m, order); i++)
2500  coef[start] += AAC_MUL26(tmp[i], lpc[i - 1]);
2501  for (i = order; i > 0; i--)
2502  tmp[i] = tmp[i - 1];
2503  }
2504  }
2505  }
2506  }
2507 }
2508 
2509 /**
2510  * Apply windowing and MDCT to obtain the spectral
2511  * coefficient from the predicted sample by LTP.
2512  */
2515 {
2516  const INTFLOAT *lwindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_long_1024) : AAC_RENAME(ff_sine_1024);
2517  const INTFLOAT *swindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_short_128) : AAC_RENAME(ff_sine_128);
2518  const INTFLOAT *lwindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_long_1024) : AAC_RENAME(ff_sine_1024);
2519  const INTFLOAT *swindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_short_128) : AAC_RENAME(ff_sine_128);
2520 
2521  if (ics->window_sequence[0] != LONG_STOP_SEQUENCE) {
2522  ac->fdsp->vector_fmul(in, in, lwindow_prev, 1024);
2523  } else {
2524  memset(in, 0, 448 * sizeof(*in));
2525  ac->fdsp->vector_fmul(in + 448, in + 448, swindow_prev, 128);
2526  }
2527  if (ics->window_sequence[0] != LONG_START_SEQUENCE) {
2528  ac->fdsp->vector_fmul_reverse(in + 1024, in + 1024, lwindow, 1024);
2529  } else {
2530  ac->fdsp->vector_fmul_reverse(in + 1024 + 448, in + 1024 + 448, swindow, 128);
2531  memset(in + 1024 + 576, 0, 448 * sizeof(*in));
2532  }
2533  ac->mdct_ltp.mdct_calc(&ac->mdct_ltp, out, in);
2534 }
2535 
2536 /**
2537  * Apply the long term prediction
2538  */
2540 {
2541  const LongTermPrediction *ltp = &sce->ics.ltp;
2542  const uint16_t *offsets = sce->ics.swb_offset;
2543  int i, sfb;
2544 
2545  if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
2546  INTFLOAT *predTime = sce->ret;
2547  INTFLOAT *predFreq = ac->buf_mdct;
2548  int16_t num_samples = 2048;
2549 
2550  if (ltp->lag < 1024)
2551  num_samples = ltp->lag + 1024;
2552  for (i = 0; i < num_samples; i++)
2553  predTime[i] = AAC_MUL30(sce->ltp_state[i + 2048 - ltp->lag], ltp->coef);
2554  memset(&predTime[i], 0, (2048 - i) * sizeof(*predTime));
2555 
2556  ac->windowing_and_mdct_ltp(ac, predFreq, predTime, &sce->ics);
2557 
2558  if (sce->tns.present)
2559  ac->apply_tns(predFreq, &sce->tns, &sce->ics, 0);
2560 
2561  for (sfb = 0; sfb < FFMIN(sce->ics.max_sfb, MAX_LTP_LONG_SFB); sfb++)
2562  if (ltp->used[sfb])
2563  for (i = offsets[sfb]; i < offsets[sfb + 1]; i++)
2564  sce->coeffs[i] += predFreq[i];
2565  }
2566 }
2567 
2568 /**
2569  * Update the LTP buffer for next frame
2570  */
2572 {
2573  IndividualChannelStream *ics = &sce->ics;
2574  INTFLOAT *saved = sce->saved;
2575  INTFLOAT *saved_ltp = sce->coeffs;
2576  const INTFLOAT *lwindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_long_1024) : AAC_RENAME(ff_sine_1024);
2577  const INTFLOAT *swindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_short_128) : AAC_RENAME(ff_sine_128);
2578  int i;
2579 
2580  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2581  memcpy(saved_ltp, saved, 512 * sizeof(*saved_ltp));
2582  memset(saved_ltp + 576, 0, 448 * sizeof(*saved_ltp));
2583  ac->fdsp->vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960, &swindow[64], 64);
2584 
2585  for (i = 0; i < 64; i++)
2586  saved_ltp[i + 512] = AAC_MUL31(ac->buf_mdct[1023 - i], swindow[63 - i]);
2587  } else if (ics->window_sequence[0] == LONG_START_SEQUENCE) {
2588  memcpy(saved_ltp, ac->buf_mdct + 512, 448 * sizeof(*saved_ltp));
2589  memset(saved_ltp + 576, 0, 448 * sizeof(*saved_ltp));
2590  ac->fdsp->vector_fmul_reverse(saved_ltp + 448, ac->buf_mdct + 960, &swindow[64], 64);
2591 
2592  for (i = 0; i < 64; i++)
2593  saved_ltp[i + 512] = AAC_MUL31(ac->buf_mdct[1023 - i], swindow[63 - i]);
2594  } else { // LONG_STOP or ONLY_LONG
2595  ac->fdsp->vector_fmul_reverse(saved_ltp, ac->buf_mdct + 512, &lwindow[512], 512);
2596 
2597  for (i = 0; i < 512; i++)
2598  saved_ltp[i + 512] = AAC_MUL31(ac->buf_mdct[1023 - i], lwindow[511 - i]);
2599  }
2600 
2601  memcpy(sce->ltp_state, sce->ltp_state+1024, 1024 * sizeof(*sce->ltp_state));
2602  memcpy(sce->ltp_state+1024, sce->ret, 1024 * sizeof(*sce->ltp_state));
2603  memcpy(sce->ltp_state+2048, saved_ltp, 1024 * sizeof(*sce->ltp_state));
2604 }
2605 
2606 /**
2607  * Conduct IMDCT and windowing.
2608  */
2610 {
2611  IndividualChannelStream *ics = &sce->ics;
2612  INTFLOAT *in = sce->coeffs;
2613  INTFLOAT *out = sce->ret;
2614  INTFLOAT *saved = sce->saved;
2615  const INTFLOAT *swindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_short_128) : AAC_RENAME(ff_sine_128);
2616  const INTFLOAT *lwindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_long_1024) : AAC_RENAME(ff_sine_1024);
2617  const INTFLOAT *swindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_short_128) : AAC_RENAME(ff_sine_128);
2618  INTFLOAT *buf = ac->buf_mdct;
2619  INTFLOAT *temp = ac->temp;
2620  int i;
2621 
2622  // imdct
2623  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2624  for (i = 0; i < 1024; i += 128)
2625  ac->mdct_small.imdct_half(&ac->mdct_small, buf + i, in + i);
2626  } else {
2627  ac->mdct.imdct_half(&ac->mdct, buf, in);
2628 #if USE_FIXED
2629  for (i=0; i<1024; i++)
2630  buf[i] = (buf[i] + 4) >> 3;
2631 #endif /* USE_FIXED */
2632  }
2633 
2634  /* window overlapping
2635  * NOTE: To simplify the overlapping code, all 'meaningless' short to long
2636  * and long to short transitions are considered to be short to short
2637  * transitions. This leaves just two cases (long to long and short to short)
2638  * with a little special sauce for EIGHT_SHORT_SEQUENCE.
2639  */
2640  if ((ics->window_sequence[1] == ONLY_LONG_SEQUENCE || ics->window_sequence[1] == LONG_STOP_SEQUENCE) &&
2642  ac->fdsp->vector_fmul_window( out, saved, buf, lwindow_prev, 512);
2643  } else {
2644  memcpy( out, saved, 448 * sizeof(*out));
2645 
2646  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2647  ac->fdsp->vector_fmul_window(out + 448 + 0*128, saved + 448, buf + 0*128, swindow_prev, 64);
2648  ac->fdsp->vector_fmul_window(out + 448 + 1*128, buf + 0*128 + 64, buf + 1*128, swindow, 64);
2649  ac->fdsp->vector_fmul_window(out + 448 + 2*128, buf + 1*128 + 64, buf + 2*128, swindow, 64);
2650  ac->fdsp->vector_fmul_window(out + 448 + 3*128, buf + 2*128 + 64, buf + 3*128, swindow, 64);
2651  ac->fdsp->vector_fmul_window(temp, buf + 3*128 + 64, buf + 4*128, swindow, 64);
2652  memcpy( out + 448 + 4*128, temp, 64 * sizeof(*out));
2653  } else {
2654  ac->fdsp->vector_fmul_window(out + 448, saved + 448, buf, swindow_prev, 64);
2655  memcpy( out + 576, buf + 64, 448 * sizeof(*out));
2656  }
2657  }
2658 
2659  // buffer update
2660  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2661  memcpy( saved, temp + 64, 64 * sizeof(*saved));
2662  ac->fdsp->vector_fmul_window(saved + 64, buf + 4*128 + 64, buf + 5*128, swindow, 64);
2663  ac->fdsp->vector_fmul_window(saved + 192, buf + 5*128 + 64, buf + 6*128, swindow, 64);
2664  ac->fdsp->vector_fmul_window(saved + 320, buf + 6*128 + 64, buf + 7*128, swindow, 64);
2665  memcpy( saved + 448, buf + 7*128 + 64, 64 * sizeof(*saved));
2666  } else if (ics->window_sequence[0] == LONG_START_SEQUENCE) {
2667  memcpy( saved, buf + 512, 448 * sizeof(*saved));
2668  memcpy( saved + 448, buf + 7*128 + 64, 64 * sizeof(*saved));
2669  } else { // LONG_STOP or ONLY_LONG
2670  memcpy( saved, buf + 512, 512 * sizeof(*saved));
2671  }
2672 }
2673 
2674 /**
2675  * Conduct IMDCT and windowing.
