FFmpeg  4.4.5
aacenc.c
Go to the documentation of this file.
1 /*
2  * AAC encoder
3  * Copyright (C) 2008 Konstantin Shishkov
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 /**
23  * @file
24  * AAC encoder
25  */
26 
27 /***********************************
28  * TODOs:
29  * add sane pulse detection
30  ***********************************/
31 #include <float.h>
32 
33 #include "libavutil/libm.h"
34 #include "libavutil/float_dsp.h"
35 #include "libavutil/opt.h"
36 #include "avcodec.h"
37 #include "put_bits.h"
38 #include "internal.h"
39 #include "mpeg4audio.h"
40 #include "sinewin.h"
41 #include "profiles.h"
42 
43 #include "aac.h"
44 #include "aactab.h"
45 #include "aacenc.h"
46 #include "aacenctab.h"
47 #include "aacenc_utils.h"
48 
49 #include "psymodel.h"
50 
51 static void put_pce(PutBitContext *pb, AVCodecContext *avctx)
52 {
53  int i, j;
54  AACEncContext *s = avctx->priv_data;
55  AACPCEInfo *pce = &s->pce;
56  const int bitexact = avctx->flags & AV_CODEC_FLAG_BITEXACT;
57  const char *aux_data = bitexact ? "Lavc" : LIBAVCODEC_IDENT;
58 
59  put_bits(pb, 4, 0);
60 
61  put_bits(pb, 2, avctx->profile);
62  put_bits(pb, 4, s->samplerate_index);
63 
64  put_bits(pb, 4, pce->num_ele[0]); /* Front */
65  put_bits(pb, 4, pce->num_ele[1]); /* Side */
66  put_bits(pb, 4, pce->num_ele[2]); /* Back */
67  put_bits(pb, 2, pce->num_ele[3]); /* LFE */
68  put_bits(pb, 3, 0); /* Assoc data */
69  put_bits(pb, 4, 0); /* CCs */
70 
71  put_bits(pb, 1, 0); /* Stereo mixdown */
72  put_bits(pb, 1, 0); /* Mono mixdown */
73  put_bits(pb, 1, 0); /* Something else */
74 
75  for (i = 0; i < 4; i++) {
76  for (j = 0; j < pce->num_ele[i]; j++) {
77  if (i < 3)
78  put_bits(pb, 1, pce->pairing[i][j]);
79  put_bits(pb, 4, pce->index[i][j]);
80  }
81  }
82 
83  align_put_bits(pb);
84  put_bits(pb, 8, strlen(aux_data));
85  ff_put_string(pb, aux_data, 0);
86 }
87 
88 /**
89  * Make AAC audio config object.
90  * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
91  */
93 {
94  PutBitContext pb;
95  AACEncContext *s = avctx->priv_data;
96  int channels = (!s->needs_pce)*(s->channels - (s->channels == 8 ? 1 : 0));
97  const int max_size = 32;
98 
99  avctx->extradata = av_mallocz(max_size);
100  if (!avctx->extradata)
101  return AVERROR(ENOMEM);
102 
103  init_put_bits(&pb, avctx->extradata, max_size);
104  put_bits(&pb, 5, s->profile+1); //profile
105  put_bits(&pb, 4, s->samplerate_index); //sample rate index
106  put_bits(&pb, 4, channels);
107  //GASpecificConfig
108  put_bits(&pb, 1, 0); //frame length - 1024 samples
109  put_bits(&pb, 1, 0); //does not depend on core coder
110  put_bits(&pb, 1, 0); //is not extension
111  if (s->needs_pce)
112  put_pce(&pb, avctx);
113 
114  //Explicitly Mark SBR absent
115  put_bits(&pb, 11, 0x2b7); //sync extension
116  put_bits(&pb, 5, AOT_SBR);
117  put_bits(&pb, 1, 0);
118  flush_put_bits(&pb);
119  avctx->extradata_size = put_bits_count(&pb) >> 3;
120 
121  return 0;
122 }
123 
125 {
126  ++s->quantize_band_cost_cache_generation;
127  if (s->quantize_band_cost_cache_generation == 0) {
128  memset(s->quantize_band_cost_cache, 0, sizeof(s->quantize_band_cost_cache));
129  s->quantize_band_cost_cache_generation = 1;
130  }
131 }
132 
133 #define WINDOW_FUNC(type) \
134 static void apply_ ##type ##_window(AVFloatDSPContext *fdsp, \
135  SingleChannelElement *sce, \
136  const float *audio)
137 
138 WINDOW_FUNC(only_long)
139 {
140  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
141  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
142  float *out = sce->ret_buf;
143 
144  fdsp->vector_fmul (out, audio, lwindow, 1024);
145  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, pwindow, 1024);
146 }
147 
148 WINDOW_FUNC(long_start)
149 {
150  const float *lwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
151  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
152  float *out = sce->ret_buf;
153 
154  fdsp->vector_fmul(out, audio, lwindow, 1024);
155  memcpy(out + 1024, audio + 1024, sizeof(out[0]) * 448);
156  fdsp->vector_fmul_reverse(out + 1024 + 448, audio + 1024 + 448, swindow, 128);
157  memset(out + 1024 + 576, 0, sizeof(out[0]) * 448);
158 }
159 
160 WINDOW_FUNC(long_stop)
161 {
162  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
163  const float *swindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
164  float *out = sce->ret_buf;
165 
166  memset(out, 0, sizeof(out[0]) * 448);
167  fdsp->vector_fmul(out + 448, audio + 448, swindow, 128);
168  memcpy(out + 576, audio + 576, sizeof(out[0]) * 448);
169  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, lwindow, 1024);
170 }
171 
172 WINDOW_FUNC(eight_short)
173 {
174  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
175  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
176  const float *in = audio + 448;
177  float *out = sce->ret_buf;