2676  */
2678 {
2679 #if !USE_FIXED
2680  IndividualChannelStream *ics = &sce->ics;
2681  INTFLOAT *in = sce->coeffs;
2682  INTFLOAT *out = sce->ret;
2683  INTFLOAT *saved = sce->saved;
2684  const INTFLOAT *swindow = ics->use_kb_window[0] ? AAC_RENAME(ff_aac_kbd_short_120) : AAC_RENAME(ff_sine_120);
2685  const INTFLOAT *lwindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_long_960) : AAC_RENAME(ff_sine_960);
2686  const INTFLOAT *swindow_prev = ics->use_kb_window[1] ? AAC_RENAME(ff_aac_kbd_short_120) : AAC_RENAME(ff_sine_120);
2687  INTFLOAT *buf = ac->buf_mdct;
2688  INTFLOAT *temp = ac->temp;
2689  int i;
2690 
2691  // imdct
2692  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2693  for (i = 0; i < 8; i++)
2694  ac->mdct120->imdct_half(ac->mdct120, buf + i * 120, in + i * 128, 1);
2695  } else {
2696  ac->mdct960->imdct_half(ac->mdct960, buf, in, 1);
2697  }
2698 
2699  /* window overlapping
2700  * NOTE: To simplify the overlapping code, all 'meaningless' short to long
2701  * and long to short transitions are considered to be short to short
2702  * transitions. This leaves just two cases (long to long and short to short)
2703  * with a little special sauce for EIGHT_SHORT_SEQUENCE.
2704  */
2705 
2706  if ((ics->window_sequence[1] == ONLY_LONG_SEQUENCE || ics->window_sequence[1] == LONG_STOP_SEQUENCE) &&
2708  ac->fdsp->vector_fmul_window( out, saved, buf, lwindow_prev, 480);
2709  } else {
2710  memcpy( out, saved, 420 * sizeof(*out));
2711 
2712  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2713  ac->fdsp->vector_fmul_window(out + 420 + 0*120, saved + 420, buf + 0*120, swindow_prev, 60);
2714  ac->fdsp->vector_fmul_window(out + 420 + 1*120, buf + 0*120 + 60, buf + 1*120, swindow, 60);
2715  ac->fdsp->vector_fmul_window(out + 420 + 2*120, buf + 1*120 + 60, buf + 2*120, swindow, 60);
2716  ac->fdsp->vector_fmul_window(out + 420 + 3*120, buf + 2*120 + 60, buf + 3*120, swindow, 60);
2717  ac->fdsp->vector_fmul_window(temp, buf + 3*120 + 60, buf + 4*120, swindow, 60);
2718  memcpy( out + 420 + 4*120, temp, 60 * sizeof(*out));
2719  } else {
2720  ac->fdsp->vector_fmul_window(out + 420, saved + 420, buf, swindow_prev, 60);
2721  memcpy( out + 540, buf + 60, 420 * sizeof(*out));
2722  }
2723  }
2724 
2725  // buffer update
2726  if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) {
2727  memcpy( saved, temp + 60, 60 * sizeof(*saved));
2728  ac->fdsp->vector_fmul_window(saved + 60, buf + 4*120 + 60, buf + 5*120, swindow, 60);
2729  ac->fdsp->vector_fmul_window(saved + 180, buf + 5*120 + 60, buf + 6*120, swindow, 60);
2730  ac->fdsp->vector_fmul_window(saved + 300, buf + 6*120 + 60, buf + 7*120, swindow, 60);
2731  memcpy( saved + 420, buf + 7*120 + 60, 60 * sizeof(*saved));
2732  } else if (ics->window_sequence[0] == LONG_START_SEQUENCE) {
2733  memcpy( saved, buf + 480, 420 * sizeof(*saved));
2734  memcpy( saved + 420, buf + 7*120 + 60, 60 * sizeof(*saved));
2735  } else { // LONG_STOP or ONLY_LONG
2736  memcpy( saved, buf + 480, 480 * sizeof(*saved));
2737  }
2738 #endif
2739 }
2741 {
2742  IndividualChannelStream *ics = &sce->ics;
2743  INTFLOAT *in = sce->coeffs;
2744  INTFLOAT *out = sce->ret;
2745  INTFLOAT *saved = sce->saved;
2746  INTFLOAT *buf = ac->buf_mdct;
2747 #if USE_FIXED
2748  int i;
2749 #endif /* USE_FIXED */
2750 
2751  // imdct
2752  ac->mdct.imdct_half(&ac->mdct_ld, buf, in);
2753 
2754 #if USE_FIXED
2755  for (i = 0; i < 1024; i++)
2756  buf[i] = (buf[i] + 2) >> 2;
2757 #endif /* USE_FIXED */
2758 
2759  // window overlapping
2760  if (ics->use_kb_window[1]) {
2761  // AAC LD uses a low overlap sine window instead of a KBD window
2762  memcpy(out, saved, 192 * sizeof(*out));
2763  ac->fdsp->vector_fmul_window(out + 192, saved + 192, buf, AAC_RENAME(ff_sine_128), 64);
2764  memcpy( out + 320, buf + 64, 192 * sizeof(*out));
2765  } else {
2766  ac->fdsp->vector_fmul_window(out, saved, buf, AAC_RENAME(ff_sine_512), 256);
2767  }
2768 
2769  // buffer update
2770  memcpy(saved, buf + 256, 256 * sizeof(*saved));
2771 }
2772 
2774 {
2775  INTFLOAT *in = sce->coeffs;
2776  INTFLOAT *out = sce->ret;
2777  INTFLOAT *saved = sce->saved;
2778  INTFLOAT *buf = ac->buf_mdct;
2779  int i;
2780  const int n = ac->oc[1].m4ac.frame_length_short ? 480 : 512;
2781  const int n2 = n >> 1;
2782  const int n4 = n >> 2;
2783  const INTFLOAT *const window = n == 480 ? AAC_RENAME(ff_aac_eld_window_480) :
2785 
2786  // Inverse transform, mapped to the conventional IMDCT by
2787  // Chivukula, R.K.; Reznik, Y.A.; Devarajan, V.,
2788  // "Efficient algorithms for MPEG-4 AAC-ELD, AAC-LD and AAC-LC filterbanks,"
2789  // International Conference on Audio, Language and Image Processing, ICALIP 2008.
2790  // URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4590245&isnumber=4589950
2791  for (i = 0; i < n2; i+=2) {
2792  INTFLOAT temp;
2793  temp = in[i ]; in[i ] = -in[n - 1 - i]; in[n - 1 - i] = temp;
2794  temp = -in[i + 1]; in[i + 1] = in[n - 2 - i]; in[n - 2 - i] = temp;
2795  }
2796 #if !USE_FIXED
2797  if (n == 480)
2798  ac->mdct480->imdct_half(ac->mdct480, buf, in, 1);
2799  else
2800 #endif
2801  ac->mdct.imdct_half(&ac->mdct_ld, buf, in);
2802 
2803 #if USE_FIXED
2804  for (i = 0; i < 1024; i++)
2805  buf[i] = (buf[i] + 1) >> 1;
2806 #endif /* USE_FIXED */
2807 
2808  for (i = 0; i < n; i+=2) {
2809  buf[i] = -buf[i];
2810  }
2811  // Like with the regular IMDCT at this point we still have the middle half
2812  // of a transform but with even symmetry on the left and odd symmetry on
2813  // the right
2814 
2815  // window overlapping
2816  // The spec says to use samples [0..511] but the reference decoder uses
2817  // samples [128..639].
2818  for (i = n4; i < n2; i ++) {
2819  out[i - n4] = AAC_MUL31( buf[ n2 - 1 - i] , window[i - n4]) +
2820  AAC_MUL31( saved[ i + n2] , window[i + n - n4]) +
2821  AAC_MUL31(-saved[n + n2 - 1 - i] , window[i + 2*n - n4]) +
2822  AAC_MUL31(-saved[ 2*n + n2 + i] , window[i + 3*n - n4]);
2823  }
2824  for (i = 0; i < n2; i ++) {
2825  out[n4 + i] = AAC_MUL31( buf[ i] , window[i + n2 - n4]) +
2826  AAC_MUL31(-saved[ n - 1 - i] , window[i + n2 + n - n4]) +
2827  AAC_MUL31(-saved[ n + i] , window[i + n2 + 2*n - n4]) +
2828  AAC_MUL31( saved[2*n + n - 1 - i] , window[i + n2 + 3*n - n4]);
2829  }
2830  for (i = 0; i < n4; i ++) {
2831  out[n2 + n4 + i] = AAC_MUL31( buf[ i + n2] , window[i + n - n4]) +
2832  AAC_MUL31(-saved[n2 - 1 - i] , window[i + 2*n - n4]) +
2833  AAC_MUL31(-saved[n + n2 + i] , window[i + 3*n - n4]);
2834  }
2835 
2836  // buffer update
2837  memmove(saved + n, saved, 2 * n * sizeof(*saved));
2838  memcpy( saved, buf, n * sizeof(*saved));
2839 }
2840 
2841 /**
2842  * channel coupling transformation interface
2843  *
2844  * @param apply_coupling_method pointer to (in)dependent coupling function
2845  */
2847  enum RawDataBlockType type, int elem_id,
2848  enum CouplingPoint coupling_point,
2849  void (*apply_coupling_method)(AACContext *ac, SingleChannelElement *target, ChannelElement *cce, int index))
2850 {
2851  int i, c;
2852 
2853  for (i = 0; i < MAX_ELEM_ID; i++) {
2854  ChannelElement *cce = ac->che[TYPE_CCE][i];
2855  int index = 0;
2856 
2857  if (cce && cce->coup.coupling_point == coupling_point) {
2858  ChannelCoupling *coup = &cce->coup;
2859 
2860  for (c = 0; c <= coup->num_coupled; c++) {
2861  if (coup->type[c] == type && coup->id_select[c] == elem_id) {
2862  if (coup->ch_select[c] != 1) {
2863  apply_coupling_method(ac, &cc->ch[0], cce, index);
2864  if (coup->ch_select[c] != 0)
2865  index++;
2866  }
2867  if (coup->ch_select[c] != 2)
2868  apply_coupling_method(ac, &cc->ch[1], cce, index++);
2869  } else
2870  index += 1 + (coup->ch_select[c] == 3);
2871  }
2872  }
2873  }
2874 }
2875 
2876 /**
2877  * Convert spectral data to samples, applying all supported tools as appropriate.
2878  */
2879 static void spectral_to_sample(AACContext *ac, int samples)
2880 {
2881  int i, type;
2883  switch (ac->oc[1].m4ac.object_type) {
2884  case AOT_ER_AAC_LD:
2886  break;
2887  case AOT_ER_AAC_ELD:
2889  break;
2890  default:
2891  if (ac->oc[1].m4ac.frame_length_short)
2893  else
2895  }
2896  for (type = 3; type >= 0; type--) {
2897  for (i = 0; i < MAX_ELEM_ID; i++) {
2898  ChannelElement *che = ac->che[type][i];
2899  if (che && che->present) {
2900  if (type <= TYPE_CPE)
2902  if (ac->oc[1].m4ac.object_type == AOT_AAC_LTP) {
2903  if (che->ch[0].ics.predictor_present) {
2904  if (che->ch[0].ics.ltp.present)
2905  ac->apply_ltp(ac, &che->ch[0]);
2906  if (che->ch[1].ics.ltp.present && type == TYPE_CPE)
2907  ac->apply_ltp(ac, &che->ch[1]);
2908  }
2909  }
2910  if (che->ch[0].tns.present)
2911  ac->apply_tns(che->ch[0].coeffs, &che->ch[0].tns, &che->ch[0].ics, 1);
2912  if (che->ch[1].tns.present)
2913  ac->apply_tns(che->ch[1].coeffs, &che->ch[1].tns, &che->ch[1].ics, 1);
2914  if (type <= TYPE_CPE)
2916  if (type != TYPE_CCE || che->coup.coupling_point == AFTER_IMDCT) {
2917  imdct_and_window(ac, &che->ch[0]);
2918  if (ac->oc[1].m4ac.object_type == AOT_AAC_LTP)
2919  ac->update_ltp(ac, &che->ch[0]);
2920  if (type == TYPE_CPE) {
2921  imdct_and_window(ac, &che->ch[1]);
2922  if (ac->oc[1].m4ac.object_type == AOT_AAC_LTP)
2923  ac->update_ltp(ac, &che->ch[1]);
2924  }
2925  if (ac->oc[1].m4ac.sbr > 0) {
2926  AAC_RENAME(ff_sbr_apply)(ac, &che->sbr, type, che->ch[0].ret, che->ch[1].ret);
2927  }
2928  }
2929  if (type <= TYPE_CCE)
2931 
2932 #if USE_FIXED
2933  {
2934  int j;
2935  /* preparation for resampler */
2936  for(j = 0; j<samples; j++){
2937  che->ch[0].ret[j] = (int32_t)av_clip64((int64_t)che->ch[0].ret[j]*128, INT32_MIN, INT32_MAX-0x8000)+0x8000;
2938  if(type == TYPE_CPE)
2939  che->ch[1].ret[j] = (int32_t)av_clip64((int64_t)che->ch[1].ret[j]*128, INT32_MIN, INT32_MAX-0x8000)+0x8000;
2940  }
2941  }
2942 #endif /* USE_FIXED */
2943  che->present = 0;
2944  } else if (che) {
2945  av_log(ac->avctx, AV_LOG_VERBOSE, "ChannelElement %d.%d missing \n", type, i);
2946  }
2947  }
2948  }
2949 }
2950 
2952 {
2953  int size;
2954  AACADTSHeaderInfo hdr_info;
2955  uint8_t layout_map[MAX_ELEM_ID*4][3];
2956  int layout_map_tags, ret;
2957 
2958  size = avpriv_aac_parse_header(gb, &hdr_info);
2959  if (size > 0) {
2960  if (!ac->warned_num_aac_frames && hdr_info.num_aac_frames != 1) {
2961  // This is 2 for "VLB " audio in NSV files.
2962  // See samples/nsv/vlb_audio.
2964  "More than one AAC RDB per ADTS frame");
2965  ac->warned_num_aac_frames = 1;
2966  }
2968  if (hdr_info.chan_config) {
2969  ac->oc[1].m4ac.chan_config = hdr_info.chan_config;
2970  if ((ret = set_default_channel_config(ac->avctx,
2971  layout_map,
2972  &layout_map_tags,
2973  hdr_info.chan_config)) < 0)
2974  return ret;
2975  if ((ret = output_configure(ac, layout_map, layout_map_tags,
2976  FFMAX(ac->oc[1].status,
2977  OC_TRIAL_FRAME), 0)) < 0)
2978  return ret;
2979  } else {
2980  ac->oc[1].m4ac.chan_config = 0;
2981  /**
2982  * dual mono frames in Japanese DTV can have chan_config 0
2983  * WITHOUT specifying PCE.
2984  * thus, set dual mono as default.
2985  */
2986  if (ac->dmono_mode && ac->oc[0].status == OC_NONE) {
2987  layout_map_tags = 2;
2988  layout_map[0][0] = layout_map[1][0] = TYPE_SCE;
2989  layout_map[0][2] = layout_map[1][2] = AAC_CHANNEL_FRONT;
2990  layout_map[0][1] = 0;
2991  layout_map[1][1] = 1;
2992  if (output_configure(ac, layout_map, layout_map_tags,
2993  OC_TRIAL_FRAME, 0))
2994  return -7;
2995  }
2996  }
2997  ac->oc[1].m4ac.sample_rate = hdr_info.sample_rate;
2998  ac->oc[1].m4ac.sampling_index = hdr_info.sampling_index;
2999  ac->oc[1].m4ac.object_type = hdr_info.object_type;
3000  ac->oc[1].m4ac.frame_length_short = 0;
3001  if (ac->oc[0].status != OC_LOCKED ||
3002  ac->oc[0].m4ac.chan_config != hdr_info.chan_config ||
3003  ac->oc[0].m4ac.sample_rate != hdr_info.sample_rate) {
3004  ac->oc[1].m4ac.sbr = -1;
3005  ac->oc[1].m4ac.ps = -1;
3006  }
3007  if (!hdr_info.crc_absent)
3008  skip_bits(gb, 16);
3009  }
3010  return size;
3011 }
3012 
3013 static int aac_decode_er_frame(AVCodecContext *avctx, void *data,
3014  int *got_frame_ptr, GetBitContext *gb)
3015 {
3016  AACContext *ac = avctx->priv_data;
3017  const MPEG4AudioConfig *const m4ac = &ac->oc[1].m4ac;
3018  ChannelElement *che;
3019  int err, i;
3020  int samples = m4ac->frame_length_short ? 960 : 1024;
3021  int chan_config = m4ac->chan_config;
3022  int aot = m4ac->object_type;
3023 
3024  if (aot == AOT_ER_AAC_LD || aot == AOT_ER_AAC_ELD)
3025  samples >>= 1;
3026 
3027  ac->frame = data;
3028 
3029  if ((err = frame_configure_elements(avctx)) < 0)
3030  return err;
3031 
3032  // The FF_PROFILE_AAC_* defines are all object_type - 1
3033  // This may lead to an undefined profile being signaled
3034  ac->avctx->profile = aot - 1;
3035 
3036  ac->tags_mapped = 0;
3037 
3038  if (chan_config < 0 || (chan_config >= 8 && chan_config < 11) || chan_config >= 13) {
3039  avpriv_request_sample(avctx, "Unknown ER channel configuration %d",
3040  chan_config);
3041  return AVERROR_INVALIDDATA;
3042  }
3043  for (i = 0; i < tags_per_config[chan_config]; i++) {
3044  const int elem_type = aac_channel_layout_map[chan_config-1][i][0];
3045  const int elem_id = aac_channel_layout_map[chan_config-1][i][1];
3046  if (!(che=get_che(ac, elem_type, elem_id))) {
3047  av_log(ac->avctx, AV_LOG_ERROR,
3048  "channel element %d.%d is not allocated\n",
3049  elem_type, elem_id);
3050  return AVERROR_INVALIDDATA;
3051  }
3052  che->present = 1;
3053  if (aot != AOT_ER_AAC_ELD)
3054  skip_bits(gb, 4);
3055  switch (elem_type) {
3056  case TYPE_SCE:
3057  err = decode_ics(ac, &che->ch[0], gb, 0, 0);
3058  break;
3059  case TYPE_CPE:
3060  err = decode_cpe(ac, gb, che);
3061  break;
3062  case TYPE_LFE:
3063  err = decode_ics(ac, &che->ch[0], gb, 0, 0);
3064  break;
3065  }
3066  if (err < 0)
3067  return err;
3068  }
3069 
3070  spectral_to_sample(ac, samples);
3071 
3072  if (!ac->frame->data[0] && samples) {
3073  av_log(avctx, AV_LOG_ERROR, "no frame data found\n");
3074  return AVERROR_INVALIDDATA;
3075  }
3076 
3077  ac->frame->nb_samples = samples;
3078  ac->frame->sample_rate = avctx->sample_rate;
3079  *got_frame_ptr = 1;
3080 
3081  skip_bits_long(gb, get_bits_left(gb));
3082  return 0;
3083 }
3084 
3085 static int aac_decode_frame_int(AVCodecContext *avctx, void *data,
3086  int *got_frame_ptr, GetBitContext *gb, AVPacket *avpkt)
3087 {
3088  AACContext *ac = avctx->priv_data;
3089  ChannelElement *che = NULL, *che_prev = NULL;
3090  enum RawDataBlockType elem_type, che_prev_type = TYPE_END;
3091  int err, elem_id;
3092  int samples = 0, multiplier, audio_found = 0, pce_found = 0;
3093  int is_dmono, sce_count = 0;
3094  int payload_alignment;
3095 
3096  ac->frame = data;
3097 
3098  if (show_bits(gb, 12) == 0xfff) {
3099  if ((err = parse_adts_frame_header(ac, gb)) < 0) {
3100  av_log(avctx, AV_LOG_ERROR, "Error decoding AAC frame header.\n");
3101  goto fail;
3102  }
3103  if (ac->oc[1].m4ac.sampling_index > 12) {
3104  av_log(ac->avctx, AV_LOG_ERROR, "invalid sampling rate index %d\n", ac->oc[1].m4ac.sampling_index);
3105  err = AVERROR_INVALIDDATA;
3106  goto fail;
3107  }
3108  }
3109 
3110  if ((err = frame_configure_elements(avctx)) < 0)
3111  goto fail;
3112 
3113  // The FF_PROFILE_AAC_* defines are all object_type - 1
3114  // This may lead to an undefined profile being signaled
3115  ac->avctx->profile = ac->oc[1].m4ac.object_type - 1;
3116 
3117  payload_alignment = get_bits_count(gb);
3118  ac->tags_mapped = 0;
3119  // parse
3120  while ((elem_type = get_bits(gb, 3)) != TYPE_END) {
3121  elem_id = get_bits(gb, 4);
3122 
3123  if (avctx->debug & FF_DEBUG_STARTCODE)
3124  av_log(avctx, AV_LOG_DEBUG, "Elem type:%x id:%x\n", elem_type, elem_id);
3125 
3126  if (!avctx->channels && elem_type != TYPE_PCE) {
3127  err = AVERROR_INVALIDDATA;
3128  goto fail;
3129  }
3130 
3131  if (elem_type < TYPE_DSE) {
3132  if (!(che=get_che(ac, elem_type, elem_id))) {