178  int w;
179 
180  for (w = 0; w < 8; w++) {
181  fdsp->vector_fmul (out, in, w ? pwindow : swindow, 128);
182  out += 128;
183  in += 128;
184  fdsp->vector_fmul_reverse(out, in, swindow, 128);
185  out += 128;
186  }
187 }
188 
189 static void (*const apply_window[4])(AVFloatDSPContext *fdsp,
191  const float *audio) = {
192  [ONLY_LONG_SEQUENCE] = apply_only_long_window,
193  [LONG_START_SEQUENCE] = apply_long_start_window,
194  [EIGHT_SHORT_SEQUENCE] = apply_eight_short_window,
195  [LONG_STOP_SEQUENCE] = apply_long_stop_window
196 };
197 
199  float *audio)
200 {
201  int i;
202  const float *output = sce->ret_buf;
203 
204  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, audio);
205 
207  s->mdct1024.mdct_calc(&s->mdct1024, sce->coeffs, output);
208  else
209  for (i = 0; i < 1024; i += 128)
210  s->mdct128.mdct_calc(&s->mdct128, &sce->coeffs[i], output + i*2);
211  memcpy(audio, audio + 1024, sizeof(audio[0]) * 1024);
212  memcpy(sce->pcoeffs, sce->coeffs, sizeof(sce->pcoeffs));
213 }
214 
215 /**
216  * Encode ics_info element.
217  * @see Table 4.6 (syntax of ics_info)
218  */
220 {
221  int w;
222 
223  put_bits(&s->pb, 1, 0); // ics_reserved bit
224  put_bits(&s->pb, 2, info->window_sequence[0]);
225  put_bits(&s->pb, 1, info->use_kb_window[0]);
226  if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
227  put_bits(&s->pb, 6, info->max_sfb);
228  put_bits(&s->pb, 1, !!info->predictor_present);
229  } else {
230  put_bits(&s->pb, 4, info->max_sfb);
231  for (w = 1; w < 8; w++)
232  put_bits(&s->pb, 1, !info->group_len[w]);
233  }
234 }
235 
236 /**
237  * Encode MS data.
238  * @see 4.6.8.1 "Joint Coding - M/S Stereo"
239  */
241 {
242  int i, w;
243 
244  put_bits(pb, 2, cpe->ms_mode);
245  if (cpe->ms_mode == 1)
246  for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
247  for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
248  put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
249 }
250 
251 /**
252  * Produce integer coefficients from scalefactors provided by the model.
253  */
254 static void adjust_frame_information(ChannelElement *cpe, int chans)
255 {
256  int i, w, w2, g, ch;
257  int maxsfb, cmaxsfb;
258 
259  for (ch = 0; ch < chans; ch++) {
260  IndividualChannelStream *ics = &cpe->ch[ch].ics;
261  maxsfb = 0;
262  cpe->ch[ch].pulse.num_pulse = 0;
263  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
264  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
265  for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w*16+cmaxsfb-1]; cmaxsfb--)
266  ;
267  maxsfb = FFMAX(maxsfb, cmaxsfb);
268  }
269  }
270  ics->max_sfb = maxsfb;
271 
272  //adjust zero bands for window groups
273  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
274  for (g = 0; g < ics->max_sfb; g++) {
275  i = 1;
276  for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
277  if (!cpe->ch[ch].zeroes[w2*16 + g]) {
278  i = 0;
279  break;
280  }
281  }
282  cpe->ch[ch].zeroes[w*16 + g] = i;
283  }
284  }
285  }
286 
287  if (chans > 1 && cpe->common_window) {
288  IndividualChannelStream *ics0 = &cpe->ch[0].ics;
289  IndividualChannelStream *ics1 = &cpe->ch[1].ics;
290  int msc = 0;
291  ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
292  ics1->max_sfb = ics0->max_sfb;
293  for (w = 0; w < ics0->num_windows*16; w += 16)
294  for (i = 0; i < ics0->max_sfb; i++)
295  if (cpe->ms_mask[w+i])
296  msc++;
297  if (msc == 0 || ics0->max_sfb == 0)
298  cpe->ms_mode = 0;
299  else
300  cpe->ms_mode = msc < ics0->max_sfb * ics0->num_windows ? 1 : 2;
301  }
302 }
303 
305 {
306  int w, w2, g, i;
307  IndividualChannelStream *ics = &cpe->ch[0].ics;
308  if (!cpe->common_window)
309  return;
310  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
311  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
312  int start = (w+w2) * 128;
313  for (g = 0; g < ics->num_swb; g++) {
314  int p = -1 + 2 * (cpe->ch[1].band_type[w*16+g] - 14);
315  float scale = cpe->ch[0].is_ener[w*16+g];
316  if (!cpe->is_mask[w*16 + g]) {
317  start += ics->swb_sizes[g];
318  continue;
319  }
320  if (cpe->ms_mask[w*16 + g])
321  p *= -1;
322  for (i = 0; i < ics->swb_sizes[g]; i++) {
323  float sum = (cpe->ch[0].coeffs[start+i] + p*cpe->ch[1].coeffs[start+i])*scale;
324  cpe->ch[0].coeffs[start+i] = sum;
325  cpe->ch[1].coeffs[start+i] = 0.0f;
326  }
327  start += ics->swb_sizes[g];
328  }
329  }
330  }
331 }
332 
334 {
335  int w, w2, g, i;
336  IndividualChannelStream *ics = &cpe->ch[0].ics;
337  if (!cpe->common_window)
338  return;
339  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
340  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
341  int start = (w+w2) * 128;
342  for (g = 0; g < ics->num_swb; g++) {
343  /* ms_mask can be used for other purposes in PNS and I/S,
344  * so must not apply M/S if any band uses either, even if
345  * ms_mask is set.