3133  av_log(ac->avctx, AV_LOG_ERROR, "channel element %d.%d is not allocated\n",
3134  elem_type, elem_id);
3135  err = AVERROR_INVALIDDATA;
3136  goto fail;
3137  }
3138  samples = ac->oc[1].m4ac.frame_length_short ? 960 : 1024;
3139  che->present = 1;
3140  }
3141 
3142  switch (elem_type) {
3143 
3144  case TYPE_SCE:
3145  err = decode_ics(ac, &che->ch[0], gb, 0, 0);
3146  audio_found = 1;
3147  sce_count++;
3148  break;
3149 
3150  case TYPE_CPE:
3151  err = decode_cpe(ac, gb, che);
3152  audio_found = 1;
3153  break;
3154 
3155  case TYPE_CCE:
3156  err = decode_cce(ac, gb, che);
3157  break;
3158 
3159  case TYPE_LFE:
3160  err = decode_ics(ac, &che->ch[0], gb, 0, 0);
3161  audio_found = 1;
3162  break;
3163 
3164  case TYPE_DSE:
3165  err = skip_data_stream_element(ac, gb);
3166  break;
3167 
3168  case TYPE_PCE: {
3169  uint8_t layout_map[MAX_ELEM_ID*4][3];
3170  int tags;
3171 
3172  int pushed = push_output_configuration(ac);
3173  if (pce_found && !pushed) {
3174  err = AVERROR_INVALIDDATA;
3175  goto fail;
3176  }
3177 
3178  tags = decode_pce(avctx, &ac->oc[1].m4ac, layout_map, gb,
3179  payload_alignment);
3180  if (tags < 0) {
3181  err = tags;
3182  break;
3183  }
3184  if (pce_found) {
3185  av_log(avctx, AV_LOG_ERROR,
3186  "Not evaluating a further program_config_element as this construct is dubious at best.\n");
3188  } else {
3189  err = output_configure(ac, layout_map, tags, OC_TRIAL_PCE, 1);
3190  if (!err)
3191  ac->oc[1].m4ac.chan_config = 0;
3192  pce_found = 1;
3193  }
3194  break;
3195  }
3196 
3197  case TYPE_FIL:
3198  if (elem_id == 15)
3199  elem_id += get_bits(gb, 8) - 1;
3200  if (get_bits_left(gb) < 8 * elem_id) {
3201  av_log(avctx, AV_LOG_ERROR, "TYPE_FIL: "overread_err);
3202  err = AVERROR_INVALIDDATA;
3203  goto fail;
3204  }
3205  while (elem_id > 0)
3206  elem_id -= decode_extension_payload(ac, gb, elem_id, che_prev, che_prev_type);
3207  err = 0; /* FIXME */
3208  break;
3209 
3210  default:
3211  err = AVERROR_BUG; /* should not happen, but keeps compiler happy */
3212  break;
3213  }
3214 
3215  if (elem_type < TYPE_DSE) {
3216  che_prev = che;
3217  che_prev_type = elem_type;
3218  }
3219 
3220  if (err)
3221  goto fail;
3222 
3223  if (get_bits_left(gb) < 3) {
3224  av_log(avctx, AV_LOG_ERROR, overread_err);
3225  err = AVERROR_INVALIDDATA;
3226  goto fail;
3227  }
3228  }
3229 
3230  if (!avctx->channels) {
3231  *got_frame_ptr = 0;
3232  return 0;
3233  }
3234 
3235  multiplier = (ac->oc[1].m4ac.sbr == 1) ? ac->oc[1].m4ac.ext_sample_rate > ac->oc[1].m4ac.sample_rate : 0;
3236  samples <<= multiplier;
3237 
3238  spectral_to_sample(ac, samples);
3239 
3240  if (ac->oc[1].status && audio_found) {
3241  avctx->sample_rate = ac->oc[1].m4ac.sample_rate << multiplier;
3242  avctx->frame_size = samples;
3243  ac->oc[1].status = OC_LOCKED;
3244  }
3245 
3246  if (multiplier)
3247  avctx->internal->skip_samples_multiplier = 2;
3248 
3249  if (!ac->frame->data[0] && samples) {
3250  av_log(avctx, AV_LOG_ERROR, "no frame data found\n");
3251  err = AVERROR_INVALIDDATA;
3252  goto fail;
3253  }
3254 
3255  if (samples) {
3256  ac->frame->nb_samples = samples;
3257  ac->frame->sample_rate = avctx->sample_rate;
3258  } else
3259  av_frame_unref(ac->frame);
3260  *got_frame_ptr = !!samples;
3261 
3262  /* for dual-mono audio (SCE + SCE) */
3263  is_dmono = ac->dmono_mode && sce_count == 2 &&
3265  if (is_dmono) {
3266  if (ac->dmono_mode == 1)
3267  ((AVFrame *)data)->data[1] =((AVFrame *)data)->data[0];
3268  else if (ac->dmono_mode == 2)
3269  ((AVFrame *)data)->data[0] =((AVFrame *)data)->data[1];
3270  }
3271 
3272  return 0;
3273 fail:
3275  return err;
3276 }
3277 
3278 static int aac_decode_frame(AVCodecContext *avctx, void *data,
3279  int *got_frame_ptr, AVPacket *avpkt)
3280 {
3281  AACContext *ac = avctx->priv_data;
3282  const uint8_t *buf = avpkt->data;
3283  int buf_size = avpkt->size;
3284  GetBitContext gb;
3285  int buf_consumed;
3286  int buf_offset;
3287  int err;
3288  int new_extradata_size;
3289  const uint8_t *new_extradata = av_packet_get_side_data(avpkt,
3291  &new_extradata_size);
3292  int jp_dualmono_size;
3293  const uint8_t *jp_dualmono = av_packet_get_side_data(avpkt,
3295  &jp_dualmono_size);
3296 
3297  if (new_extradata && 0) {
3298  av_free(avctx->extradata);
3299  avctx->extradata = av_mallocz(new_extradata_size +
3301  if (!avctx->extradata)
3302  return AVERROR(ENOMEM);
3303  avctx->extradata_size = new_extradata_size;
3304  memcpy(avctx->extradata, new_extradata, new_extradata_size);
3306  if (decode_audio_specific_config(ac, ac->avctx, &ac->oc[1].m4ac,
3307  avctx->extradata,
3308  avctx->extradata_size*8LL, 1) < 0) {
3310  return AVERROR_INVALIDDATA;
3311  }
3312  }
3313 
3314  ac->dmono_mode = 0;
3315  if (jp_dualmono && jp_dualmono_size > 0)
3316  ac->dmono_mode = 1 + *jp_dualmono;
3317  if (ac->force_dmono_mode >= 0)
3318  ac->dmono_mode = ac->force_dmono_mode;
3319 
3320  if (INT_MAX / 8 <= buf_size)
3321  return AVERROR_INVALIDDATA;
3322 
3323  if ((err = init_get_bits8(&gb, buf, buf_size)) < 0)
3324  return err;
3325 
3326  switch (ac->oc[1].m4ac.object_type) {
3327  case AOT_ER_AAC_LC:
3328  case AOT_ER_AAC_LTP:
3329  case AOT_ER_AAC_LD:
3330  case AOT_ER_AAC_ELD:
3331  err = aac_decode_er_frame(avctx, data, got_frame_ptr, &gb);
3332  break;
3333  default:
3334  err = aac_decode_frame_int(avctx, data, got_frame_ptr, &gb, avpkt);
3335  }
3336  if (err < 0)
3337  return err;
3338 
3339  buf_consumed = (get_bits_count(&gb) + 7) >> 3;
3340  for (buf_offset = buf_consumed; buf_offset < buf_size; buf_offset++)
3341  if (buf[buf_offset])
3342  break;
3343 
3344  return buf_size > buf_offset ? buf_consumed : buf_size;
3345 }
3346 
3348 {
3349  AACContext *ac = avctx->priv_data;
3350  int i, type;
3351 
3352  for (i = 0; i < MAX_ELEM_ID; i++) {
3353  for (type = 0; type < 4; type++) {
3354  if (ac->che[type][i])
3356  av_freep(&ac->che[type][i]);
3357  }
3358  }
3359 
3360  ff_mdct_end(&ac->mdct);
3361  ff_mdct_end(&ac->mdct_small);
3362  ff_mdct_end(&ac->mdct_ld);
3363  ff_mdct_end(&ac->mdct_ltp);
3364 #if !USE_FIXED
3365  ff_mdct15_uninit(&ac->mdct120);
3366  ff_mdct15_uninit(&ac->mdct480);
3367  ff_mdct15_uninit(&ac->mdct960);
3368 #endif
3369  av_freep(&ac->fdsp);
3370  return 0;
3371 }
3372 
3373 static void aacdec_init(AACContext *c)
3374 {
3376  c->apply_ltp = apply_ltp;
3377  c->apply_tns = apply_tns;
3379  c->update_ltp = update_ltp;
3380 #if USE_FIXED
3383 #endif
3384 
3385 #if !USE_FIXED
3386  if(ARCH_MIPS)
3388 #endif /* !USE_FIXED */
3389 }
3390 /**
3391  * AVOptions for Japanese DTV specific extensions (ADTS only)
3392  */
3393 #define AACDEC_FLAGS AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
3394 static const AVOption options[] = {
3395  {"dual_mono_mode", "Select the channel to decode for dual mono",
3396  offsetof(AACContext, force_dmono_mode), AV_OPT_TYPE_INT, {.i64=-1}, -1, 2,
3397  AACDEC_FLAGS, "dual_mono_mode"},
3398 
3399  {"auto", "autoselection", 0, AV_OPT_TYPE_CONST, {.i64=-1}, INT_MIN, INT_MAX, AACDEC_FLAGS, "dual_mono_mode"},
3400  {"main", "Select Main/Left channel", 0, AV_OPT_TYPE_CONST, {.i64= 1}, INT_MIN, INT_MAX, AACDEC_FLAGS, "dual_mono_mode"},
3401  {"sub" , "Select Sub/Right channel", 0, AV_OPT_TYPE_CONST, {.i64= 2}, INT_MIN, INT_MAX, AACDEC_FLAGS, "dual_mono_mode"},
3402  {"both", "Select both channels", 0, AV_OPT_TYPE_CONST, {.i64= 0}, INT_MIN, INT_MAX, AACDEC_FLAGS, "dual_mono_mode"},
3403 
3404  {NULL},
3405 };
3406 
3407 static const AVClass aac_decoder_class = {
3408  .class_name = "AAC decoder",
3409  .item_name = av_default_item_name,
3410  .option = options,
3411  .version = LIBAVUTIL_VERSION_INT,
3412 };
int predictor_initialized
Definition: aac.h:187
float UINTFLOAT
Definition: aac_defines.h:87
static float * VMUL4S(float *dst, const float *v, unsigned idx, unsigned sign, const float *scale)
Definition: aacdec.c:124
AVFloatDSPContext * fdsp
Definition: aac.h:333
static void apply_prediction(AACContext *ac, SingleChannelElement *sce)
Apply AAC-Main style frequency domain prediction.
float, planar
Definition: samplefmt.h:69
static void imdct_and_windowing(AACContext *ac, SingleChannelElement *sce)
Conduct IMDCT and windowing.
static void apply_ltp(AACContext *ac, SingleChannelElement *sce)
Apply the long term prediction.
#define NULL
Definition: coverity.c:32
const char * s
Definition: avisynth_c.h:768
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
Definition: aac.h:60
static int decode_pulses(Pulse *pulse, GetBitContext *gb, const uint16_t *swb_offset, int num_swb)
Decode pulse data; reference: table 4.7.
uint8_t use_kb_window[2]
If set, use Kaiser-Bessel window, otherwise use a sine window.
Definition: aac.h:177
float ff_aac_kbd_short_120[120]
Definition: aactab.c:41
INTFLOAT buf_mdct[1024]
Definition: aac.h:316
void(* subband_scale)(int *dst, int *src, int scale, int offset, int len)
Definition: aac.h:369
#define overread_err
This structure describes decoded (raw) audio or video data.
Definition: frame.h:201
uint8_t object_type
Definition: aacadtsdec.h:36
AVOption.
Definition: opt.h:246
ptrdiff_t const GLvoid * data
Definition: opengl_enc.c:101
static void flush(AVCodecContext *avctx)
static const int8_t tags_per_config[16]
Definition: aacdectab.h:38
AVCodecContext * avctx
Definition: aac.h:295
Definition: aac.h:224
void(* mdct_calc)(struct FFTContext *s, FFTSample *output, const FFTSample *input)
Definition: fft.h:109
static int * DEC_UPAIR(int *dst, unsigned idx, unsigned sign)
Definition: aacdec_fixed.c:125
static AVOnce aac_table_init
float re
Definition: fft.c:82
#define AAC_MUL26(x, y)
Definition: aac_defines.h:100
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
Definition: get_bits.h:261
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:182
av_cold void ff_kbd_window_init(float *window, float alpha, int n)
Generate a Kaiser-Bessel Derived Window.
Definition: kbdwin.c:26
#define LIBAVUTIL_VERSION_INT
Definition: version.h:86
#define SCALE_DIFF_ZERO
codebook index corresponding to zero scalefactor indices difference
Definition: aac.h:152
else temp
Definition: vf_mcdeint.c:256
Definition: aac.h:63
static const float cce_scale[]
static void skip_bits_long(GetBitContext *s, int n)
Definition: get_bits.h:204
const char * g
Definition: vf_curves.c:112
static float * VMUL2S(float *dst, const float *v, unsigned idx, unsigned sign, const float *scale)
Definition: aacdec.c:107
#define AACDEC_FLAGS
AVOptions for Japanese DTV specific extensions (ADTS only)
static void imdct_and_windowing_eld(AACContext *ac, SingleChannelElement *sce)
#define INIT_VLC_STATIC(vlc, bits, a, b, c, d, e, f, g, static_size)
Definition: vlc.h:75
static void aacdec_init(AACContext *ac)
#define FIXR10(x)
Definition: aac_defines.h:93
static int * DEC_SQUAD(int *dst, unsigned idx)
Definition: aacdec_fixed.c:115
static int decode_audio_specific_config_gb(AACContext *ac, AVCodecContext *avctx, MPEG4AudioConfig *m4ac, GetBitContext *gb, int get_bit_alignment, int sync_extension)
Decode audio specific configuration; reference: table 1.13.
Definition: aac.h:56
Definition: aac.h:57
ChannelElement * che[4][MAX_ELEM_ID]
Definition: aac.h:305
int size
Definition: avcodec.h:1680
const char * b
Definition: vf_curves.c:113
float(* scalarproduct_float)(const float *v1, const float *v2, int len)
Calculate the scalar product of two vectors of floats.
Definition: float_dsp.h:175
int av_log2(unsigned v)
Definition: intmath.c:26
INTFLOAT * ret
PCM output.
Definition: aac.h:269
int present
Definition: aac.h:276
static void update_ltp(AACContext *ac, SingleChannelElement *sce)
Update the LTP buffer for next frame.
void(* update_ltp)(AACContext *ac, SingleChannelElement *sce)
Definition: aac.h:367
void(* imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce)
Definition: aac.h:361
static void vector_pow43(int *coefs, int len)
Definition: aacdec_fixed.c:151
uint64_t channel_layout
Definition: aac.h:128
INTFLOAT sf[120]
scalefactors
Definition: aac.h:255
void(* vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len)
Calculate the entry wise product of two vectors of floats, and store the result in a vector of floats...
Definition: float_dsp.h:154
#define AV_EF_BITSTREAM
detect bitstream specification deviations
Definition: avcodec.h:3059
static int decode_ga_specific_config(AACContext *ac, AVCodecContext *avctx, GetBitContext *gb, int get_bit_alignment, MPEG4AudioConfig *m4ac, int channel_config)
Decode GA "General Audio" specific configuration; reference: table 4.1.
uint8_t ms_mask[128]
Set if mid/side stereo is used for each scalefactor window band.
Definition: aac.h:281
static void apply_independent_coupling(AACContext *ac, SingleChannelElement *target, ChannelElement *cce, int index)
Apply independent channel coupling (applied after IMDCT).
Definition: aacdec.c:246
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:222
static void subband_scale(int *dst, int *src, int scale, int offset, int len)
Definition: aacdec_fixed.c:165
#define MAX_LTP_LONG_SFB
Definition: aac.h:51
#define GET_GAIN(x, y)
Definition: aac_defines.h:98
Dynamic Range Control - decoded from the bitstream but not processed further.
Definition: aac.h:211
static av_cold int che_configure(AACContext *ac, enum ChannelPosition che_pos, int type, int id, int *channels)
Check for the channel element in the current channel position configuration.
static VLC vlc_scalefactors
#define NOISE_PRE
preamble for NOISE_BT, put in bitstream with the first noise band
Definition: aac.h:156
#define FF_PROFILE_AAC_HE_V2
Definition: avcodec.h:3275
static av_always_inline void predict(PredictorState *ps, float *coef, int output_enable)
Definition: aacdec.c:174
enum RawDataBlockType type[8]
Type of channel element to be coupled - SCE or CPE.
Definition: aac.h:237
int profile
profile
Definition: avcodec.h:3266
static int output_configure(AACContext *ac, uint8_t layout_map[MAX_ELEM_ID *4][3], int tags, enum OCStatus oc_type, int get_new_frame)
Configure output channel order based on the current program configuration element.
ChannelPosition
Definition: aac.h:94
static void decode_channel_map(uint8_t layout_map[][3], enum ChannelPosition type, GetBitContext *gb, int n)
Decode an array of 4 bit element IDs, optionally interleaved with a stereo/mono switching bit...
static void decode(AVCodecContext *dec_ctx, AVPacket *pkt, AVFrame *frame, FILE *outfile)
Definition: decode_audio.c:42
static int aac_decode_er_frame(AVCodecContext *avctx, void *data, int *got_frame_ptr, GetBitContext *gb)
Spectral data are scaled white noise not coded in the bitstream.
Definition: aac.h:87
Definition: aac.h:58
static int decode_cce(AACContext *ac, GetBitContext *gb, ChannelElement *che)
Decode coupling_channel_element; reference: table 4.8.
#define USE_FIXED
Definition: aac_defines.h:25
static av_always_inline int lcg_random(unsigned previous_val)
linear congruential pseudorandom number generator
int band_incr
Number of DRC bands greater than 1 having DRC info.