346  */
347  if (!cpe->ms_mask[w*16 + g] || cpe->is_mask[w*16 + g]
348  || cpe->ch[0].band_type[w*16 + g] >= NOISE_BT
349  || cpe->ch[1].band_type[w*16 + g] >= NOISE_BT) {
350  start += ics->swb_sizes[g];
351  continue;
352  }
353  for (i = 0; i < ics->swb_sizes[g]; i++) {
354  float L = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) * 0.5f;
355  float R = L - cpe->ch[1].coeffs[start+i];
356  cpe->ch[0].coeffs[start+i] = L;
357  cpe->ch[1].coeffs[start+i] = R;
358  }
359  start += ics->swb_sizes[g];
360  }
361  }
362  }
363 }
364 
365 /**
366  * Encode scalefactor band coding type.
367  */
369 {
370  int w;
371 
372  if (s->coder->set_special_band_scalefactors)
373  s->coder->set_special_band_scalefactors(s, sce);
374 
375  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
376  s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
377 }
378 
379 /**
380  * Encode scalefactors.
381  */
384 {
385  int diff, off_sf = sce->sf_idx[0], off_pns = sce->sf_idx[0] - NOISE_OFFSET;
386  int off_is = 0, noise_flag = 1;
387  int i, w;
388 
389  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
390  for (i = 0; i < sce->ics.max_sfb; i++) {
391  if (!sce->zeroes[w*16 + i]) {
392  if (sce->band_type[w*16 + i] == NOISE_BT) {
393  diff = sce->sf_idx[w*16 + i] - off_pns;
394  off_pns = sce->sf_idx[w*16 + i];
395  if (noise_flag-- > 0) {
397  continue;
398  }
399  } else if (sce->band_type[w*16 + i] == INTENSITY_BT ||
400  sce->band_type[w*16 + i] == INTENSITY_BT2) {
401  diff = sce->sf_idx[w*16 + i] - off_is;
402  off_is = sce->sf_idx[w*16 + i];
403  } else {
404  diff = sce->sf_idx[w*16 + i] - off_sf;
405  off_sf = sce->sf_idx[w*16 + i];
406  }
408  av_assert0(diff >= 0 && diff <= 120);
410  }
411  }
412  }
413 }
414 
415 /**
416  * Encode pulse data.
417  */
418 static void encode_pulses(AACEncContext *s, Pulse *pulse)
419 {
420  int i;
421 
422  put_bits(&s->pb, 1, !!pulse->num_pulse);
423  if (!pulse->num_pulse)
424  return;
425 
426  put_bits(&s->pb, 2, pulse->num_pulse - 1);
427  put_bits(&s->pb, 6, pulse->start);
428  for (i = 0; i < pulse->num_pulse; i++) {
429  put_bits(&s->pb, 5, pulse->pos[i]);
430  put_bits(&s->pb, 4, pulse->amp[i]);
431  }
432 }
433 
434 /**
435  * Encode spectral coefficients processed by psychoacoustic model.
436  */
438 {
439  int start, i, w, w2;
440 
441  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
442  start = 0;
443  for (i = 0; i < sce->ics.max_sfb; i++) {
444  if (sce->zeroes[w*16 + i]) {
445  start += sce->ics.swb_sizes[i];
446  continue;
447  }
448  for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
449  s->coder->quantize_and_encode_band(s, &s->pb,
450  &sce->coeffs[start + w2*128],
451  NULL, sce->ics.swb_sizes[i],
452  sce->sf_idx[w*16 + i],
453  sce->band_type[w*16 + i],
454  s->lambda,
455  sce->ics.window_clipping[w]);
456  }
457  start += sce->ics.swb_sizes[i];
458  }
459  }
460 }
461 
462 /**
463  * Downscale spectral coefficients for near-clipping windows to avoid artifacts
464  */
466 {
467  int start, i, j, w;
468 
469  if (sce->ics.clip_avoidance_factor < 1.0f) {
470  for (w = 0; w < sce->ics.num_windows; w++) {
471  start = 0;
472  for (i = 0; i < sce->ics.max_sfb; i++) {
473  float *swb_coeffs = &sce->coeffs[start + w*128];
474  for (j = 0; j < sce->ics.swb_sizes[i]; j++)
475  swb_coeffs[j] *= sce->ics.clip_avoidance_factor;
476  start += sce->ics.swb_sizes[i];
477  }
478  }
479  }
480 }
481 
482 /**
483  * Encode one channel of audio data.
484  */
487  int common_window)
488 {
489  put_bits(&s->pb, 8, sce->sf_idx[0]);
490  if (!common_window) {
491  put_ics_info(s, &sce->ics);
492  if (s->coder->encode_main_pred)
493  s->coder->encode_main_pred(s, sce);
494  if (s->coder->encode_ltp_info)
495  s->coder->encode_ltp_info(s, sce, 0);
496  }
497  encode_band_info(s, sce);
498  encode_scale_factors(avctx, s, sce);
499  encode_pulses(s, &sce->pulse);
500  put_bits(&s->pb, 1, !!sce->tns.present);
501  if (s->coder->encode_tns_info)
502  s->coder->encode_tns_info(s, sce);
503  put_bits(&s->pb, 1, 0); //ssr
505  return 0;
506 }
507 
508 /**
509  * Write some auxiliary information about the created AAC file.
510  */
511 static void put_bitstream_info(AACEncContext *s, const char *name)
512 {
513  int i, namelen, padbits;
514 
515  namelen = strlen(name) + 2;
516  put_bits(&s->pb, 3, TYPE_FIL);
517  put_bits(&s->pb, 4, FFMIN(namelen, 15));
518  if (namelen >= 15)
519  put_bits(&s->pb, 8, namelen - 14);
520  put_bits(&s->pb, 4, 0); //extension type - filler
521  padbits = -put_bits_count(&s->pb) & 7;
522  align_put_bits(&s->pb);
523  for (i = 0; i < namelen - 2; i++)
524  put_bits(&s->pb, 8, name[i]);
525  put_bits(&s->pb, 12 - padbits, 0);
526 }
527 
528 /*
529  * Copy input samples.
530  * Channels are reordered from libavcodec's default order to AAC order.
531  */
533 {
534  int ch;
535  int end = 2048 + (frame ? frame->nb_samples : 0);
536  const uint8_t *channel_map = s->reorder_map;
537 
538  /* copy and remap input samples */
539  for (ch = 0; ch < s->channels; ch++) {
540  /* copy last 1024 samples of previous frame to the start of the current frame */
541  memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0]));
542 
543  /* copy new samples and zero any remaining samples */
544  if (frame) {
545  memcpy(&s->planar_samples[ch][2048],
546  frame->extended_data[channel_map[ch]],
547  frame->nb_samples * sizeof(s->planar_samples[0][0]));
548  }
549  memset(&s->planar_samples[ch][end], 0,
550  (3072 - end) * sizeof(s->planar_samples[0][0]));
551  }
552 }
553 
554 static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
555  const AVFrame *frame, int *got_packet_ptr)
556 {
557  AACEncContext *s = avctx->priv_data;
558  float **samples = s->planar_samples, *samples2, *la, *overlap;
559  ChannelElement *cpe;
562  int i, its, ch, w, chans, tag, start_ch, ret, frame_bits;
563  int target_bits, rate_bits, too_many_bits, too_few_bits;
564  int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
565  int chan_el_counter[4];
567 
568  /* add current frame to queue */
569  if (frame) {
570  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
571  return ret;
572  } else {
573  if (!s->afq.remaining_samples || (!s->afq.frame_alloc && !s->afq.frame_count))
574  return 0;
575  }
576 
578  if (s->psypp)
579  ff_psy_preprocess(s->psypp, s->planar_samples, s->channels);
580 
581  if (!avctx->frame_number)
582  return 0;
583 
584  start_ch = 0;
585  for (i = 0; i < s->chan_map[0]; i++) {
586  FFPsyWindowInfo* wi = windows + start_ch;
587  tag = s->chan_map[i+1];
588  chans = tag == TYPE_CPE ? 2 : 1;
589  cpe = &s->cpe[i];
590  for (ch = 0; ch < chans; ch++) {
591  int k;
592  float clip_avoidance_factor;
593  sce = &cpe->ch[ch];
594  ics = &sce->ics;
595  s->cur_channel = start_ch + ch;
596  overlap = &samples[s->cur_channel][0];
597  samples2 = overlap + 1024;
598  la = samples2 + (448+64);
599  if (!frame)
600  la = NULL;
601  if (tag == TYPE_LFE) {
602  wi[ch].window_type[0] = wi[ch].window_type[1] = ONLY_LONG_SEQUENCE;
603  wi[ch].window_shape = 0;
604  wi[ch].num_windows = 1;
605  wi[ch].grouping[0] = 1;
606  wi[ch].clipping[0] = 0;
607 
608  /* Only the lowest 12 coefficients are used in a LFE channel.
609  * The expression below results in only the bottom 8 coefficients
610  * being used for 11.025kHz to 16kHz sample rates.