Definition: aac.h:216
const uint8_t ff_aac_num_swb_128[]
Definition: aactab.c:61
#define AAC_RENAME_32(x)
Definition: aac_defines.h:85
void ff_cbrt_tableinit(void)
Definition: cbrt_tablegen.h:40
int dmono_mode
0->not dmono, 1->use first channel, 2->use second channel
Definition: aac.h:351
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
N Error Resilient Long Term Prediction.
Definition: mpeg4audio.h:86
float INTFLOAT
Definition: aac_defines.h:86
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
Definition: log.h:72
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
void void avpriv_request_sample(void *avc, const char *msg,...) av_printf_format(2
Log a generic warning message about a missing feature.
static int decode_ics_info(AACContext *ac, IndividualChannelStream *ics, GetBitContext *gb)
Decode Individual Channel Stream info; reference: table 4.6.
void(* vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, int len)
Overlap/add with window function.
Definition: float_dsp.h:119
Definition: aac.h:67
BandType
Definition: aac.h:82
uint8_t bits
Definition: crc.c:296
enum AVSampleFormat sample_fmt
audio sample format
Definition: avcodec.h:2531
uint8_t
#define FIXR(x)
Definition: aac_defines.h:92
#define av_cold
Definition: attributes.h:82
float ff_aac_kbd_long_960[960]
Definition: aactab.c:40
uint8_t layout_map[MAX_ELEM_ID *4][3]
Definition: aac.h:125
Output configuration under trial specified by an inband PCE.
Definition: aac.h:117
const uint16_t *const ff_swb_offset_480[]
Definition: aactab.c:1360
#define FF_DEBUG_PICT_INFO
Definition: avcodec.h:3004
int warned_960_sbr
Definition: aac.h:358
SingleChannelElement ch[2]
Definition: aac.h:284
const uint16_t *const ff_swb_offset_512[]
Definition: aactab.c:1352
Definition: aac.h:59
const uint8_t ff_tns_max_bands_480[]
Definition: aactab.c:1402
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
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_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);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
TemporalNoiseShaping tns
Definition: aac.h:250
N Error Resilient Low Delay.
Definition: mpeg4audio.h:90
static int decode_extension_payload(AACContext *ac, GetBitContext *gb, int cnt, ChannelElement *che, enum RawDataBlockType elem_type)
Decode extension data (incomplete); reference: table 4.51.
const uint8_t ff_aac_scalefactor_bits[121]
Definition: aactab.c:92
CouplingPoint
The point during decoding at which channel coupling is applied.
Definition: aac.h:106
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1876
int num_coupled
number of target elements
Definition: aac.h:236
#define AV_CH_LOW_FREQUENCY
av_cold int ff_mdct15_init(MDCT15Context **ps, int inverse, int N, double scale)
Definition: mdct15.c:247
int exclude_mask[MAX_CHANNELS]
Channels to be excluded from DRC processing.
Definition: aac.h:215
int n_filt[8]
Definition: aac.h:200
FFTContext mdct_ltp
Definition: aac.h:326
void(* vector_pow43)(int *coefs, int len)
Definition: aac.h:368
SingleChannelElement * output_element[MAX_CHANNELS]
Points to each SingleChannelElement.
Definition: aac.h:342
static av_cold int aac_decode_init(AVCodecContext *avctx)
uint8_t * data
Definition: avcodec.h:1679
static int get_bits_count(const GetBitContext *s)
Definition: get_bits.h:199
#define AAC_MUL31(x, y)
Definition: aac_defines.h:102
static int count_channels(uint8_t(*layout)[3], int tags)
#define ff_dlog(a,...)
Scalefactor data are intensity stereo positions (in phase).
Definition: aac.h:89
#define AV_LOG_VERBOSE
Detailed information.
Definition: log.h:192
static int sample_rate_idx(int rate)
static int decode_tns(AACContext *ac, TemporalNoiseShaping *tns, GetBitContext *gb, const IndividualChannelStream *ics)
Decode Temporal Noise Shaping data; reference: table 4.48.
#define AV_CH_BACK_LEFT
int id_select[8]
element id
Definition: aac.h:238
ptrdiff_t size
Definition: opengl_enc.c:101
const float *const ff_aac_codebook_vector_vals[]
Definition: aactab.c:1074
static av_always_inline int fixed_sqrt(int x, int bits)
Calculate the square root.
Definition: fixed_dsp.h:176
N Error Resilient Low Complexity.
Definition: mpeg4audio.h:85
ChannelElement * tag_che_map[4][MAX_ELEM_ID]
Definition: aac.h:306
#define AVOnce
Definition: thread.h:157
#define av_log(a,...)
Output configuration set in a global header but not yet locked.
Definition: aac.h:119
static void spectral_to_sample(AACContext *ac, int samples)
Convert spectral data to samples, applying all supported tools as appropriate.
int random_state
Definition: aac.h:335
MDCT15Context * mdct480
Definition: aac.h:331
#define U(x)
Definition: vp56_arith.h:37
static int parse_adts_frame_header(AACContext *ac, GetBitContext *gb)
static int get_bits_left(GetBitContext *gb)
Definition: get_bits.h:587
MPEG4AudioConfig m4ac
Definition: aac.h:124
int dyn_rng_sgn[17]
DRC sign information; 0 - positive, 1 - negative.
Definition: aac.h:213
void AAC_RENAME() ff_sbr_apply(AACContext *ac, SpectralBandReplication *sbr, int id_aac, INTFLOAT *L, INTFLOAT *R)
Apply one SBR element to one AAC element.
uint32_t ff_cbrt_tab[1<< 13]
static void pop_output_configuration(AACContext *ac)
Restore the previous output configuration if and only if the current configuration is unlocked...
static int decode_fill(AACContext *ac, GetBitContext *gb, int len)
#define UPDATE_CACHE(name, gb)
Definition: get_bits.h:160
PredictorState predictor_state[MAX_PREDICTORS]
Definition: aac.h:268
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
void(* imdct_half)(struct MDCT15Context *s, float *dst, const float *src, ptrdiff_t stride)
Definition: mdct15.h:52
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:127
const uint8_t ff_aac_num_swb_960[]
Definition: aactab.c:49
static void relative_align_get_bits(GetBitContext *gb, int reference_position)
SpectralBandReplication sbr
Definition: aac.h:287
FFTContext mdct_small
Definition: aac.h:324
void(* vector_fmul)(float *dst, const float *src0, const float *src1, int len)
Calculate the entry wise product of two vectors of floats and store the result in a vector of floats...
Definition: float_dsp.h:38
av_default_item_name
enum CouplingPoint coupling_point
The point during decoding at which coupling is applied.
Definition: aac.h:235
#define AVERROR(e)
Definition: error.h:43
const uint16_t *const ff_swb_offset_120[]
Definition: aactab.c:1378
uint8_t * av_packet_get_side_data(const AVPacket *pkt, enum AVPacketSideDataType type, int *size)
Get side information from packet.
Definition: avpacket.c:350
const uint8_t ff_aac_num_swb_1024[]
Definition: aactab.c:45
int ff_mpeg4audio_get_config_gb(MPEG4AudioConfig *c, GetBitContext *gb, int sync_extension)
Parse MPEG-4 systems extradata from a potentially unaligned GetBitContext to retrieve audio configura...
Definition: mpeg4audio.c:86
static void imdct_and_windowing_960(AACContext *ac, SingleChannelElement *sce)
Conduct IMDCT and windowing.
void(* butterflies_float)(float *av_restrict v1, float *av_restrict v2, int len)
Calculate the sum and difference of two vectors of floats.
Definition: float_dsp.h:164
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:197
float ff_aac_kbd_long_1024[1024]
Definition: aactab.c:38
INTFLOAT temp[128]
Definition: aac.h:354
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1856
static int decode_eld_specific_config(AACContext *ac, AVCodecContext *avctx, GetBitContext *gb, MPEG4AudioConfig *m4ac, int channel_config)
uint8_t sampling_index
Definition: aacadtsdec.h:37
int amp[4]
Definition: aac.h:228
static int assign_pair(struct elem_to_channel e2c_vec[MAX_ELEM_ID], uint8_t(*layout_map)[3], int offset, uint64_t left, uint64_t right, int pos)
uint8_t max_sfb
number of scalefactor bands per group
Definition: aac.h:175
#define ff_mdct_init
Definition: fft.h:169
const float ff_aac_eld_window_512[1920]
Definition: aactab.c:1411
Definition: aac.h:62
static const uint8_t offset[127][2]
Definition: vf_spp.c:92
GLsizei count
Definition: opengl_enc.c:109
#define CLOSE_READER(name, gb)
Definition: get_bits.h:131
int num_swb
number of scalefactor window bands
Definition: aac.h:183
static int count_paired_channels(uint8_t(*layout_map)[3], int tags, int pos, int *current)
#define FFMAX(a, b)
Definition: common.h:94
#define fail()
Definition: checkasm.h:109
int prog_ref_level
A reference level for the long-term program audio level for all channels combined.
Definition: aac.h:219
Output configuration locked in place.
Definition: aac.h:120
Predictor State.
Definition: aac.h:135
uint8_t chan_config
Definition: aacadtsdec.h:38
Definition: vlc.h:26
float ff_aac_pow2sf_tab[428]
Definition: aactab.c:35
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:2574
#define SKIP_BITS(name, gb, num)
Definition: get_bits.h:175
#define AAC_RENAME(x)
Definition: aac_defines.h:84
int warned_remapping_once
Definition: aac.h:308
INTFLOAT ret_buf[2048]
PCM output buffer.
Definition: aac.h:264
N Error Resilient Scalable.
Definition: mpeg4audio.h:87
static SDL_Window * window
Definition: ffplay.c:362
static void reset_predictor_group(PredictorState *ps, int group_num)
static ChannelElement * get_che(AACContext *ac, int type, int elem_id)
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:929
const uint8_t ff_aac_num_swb_512[]
Definition: aactab.c:53
int err_recognition
Error recognition; may misdetect some more or less valid parts as errors.