611  */
612  ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
613  } else {
614  wi[ch] = s->psy.model->window(&s->psy, samples2, la, s->cur_channel,
615  ics->window_sequence[0]);
616  }
617  ics->window_sequence[1] = ics->window_sequence[0];
618  ics->window_sequence[0] = wi[ch].window_type[0];
619  ics->use_kb_window[1] = ics->use_kb_window[0];
620  ics->use_kb_window[0] = wi[ch].window_shape;
621  ics->num_windows = wi[ch].num_windows;
622  ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
623  ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
624  ics->max_sfb = FFMIN(ics->max_sfb, ics->num_swb);
625  ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
626  ff_swb_offset_128 [s->samplerate_index]:
627  ff_swb_offset_1024[s->samplerate_index];
628  ics->tns_max_bands = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
629  ff_tns_max_bands_128 [s->samplerate_index]:
630  ff_tns_max_bands_1024[s->samplerate_index];
631 
632  for (w = 0; w < ics->num_windows; w++)
633  ics->group_len[w] = wi[ch].grouping[w];
634 
635  /* Calculate input sample maximums and evaluate clipping risk */
636  clip_avoidance_factor = 0.0f;
637  for (w = 0; w < ics->num_windows; w++) {
638  const float *wbuf = overlap + w * 128;
639  const int wlen = 2048 / ics->num_windows;
640  float max = 0;
641  int j;
642  /* mdct input is 2 * output */
643  for (j = 0; j < wlen; j++)
644  max = FFMAX(max, fabsf(wbuf[j]));
645  wi[ch].clipping[w] = max;
646  }
647  for (w = 0; w < ics->num_windows; w++) {
648  if (wi[ch].clipping[w] > CLIP_AVOIDANCE_FACTOR) {
649  ics->window_clipping[w] = 1;
650  clip_avoidance_factor = FFMAX(clip_avoidance_factor, wi[ch].clipping[w]);
651  } else {
652  ics->window_clipping[w] = 0;
653  }
654  }
655  if (clip_avoidance_factor > CLIP_AVOIDANCE_FACTOR) {
656  ics->clip_avoidance_factor = CLIP_AVOIDANCE_FACTOR / clip_avoidance_factor;
657  } else {
658  ics->clip_avoidance_factor = 1.0f;
659  }
660 
661  apply_window_and_mdct(s, sce, overlap);
662 
663  if (s->options.ltp && s->coder->update_ltp) {
664  s->coder->update_ltp(s, sce);
665  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
666  s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
667  }
668 
669  for (k = 0; k < 1024; k++) {
670  if (!(fabs(cpe->ch[ch].coeffs[k]) < 1E16)) { // Ensure headroom for energy calculation
671  av_log(avctx, AV_LOG_ERROR, "Input contains (near) NaN/+-Inf\n");
672  return AVERROR(EINVAL);
673  }
674  }
675  avoid_clipping(s, sce);
676  }
677  start_ch += chans;
678  }
679  if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels, 0)) < 0)
680  return ret;
681  frame_bits = its = 0;
682  do {
683  init_put_bits(&s->pb, avpkt->data, avpkt->size);
684 
685  if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & AV_CODEC_FLAG_BITEXACT))
687  start_ch = 0;
688  target_bits = 0;
689  memset(chan_el_counter, 0, sizeof(chan_el_counter));
690  for (i = 0; i < s->chan_map[0]; i++) {
691  FFPsyWindowInfo* wi = windows + start_ch;
692  const float *coeffs[2];
693  tag = s->chan_map[i+1];
694  chans = tag == TYPE_CPE ? 2 : 1;
695  cpe = &s->cpe[i];
696  cpe->common_window = 0;
697  memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
698  memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
699  put_bits(&s->pb, 3, tag);
700  put_bits(&s->pb, 4, chan_el_counter[tag]++);
701  for (ch = 0; ch < chans; ch++) {
702  sce = &cpe->ch[ch];
703  coeffs[ch] = sce->coeffs;
704  sce->ics.predictor_present = 0;
705  sce->ics.ltp.present = 0;
706  memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
707  memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
708  memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
709  for (w = 0; w < 128; w++)
710  if (sce->band_type[w] > RESERVED_BT)
711  sce->band_type[w] = 0;
712  }
713  s->psy.bitres.alloc = -1;
714  s->psy.bitres.bits = s->last_frame_pb_count / s->channels;
715  s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
716  if (s->psy.bitres.alloc > 0) {
717  /* Lambda unused here on purpose, we need to take psy's unscaled allocation */
718  target_bits += s->psy.bitres.alloc
719  * (s->lambda / (avctx->global_quality ? avctx->global_quality : 120));
720  s->psy.bitres.alloc /= chans;
721  }
722  s->cur_type = tag;
723  for (ch = 0; ch < chans; ch++) {
724  s->cur_channel = start_ch + ch;
725  if (s->options.pns && s->coder->mark_pns)
726  s->coder->mark_pns(s, avctx, &cpe->ch[ch]);
727  s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
728  }
729  if (chans > 1
730  && wi[0].window_type[0] == wi[1].window_type[0]
731  && wi[0].window_shape == wi[1].window_shape) {
732 
733  cpe->common_window = 1;
734  for (w = 0; w < wi[0].num_windows; w++) {
735  if (wi[0].grouping[w] != wi[1].grouping[w]) {
736  cpe->common_window = 0;
737  break;
738  }
739  }