Definition: avcodec.h:3050
static int aac_decode_frame(AVCodecContext *avctx, void *data, int *got_frame_ptr, AVPacket *avpkt)
int predictor_reset_group
Definition: aac.h:188
static int frame_configure_elements(AVCodecContext *avctx)
#define FFMIN(a, b)
Definition: common.h:96
int dyn_rng_ctl[17]
DRC magnitude information.
Definition: aac.h:214
signed 32 bits, planar
Definition: samplefmt.h:68
static const INTFLOAT ltp_coef[8]
Definition: aactab.h:94
static void decode_mid_side_stereo(ChannelElement *cpe, GetBitContext *gb, int ms_present)
Decode Mid/Side data; reference: table 4.54.
typedef void(APIENTRY *FF_PFNGLACTIVETEXTUREPROC)(GLenum texture)
static void apply_intensity_stereo(AACContext *ac, ChannelElement *cpe, int ms_present)
intensity stereo decoding; reference: 4.6.8.2.3
uint8_t num_aac_frames
Definition: aacadtsdec.h:39
int pos[4]
Definition: aac.h:227
MDCT15Context * mdct120
Definition: aac.h:330
Y Main.
Definition: mpeg4audio.h:71
int32_t
static unsigned int show_bits(GetBitContext *s, int n)
Show 1-25 bits.
Definition: get_bits.h:296
FFTContext mdct_ld
Definition: aac.h:325
void ff_aacdec_init_mips(AACContext *c)
Definition: aacdec_mips.c:433
int AAC_RENAME() ff_decode_sbr_extension(AACContext *ac, SpectralBandReplication *sbr, GetBitContext *gb, int crc, int cnt, int id_aac)
Decode one SBR element.
#define LAST_SKIP_BITS(name, gb, num)
Definition: get_bits.h:181
static av_always_inline int get_vlc2(GetBitContext *s, VLC_TYPE(*table)[2], int bits, int max_depth)
Parse a vlc code.
Definition: get_bits.h:554
int length[8][4]
Definition: aac.h:201
void(* vector_fmul_scalar)(float *dst, const float *src, float mul, int len)
Multiply a vector of floats by a scalar float.
Definition: float_dsp.h:85
static av_cold void aac_static_table_init(void)
void AAC_RENAME() ff_aac_sbr_ctx_close(SpectralBandReplication *sbr)
Close one SBR context.
static void apply_channel_coupling(AACContext *ac, ChannelElement *cc, enum RawDataBlockType type, int elem_id, enum CouplingPoint coupling_point, void(*apply_coupling_method)(AACContext *ac, SingleChannelElement *target, ChannelElement *cce, int index))
channel coupling transformation interface
#define AV_CH_FRONT_LEFT_OF_CENTER
#define AV_EF_EXPLODE
abort decoding on minor error detection
Definition: avcodec.h:3061
int n
Definition: avisynth_c.h:684
const uint8_t ff_tns_max_bands_1024[]
Definition: aactab.c:1394
#define GET_VLC(code, name, gb, table, bits, max_depth)
If the vlc code is invalid and max_depth=1, then no bits will be removed.
Definition: get_bits.h:477
static int AAC_RENAME() compute_lpc_coefs(const LPC_TYPE *autoc, int max_order, LPC_TYPE *lpc, int lpc_stride, int fail, int normalize)
Levinson-Durbin recursion.
Definition: lpc.h:166
#define AV_CH_FRONT_CENTER
static void decode_ltp(LongTermPrediction *ltp, GetBitContext *gb, uint8_t max_sfb)
Decode Long Term Prediction data; reference: table 4.xx.
static int aac_decode_frame_int(AVCodecContext *avctx, void *data, int *got_frame_ptr, GetBitContext *gb, AVPacket *avpkt)
static void apply_dependent_coupling(AACContext *ac, SingleChannelElement *target, ChannelElement *cce, int index)
Apply dependent channel coupling (applied before IMDCT).
Definition: aacdec.c:210
static int decode_spectrum_and_dequant(AACContext *ac, INTFLOAT coef[1024], GetBitContext *gb, const INTFLOAT sf[120], int pulse_present, const Pulse *pulse, const IndividualChannelStream *ics, enum BandType band_type[120])
Decode spectral data; reference: table 4.50.
void AAC_RENAME() ff_aac_sbr_init(void)
Initialize SBR.
int pce_instance_tag
Indicates with which program the DRC info is associated.
Definition: aac.h:212
static void windowing_and_mdct_ltp(AACContext *ac, INTFLOAT *out, INTFLOAT *in, IndividualChannelStream *ics)
Apply windowing and MDCT to obtain the spectral coefficient from the predicted sample by LTP...
N Scalable.
Definition: mpeg4audio.h:76
static const INTFLOAT *const tns_tmp2_map[4]
Definition: aactab.h:126
#define SHOW_UBITS(name, gb, num)
Definition: get_bits.h:193
static int push_output_configuration(AACContext *ac)
Save current output configuration if and only if it has been locked.
#define FF_ARRAY_ELEMS(a)
#define AV_CH_FRONT_RIGHT_OF_CENTER
int interpolation_scheme
Indicates the interpolation scheme used in the SBR QMF domain.
Definition: aac.h:217
coupling parameters
Definition: aac.h:234
int tags_mapped
Definition: aac.h:307
static void reset_all_predictors(PredictorState *ps)
MDCT15Context * mdct960
Definition: aac.h:332
static int skip_data_stream_element(AACContext *ac, GetBitContext *gb)
Skip data_stream_element; reference: table 4.10.
#define AVERROR_PATCHWELCOME
Not yet implemented in FFmpeg, patches welcome.
Definition: error.h:62
int ch_select[8]
[0] shared list of gains; [1] list of gains for right channel; [2] list of gains for left channel; [3...
Definition: aac.h:239
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:2543
int force_dmono_mode
0->not dmono, 1->use first channel, 2->use second channel
Definition: aac.h:350
The AV_PKT_DATA_NEW_EXTRADATA is used to notify the codec or the format that the extradata buffer was...
Definition: avcodec.h:1420
int order[8][4]
Definition: aac.h:203
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
#define AV_ONCE_INIT
Definition: thread.h:158
int warned_num_aac_frames
Definition: aac.h:357
#define AAC_INIT_VLC_STATIC(num, size)
Temporal Noise Shaping.
Definition: aac.h:198
int sample_rate
samples per second
Definition: avcodec.h:2523
float ff_aac_kbd_short_128[128]
Definition: aactab.c:39
void AAC_RENAME() ff_sine_window_init(INTFLOAT *window, int n)
Generate a sine window.
static const AVOption options[]
static int init_get_bits8(GetBitContext *s, const uint8_t *buffer, int byte_size)
Initialize GetBitContext.
Definition: get_bits.h:456
#define AV_CH_LAYOUT_NATIVE
Channel mask value used for AVCodecContext.request_channel_layout to indicate that the user requests ...
static int decode_scalefactors(AACContext *ac, INTFLOAT sf[120], GetBitContext *gb, unsigned int global_gain, IndividualChannelStream *ics, enum BandType band_type[120], int band_type_run_end[120])
Decode scalefactors; reference: table 4.47.
int debug
debug
Definition: avcodec.h:3003
static int decode_cpe(AACContext *ac, GetBitContext *gb, ChannelElement *cpe)
Decode a channel_pair_element; reference: table 4.4.
Long Term Prediction.
Definition: aac.h:163
static void apply_tns(INTFLOAT coef_param[1024], TemporalNoiseShaping *tns, IndividualChannelStream *ics, int decode)
Decode Temporal Noise Shaping filter coefficients and apply all-pole filters; reference: 4...
main external API structure.
Definition: avcodec.h:1761
#define AV_CH_FRONT_LEFT
int skip_samples_multiplier
Definition: internal.h:219
#define NOISE_PRE_BITS
length of preamble
Definition: aac.h:157
#define OPEN_READER(name, gb)
Definition: get_bits.h:120
int ff_get_buffer(AVCodecContext *avctx, AVFrame *frame, int flags)
Get a buffer for a frame.
Definition: decode.c:1669
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
int avpriv_aac_parse_header(GetBitContext *gbc, AACADTSHeaderInfo *hdr)
Parse AAC frame header.
Definition: aacadtsdec.c:29
void * buf
Definition: avisynth_c.h:690
#define MAX_PREDICTORS
Definition: aac.h:146
static av_always_inline float cbrtf(float x)
Definition: libm.h:61
void(* imdct_half)(struct FFTContext *s, FFTSample *output, const FFTSample *input)
Definition: fft.h:108
GLint GLenum type
Definition: opengl_enc.c:105
int extradata_size
Definition: avcodec.h:1877
void AAC_RENAME() ff_aac_sbr_ctx_init(AACContext *ac, SpectralBandReplication *sbr, int id_aac)
Initialize one SBR context.
uint8_t group_len[8]
Definition: aac.h:179
#define AVERROR_BUG
Internal bug, also see AVERROR_BUG2.
Definition: error.h:50
static unsigned int get_bits1(GetBitContext *s)
Definition: get_bits.h:313
static void skip_bits1(GetBitContext *s)
Definition: get_bits.h:338
void(* apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns, IndividualChannelStream *ics, int decode)
Definition: aac.h:363
#define MAX_ELEM_ID
Definition: aac.h:48
Describe the class of an AVClass context structure.
Definition: log.h:67
int sample_rate
Sample rate of the audio data.