740  }
741  for (ch = 0; ch < chans; ch++) { /* TNS and PNS */
742  sce = &cpe->ch[ch];
743  s->cur_channel = start_ch + ch;
744  if (s->options.tns && s->coder->search_for_tns)
745  s->coder->search_for_tns(s, sce);
746  if (s->options.tns && s->coder->apply_tns_filt)
747  s->coder->apply_tns_filt(s, sce);
748  if (sce->tns.present)
749  tns_mode = 1;
750  if (s->options.pns && s->coder->search_for_pns)
751  s->coder->search_for_pns(s, avctx, sce);
752  }
753  s->cur_channel = start_ch;
754  if (s->options.intensity_stereo) { /* Intensity Stereo */
755  if (s->coder->search_for_is)
756  s->coder->search_for_is(s, avctx, cpe);
757  if (cpe->is_mode) is_mode = 1;
759  }
760  if (s->options.pred) { /* Prediction */
761  for (ch = 0; ch < chans; ch++) {
762  sce = &cpe->ch[ch];
763  s->cur_channel = start_ch + ch;
764  if (s->options.pred && s->coder->search_for_pred)
765  s->coder->search_for_pred(s, sce);
766  if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
767  }
768  if (s->coder->adjust_common_pred)
769  s->coder->adjust_common_pred(s, cpe);
770  for (ch = 0; ch < chans; ch++) {
771  sce = &cpe->ch[ch];
772  s->cur_channel = start_ch + ch;
773  if (s->options.pred && s->coder->apply_main_pred)
774  s->coder->apply_main_pred(s, sce);
775  }
776  s->cur_channel = start_ch;
777  }
778  if (s->options.mid_side) { /* Mid/Side stereo */
779  if (s->options.mid_side == -1 && s->coder->search_for_ms)
780  s->coder->search_for_ms(s, cpe);
781  else if (cpe->common_window)
782  memset(cpe->ms_mask, 1, sizeof(cpe->ms_mask));
784  }
785  adjust_frame_information(cpe, chans);
786  if (s->options.ltp) { /* LTP */
787  for (ch = 0; ch < chans; ch++) {
788  sce = &cpe->ch[ch];
789  s->cur_channel = start_ch + ch;
790  if (s->coder->search_for_ltp)
791  s->coder->search_for_ltp(s, sce, cpe->common_window);
792  if (sce->ics.ltp.present) pred_mode = 1;
793  }
794  s->cur_channel = start_ch;
795  if (s->coder->adjust_common_ltp)
796  s->coder->adjust_common_ltp(s, cpe);
797  }
798  if (chans == 2) {
799  put_bits(&s->pb, 1, cpe->common_window);
800  if (cpe->common_window) {
801  put_ics_info(s, &cpe->ch[0].ics);
802  if (s->coder->encode_main_pred)
803  s->coder->encode_main_pred(s, &cpe->ch[0]);
804  if (s->coder->encode_ltp_info)
805  s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
806  encode_ms_info(&s->pb, cpe);
807  if (cpe->ms_mode) ms_mode = 1;
808  }
809  }
810  for (ch = 0; ch < chans; ch++) {
811  s->cur_channel = start_ch + ch;
812  encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
813  }
814  start_ch += chans;
815  }
816 
817  if (avctx->flags & AV_CODEC_FLAG_QSCALE) {
818  /* When using a constant Q-scale, don't mess with lambda */
819  break;
820  }
821 
822  /* rate control stuff
823  * allow between the nominal bitrate, and what psy's bit reservoir says to target
824  * but drift towards the nominal bitrate always
825  */
826  frame_bits = put_bits_count(&s->pb);
827  rate_bits = avctx->bit_rate * 1024 / avctx->sample_rate;
828  rate_bits = FFMIN(rate_bits, 6144 * s->channels - 3);
829  too_many_bits = FFMAX(target_bits, rate_bits);
830  too_many_bits = FFMIN(too_many_bits, 6144 * s->channels - 3);
831  too_few_bits = FFMIN(FFMAX(rate_bits - rate_bits/4, target_bits), too_many_bits);
832 
833  /* When using ABR, be strict (but only for increasing) */
834  too_few_bits = too_few_bits - too_few_bits/8;
835  too_many_bits = too_many_bits + too_many_bits/2;
836 
837  if ( its == 0 /* for steady-state Q-scale tracking */
838  || (its < 5 && (frame_bits < too_few_bits || frame_bits > too_many_bits))
839  || frame_bits >= 6144 * s->channels - 3 )
840  {
841  float ratio = ((float)rate_bits) / frame_bits;
842 
843  if (frame_bits >= too_few_bits && frame_bits <= too_many_bits) {
844  /*
845  * This path is for steady-state Q-scale tracking
846  * When frame bits fall within the stable range, we still need to adjust
847  * lambda to maintain it like so in a stable fashion (large jumps in lambda
848  * create artifacts and should be avoided), but slowly
849  */
850  ratio = sqrtf(sqrtf(ratio));
851  ratio = av_clipf(ratio, 0.9f, 1.1f);
852  } else {
853  /* Not so fast though */
854  ratio = sqrtf(ratio);
855  }
856  s->lambda = av_clipf(s->lambda * ratio, FLT_EPSILON, 65536.f);
857 
858  /* Keep iterating if we must reduce and lambda is in the sky */
859  if (ratio > 0.9f && ratio < 1.1f) {
860  break;
861  } else {
862  if (is_mode || ms_mode || tns_mode || pred_mode) {
863  for (i = 0; i < s->chan_map[0]; i++) {
864  // Must restore coeffs
865  chans = tag == TYPE_CPE ? 2 : 1;
866  cpe = &s->cpe[i];
867  for (ch = 0; ch < chans; ch++)
868  memcpy(cpe->ch[ch].coeffs, cpe->ch[ch].pcoeffs, sizeof(cpe->ch[ch].coeffs));
869  }
870  }
871  its++;
872  }
873  } else {
874  break;
875  }
876  } while (1);
877 
878  if (s->options.ltp && s->coder->ltp_insert_new_frame)
879  s->coder->ltp_insert_new_frame(s);
880 
881  put_bits(&s->pb, 3, TYPE_END);
882  flush_put_bits(&s->pb);
883 
884  s->last_frame_pb_count = put_bits_count(&s->pb);
885 
886  s->lambda_sum += s->lambda;
887  s->lambda_count++;
888 
889  ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
890  &avpkt->duration);
891 
892  avpkt->size = put_bits_count(&s->pb) >> 3;
893  *got_packet_ptr = 1;
894  return 0;
895 }
896 
898 {
899  AACEncContext *s = avctx->priv_data;
900 
901  av_log(avctx, AV_LOG_INFO, "Qavg: %.3f\n", s->lambda_count ? s->lambda_sum / s->lambda_count : NAN);
902 
903  ff_mdct_end(&s->mdct1024);
904  ff_mdct_end(&s->mdct128);
905  ff_psy_end(&s->psy);
906  ff_lpc_end(&s->lpc);
907  if (s->psypp)
908  ff_psy_preprocess_end(s->psypp);
909  av_freep(&s->buffer.samples);
910  av_freep(&s->cpe);
911  av_freep(&s->fdsp);
912  ff_af_queue_close(&s->afq);
913  return 0;
914 }
915 
917 {
918  int ret = 0;
919 
921  if (!s->fdsp)
922  return AVERROR(ENOMEM);
923 
924  // window init
926 
927  if ((ret = ff_mdct_init(&s->mdct1024, 11, 0, 32768.0)) < 0)
928  return ret;
929  if ((ret = ff_mdct_init(&s->mdct128, 8, 0, 32768.0)) < 0)
930  return ret;
931 
932  return 0;
933 }
934 
936 {
937  int ch;
938  if (!FF_ALLOCZ_TYPED_ARRAY(s->buffer.samples, s->channels * 3 * 1024) ||
939  !FF_ALLOCZ_TYPED_ARRAY(s->cpe, s->chan_map[0]))
940  return AVERROR(ENOMEM);
941 
942  for(ch = 0; ch < s->channels; ch++)
943  s->planar_samples[ch] = s->buffer.samples + 3 * 1024 * ch;
944 
945  return 0;
946 }
947 
949 {
950  AACEncContext *s = avctx->priv_data;
951  int i, ret = 0;
952  const uint8_t *sizes[2];
953  uint8_t grouping[AAC_MAX_CHANNELS];
954  int lengths[2];
955 
956  /* Constants */
957  s->last_frame_pb_count = 0;
958  avctx->frame_size = 1024;
959  avctx->initial_padding = 1024;
960  s->lambda = avctx->global_quality > 0 ? avctx->global_quality : 120;
961 
962  /* Channel map and unspecified bitrate guessing */
963  s->channels = avctx->channels;
964 
965  s->needs_pce = 1;
966  for (i = 0; i < FF_ARRAY_ELEMS(aac_normal_chan_layouts); i++) {
967  if (avctx->channel_layout == aac_normal_chan_layouts[i]) {
968  s->needs_pce = s->options.pce;
969  break;
970  }
971  }
972 
973  if (s->needs_pce) {
974  char buf[64];
975  for (i = 0; i < FF_ARRAY_ELEMS(aac_pce_configs); i++)
976  if (avctx->channel_layout == aac_pce_configs[i].layout)
977  break;
978  av_get_channel_layout_string(buf, sizeof(buf), -1, avctx->channel_layout);
979  ERROR_IF(i == FF_ARRAY_ELEMS(aac_pce_configs), "Unsupported channel layout \"%s\"\n", buf);
980  av_log(avctx, AV_LOG_INFO, "Using a PCE to encode channel layout \"%s\"\n", buf);
981  s->pce = aac_pce_configs[i];
982  s->reorder_map = s->pce.reorder_map;
983  s->chan_map = s->pce.config_map;
984  } else {
985  s->reorder_map = aac_chan_maps[s->channels - 1];
986  s->chan_map = aac_chan_configs[s->channels - 1];
987  }
988 
989  if (!avctx->bit_rate) {
990  for (i = 1; i <= s->chan_map[0]; i++) {
991  avctx->bit_rate += s->chan_map[i] == TYPE_CPE ? 128000 : /* Pair */
992  s->chan_map[i] == TYPE_LFE ? 16000 : /* LFE */
993  69000 ; /* SCE */
994  }
995  }
996 
997  /* Samplerate */
998  for (i = 0; i < 16; i++)
1000  break;
1001  s->samplerate_index = i;
1002  ERROR_IF(s->samplerate_index == 16 ||
1003  s->samplerate_index >= ff_aac_swb_size_1024_len ||
1004  s->samplerate_index >= ff_aac_swb_size_128_len,
1005  "Unsupported sample rate %d\n", avctx->sample_rate);
1006 
1007  /* Bitrate limiting */
1008  WARN_IF(1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * s->channels,
1009  "Too many bits %f > %d per frame requested, clamping to max\n",
1010  1024.0 * avctx->bit_rate / avctx->sample_rate,
1011  6144 * s->channels);
1012  avctx->bit_rate = (int64_t)FFMIN(6144 * s->channels / 1024.0 * avctx->sample_rate,
1013  avctx->bit_rate);
1014 
1015  /* Profile and option setting */
1016  avctx->profile = avctx->profile == FF_PROFILE_UNKNOWN ? FF_PROFILE_AAC_LOW :
1017  avctx->profile;
1018  for (i = 0; i < FF_ARRAY_ELEMS(aacenc_profiles); i++)
1019  if (avctx->profile == aacenc_profiles[i])
1020  break;
1021  if (avctx->profile == FF_PROFILE_MPEG2_AAC_LOW) {
1022  avctx->profile = FF_PROFILE_AAC_LOW;
1023  ERROR_IF(s->options.pred,
1024  "Main prediction unavailable in the \"mpeg2_aac_low\" profile\n");
1025  ERROR_IF(s->options.ltp,
1026  "LTP prediction unavailable in the \"mpeg2_aac_low\" profile\n");
1027  WARN_IF(s->options.pns,
1028  "PNS unavailable in the \"mpeg2_aac_low\" profile, turning off\n");
1029  s->options.pns = 0;
1030  } else if (avctx->profile == FF_PROFILE_AAC_LTP) {
1031  s->options.ltp = 1;
1032  ERROR_IF(s->options.pred,
1033  "Main prediction unavailable in the \"aac_ltp\" profile\n");