Definition: frame.h:374
static av_cold int aac_decode_close(AVCodecContext *avctx)
static void skip_bits(GetBitContext *s, int n)
Definition: get_bits.h:306
static int decode_audio_specific_config(AACContext *ac, AVCodecContext *avctx, MPEG4AudioConfig *m4ac, const uint8_t *data, int64_t bit_size, int sync_extension)
#define AAC_MUL30(x, y)
Definition: aac_defines.h:101
static uint64_t sniff_channel_order(uint8_t(*layout_map)[3], int tags)
const uint16_t *const ff_swb_offset_960[]
Definition: aactab.c:1344
static int decode_drc_channel_exclusions(DynamicRangeControl *che_drc, GetBitContext *gb)
Parse whether channels are to be excluded from Dynamic Range Compression; reference: table 4...
int index
Definition: gxfenc.c:89
static void noise_scale(int *coefs, int scale, int band_energy, int len)
Definition: aacdec_fixed.c:196
static int decode_pce(AVCodecContext *avctx, MPEG4AudioConfig *m4ac, uint8_t(*layout_map)[3], GetBitContext *gb, int byte_align_ref)
Decode program configuration element; reference: table 4.2.
static int init_get_bits(GetBitContext *s, const uint8_t *buffer, int bit_size)
Initialize GetBitContext.
Definition: get_bits.h:425
#define GET_CACHE(name, gb)
Definition: get_bits.h:197
static float * VMUL2(float *dst, const float *v, unsigned idx, const float *scale)
Definition: aacdec.c:83
OCStatus
Output configuration status.
Definition: aac.h:115
int skip_samples
Number of audio samples to skip at the start of the next decoded frame.
Definition: internal.h:187
#define MAX_CHANNELS
Definition: aac.h:47
N Error Resilient Bit-Sliced Arithmetic Coding.
Definition: mpeg4audio.h:89
#define TNS_MAX_ORDER
Definition: aac.h:50
#define FF_COMPLIANCE_STRICT
Strictly conform to all the things in the spec no matter what consequences.
Definition: avcodec.h:2983
main AAC context
Definition: aac.h:293
#define u(width,...)
const uint32_t ff_aac_scalefactor_code[121]
Definition: aactab.c:73
LongTermPrediction ltp
Definition: aac.h:180
ChannelCoupling coup
Definition: aac.h:286
Output configuration under trial specified by a frame header.
Definition: aac.h:118
int frame_length_short
Definition: mpeg4audio.h:41
static int decode_prediction(AACContext *ac, IndividualChannelStream *ics, GetBitContext *gb)
const uint8_t ff_tns_max_bands_128[]
Definition: aactab.c:1406
#define NOISE_OFFSET
subtracted from global gain, used as offset for the preamble
Definition: aac.h:158
static void imdct_and_window(TwinVQContext *tctx, enum TwinVQFrameType ftype, int wtype, float *in, float *prev, int ch)
Definition: twinvq.c:327
void av_frame_unref(AVFrame *frame)
Unreference all the buffers referenced by frame and reset the frame fields.
Definition: frame.c:505
void avpriv_report_missing_feature(void *avc, const char *msg,...) av_printf_format(2
Log a generic warning message about a missing feature.
static const int8_t filt[NUMTAPS]
Definition: af_earwax.c:39
int band_type_run_end[120]
band type run end points
Definition: aac.h:254
static int decode_band_types(AACContext *ac, enum BandType band_type[120], int band_type_run_end[120], GetBitContext *gb, IndividualChannelStream *ics)
Decode band types (section_data payload); reference: table 4.46.
#define AV_CH_BACK_CENTER
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:215
int band_top[17]
Indicates the top of the i-th DRC band in units of 4 spectral lines.
Definition: aac.h:218
#define AV_CH_SIDE_RIGHT
INTFLOAT coeffs[1024]
coefficients for IMDCT, maybe processed
Definition: aac.h:262
AVFixedDSPContext * avpriv_alloc_fixed_dsp(int bit_exact)
Allocate and initialize a fixed DSP context.
Definition: fixed_dsp.c:148
static VLC vlc_spectral[11]
enum OCStatus status
Definition: aac.h:129
INTFLOAT gain[16][120]
Definition: aac.h:242
Scalefactor data are intensity stereo positions (out of phase).
Definition: aac.h:88
N Error Resilient Enhanced Low Delay.
Definition: mpeg4audio.h:106
static int set_default_channel_config(AVCodecContext *avctx, uint8_t(*layout_map)[3], int *tags, int channel_config)
Set up channel positions based on a default channel configuration as specified in table 1...
#define M_SQRT2
Definition: mathematics.h:61
#define RANGE15(x)
Definition: aac_defines.h:97
INTFLOAT coef[8][4][TNS_MAX_ORDER]
Definition: aac.h:205
int16_t lag
Definition: aac.h:165
const uint8_t ff_aac_num_swb_120[]
Definition: aactab.c:65
DynamicRangeControl che_drc
Definition: aac.h:299
static av_always_inline void reset_predict_state(PredictorState *ps)
Definition: aacdec.c:72
AVFrame * frame
Definition: aac.h:296
OutputConfiguration oc[2]
Definition: aac.h:356
An AV_PKT_DATA_JP_DUALMONO side data packet indicates that the packet may contain "dual mono" audio s...
Definition: avcodec.h:1531
int
const uint8_t ff_aac_pred_sfb_max[]
Definition: aactab.c:69
int direction[8][4]
Definition: aac.h:202
void(* apply_ltp)(AACContext *ac, SingleChannelElement *sce)
Definition: aac.h:362
uint8_t prediction_used[41]
Definition: aac.h:190
const float ff_aac_eld_window_480[1800]
Definition: aactab.c:2378
INTFLOAT saved[1536]
overlap
Definition: aac.h:263
Single Channel Element - used for both SCE and LFE elements.
Definition: aac.h:248
#define ff_mdct_end
Definition: fft.h:170
const uint8_t ff_aac_num_swb_480[]
Definition: aactab.c:57
static double c[64]
const uint16_t *const ff_swb_offset_1024[]
Definition: aactab.c:1336
unsigned AAC_SIGNE
Definition: aac_defines.h:91
void(* windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out, INTFLOAT *in, IndividualChannelStream *ics)
Definition: aac.h:365
Definition: aac.h:61
Individual Channel Stream.
Definition: aac.h:174
INTFLOAT coef
Definition: aac.h:167
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:71
const uint16_t *const ff_aac_codebook_vector_idx[]
Definition: aactab.c:1083
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:777
av_cold void ff_mdct15_uninit(MDCT15Context **ps)
Definition: mdct15.c:43
channel element - generic struct for SCE/CPE/CCE/LFE
Definition: aac.h:275
void * priv_data
Definition: avcodec.h:1803
static int decode_ics(AACContext *ac, SingleChannelElement *sce, GetBitContext *gb, int common_window, int scale_flag)
Decode an individual_channel_stream payload; reference: table 4.44.
#define av_free(p)
#define FF_DEBUG_STARTCODE
Definition: avcodec.h:3017
const uint8_t ff_tns_max_bands_512[]
Definition: aactab.c:1398
int len
Scalefactors and spectral data are all zero.
Definition: aac.h:83
int channels
number of audio channels
Definition: avcodec.h:2524
int num_pulse
Definition: aac.h:225
static int * DEC_SPAIR(int *dst, unsigned idx)
Definition: aacdec_fixed.c:107
struct AVCodecInternal * internal
Private context used for internal data.
Definition: avcodec.h:1811
const uint8_t ff_mpeg4audio_channels[8]
Definition: mpeg4audio.c:67
static int ff_thread_once(char *control, void(*routine)(void))
Definition: thread.h:160
VLC_TYPE(* table)[2]
code, bits
Definition: vlc.h:28
Y Long Term Prediction.
Definition: mpeg4audio.h:74
uint8_t crc_absent
Definition: aacadtsdec.h:35
static const uint8_t * align_get_bits(GetBitContext *s)
Definition: get_bits.h:464
uint64_t layout
#define FF_PROFILE_AAC_HE
Definition: avcodec.h:3274
enum BandType band_type[128]
band types
Definition: aac.h:252
static int decode_dynamic_range(DynamicRangeControl *che_drc, GetBitContext *gb)
Decode dynamic range information; reference: table 4.52.
#define AV_CH_FRONT_RIGHT
#define POW_SF2_ZERO
ff_aac_pow2sf_tab index corresponding to pow(2, 0);
Definition: aac.h:154
static void imdct_and_windowing_ld(AACContext *ac, SingleChannelElement *sce)
FILE * out
Definition: movenc.c:54
FFTContext mdct
Definition: aac.h:323
int sbr
-1 implicit, 1 presence
Definition: mpeg4audio.h:34
#define av_freep(p)
void INT64 start
Definition: avisynth_c.h:690
#define av_always_inline
Definition: attributes.h:39
static void apply_mid_side_stereo(AACContext *ac, ChannelElement *cpe)
Mid/Side stereo decoding; reference: 4.6.8.1.3.
#define VLC_TYPE
Definition: vlc.h:24
#define AV_CH_SIDE_LEFT
#define FFSWAP(type, a, b)
Definition: common.h:99
int ps
-1 implicit, 1 presence
Definition: mpeg4audio.h:40
int8_t used[MAX_LTP_LONG_SFB]
Definition: aac.h:168
const uint16_t *const ff_swb_offset_128[]
Definition: aactab.c:1368
int8_t present
Definition: aac.h:164
uint32_t sample_rate
Definition: aacadtsdec.h:32
static const AVClass aac_decoder_class
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:248
uint64_t request_channel_layout
Request decoder to use this channel layout if it can (0 for default)
Definition: avcodec.h:2581
int layout_map_tags
Definition: aac.h:126
enum AVCodecID id
This structure stores compressed data.
Definition: avcodec.h:1656
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:267
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:2981
void AAC_RENAME() ff_init_ff_sine_windows(int index)
initialize the specified entry of ff_sine_windows
static int * DEC_UQUAD(int *dst, unsigned idx, unsigned sign)
Definition: aacdec_fixed.c:133
#define AV_CH_BACK_RIGHT
Y Low Complexity.
Definition: mpeg4audio.h:72
static float * VMUL4(float *dst, const float *v, unsigned idx, const float *scale)
Definition: aacdec.c:94
Output unconfigured.
Definition: aac.h:116
static const uint8_t aac_channel_layout_map[16][5][3]
Definition: aacdectab.h:40
RawDataBlockType
Definition: aac.h:55
static uint8_t tmp[11]
Definition: aes_ctr.c:26