1034  } else if (avctx->profile == FF_PROFILE_AAC_MAIN) {
1035  s->options.pred = 1;
1036  ERROR_IF(s->options.ltp,
1037  "LTP prediction unavailable in the \"aac_main\" profile\n");
1038  } else if (s->options.ltp) {
1039  avctx->profile = FF_PROFILE_AAC_LTP;
1040  WARN_IF(1,
1041  "Chainging profile to \"aac_ltp\"\n");
1042  ERROR_IF(s->options.pred,
1043  "Main prediction unavailable in the \"aac_ltp\" profile\n");
1044  } else if (s->options.pred) {
1045  avctx->profile = FF_PROFILE_AAC_MAIN;
1046  WARN_IF(1,
1047  "Chainging profile to \"aac_main\"\n");
1048  ERROR_IF(s->options.ltp,
1049  "LTP prediction unavailable in the \"aac_main\" profile\n");
1050  }
1051  s->profile = avctx->profile;
1052 
1053  /* Coder limitations */
1054  s->coder = &ff_aac_coders[s->options.coder];
1055  if (s->options.coder == AAC_CODER_ANMR) {
1057  "The ANMR coder is considered experimental, add -strict -2 to enable!\n");
1058  s->options.intensity_stereo = 0;
1059  s->options.pns = 0;
1060  }
1062  "The LPT profile requires experimental compliance, add -strict -2 to enable!\n");
1063 
1064  /* M/S introduces horrible artifacts with multichannel files, this is temporary */
1065  if (s->channels > 3)
1066  s->options.mid_side = 0;
1067 
1068  if ((ret = dsp_init(avctx, s)) < 0)
1069  return ret;
1070 
1071  if ((ret = alloc_buffers(avctx, s)) < 0)
1072  return ret;
1073 
1074  if ((ret = put_audio_specific_config(avctx)))
1075  return ret;
1076 
1077  sizes[0] = ff_aac_swb_size_1024[s->samplerate_index];
1078  sizes[1] = ff_aac_swb_size_128[s->samplerate_index];
1079  lengths[0] = ff_aac_num_swb_1024[s->samplerate_index];
1080  lengths[1] = ff_aac_num_swb_128[s->samplerate_index];
1081  for (i = 0; i < s->chan_map[0]; i++)
1082  grouping[i] = s->chan_map[i + 1] == TYPE_CPE;
1083  if ((ret = ff_psy_init(&s->psy, avctx, 2, sizes, lengths,
1084  s->chan_map[0], grouping)) < 0)
1085  return ret;
1086  s->psypp = ff_psy_preprocess_init(avctx);
1088  s->random_state = 0x1f2e3d4c;
1089 
1090  s->abs_pow34 = abs_pow34_v;
1091  s->quant_bands = quantize_bands;
1092 
1093  if (ARCH_X86)
1095 
1096  if (HAVE_MIPSDSP)
1098 
1099  ff_af_queue_init(avctx, &s->afq);
1100  ff_aac_tableinit();
1101 
1102  return 0;
1103 }
1104 
1105 #define AACENC_FLAGS AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
1106 static const AVOption aacenc_options[] = {
1107  {"aac_coder", "Coding algorithm", offsetof(AACEncContext, options.coder), AV_OPT_TYPE_INT, {.i64 = AAC_CODER_FAST}, 0, AAC_CODER_NB-1, AACENC_FLAGS, "coder"},
1108  {"anmr", "ANMR method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_ANMR}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1109  {"twoloop", "Two loop searching method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_TWOLOOP}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1110  {"fast", "Default fast search", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_FAST}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1111  {"aac_ms", "Force M/S stereo coding", offsetof(AACEncContext, options.mid_side), AV_OPT_TYPE_BOOL, {.i64 = -1}, -1, 1, AACENC_FLAGS},
1112  {"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1113  {"aac_pns", "Perceptual noise substitution", offsetof(AACEncContext, options.pns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1114  {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1115  {"aac_ltp", "Long term prediction", offsetof(AACEncContext, options.ltp), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1116  {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1117  {"aac_pce", "Forces the use of PCEs", offsetof(AACEncContext, options.pce), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1119  {NULL}
1120 };
1121 
1122 static const AVClass aacenc_class = {
1123  .class_name = "AAC encoder",
1124  .item_name = av_default_item_name,
1125  .option = aacenc_options,
1126  .version = LIBAVUTIL_VERSION_INT,
1127 };
1128 
1130  { "b", "0" },
1131  { NULL }
1132 };
1133 
1135  .name = "aac",
1136  .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
1137  .type = AVMEDIA_TYPE_AUDIO,
1138  .id = AV_CODEC_ID_AAC,
1139  .priv_data_size = sizeof(AACEncContext),
1140  .init = aac_encode_init,
1141  .encode2 = aac_encode_frame,
1142  .close = aac_encode_end,
1144  .supported_samplerates = mpeg4audio_sample_rates,
1147  .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
1149  .priv_class = &aacenc_class,
1150 };
@ EIGHT_SHORT_SEQUENCE
Definition: aac.h:79
@ LONG_STOP_SEQUENCE
Definition: aac.h:80
@ ONLY_LONG_SEQUENCE
Definition: aac.h:77
@ LONG_START_SEQUENCE
Definition: aac.h:78
#define NOISE_PRE
preamble for NOISE_BT, put in bitstream with the first noise band
Definition: aac.h:157
@ INTENSITY_BT
Scalefactor data are intensity stereo positions (in phase).
Definition: aac.h:90
@ INTENSITY_BT2
Scalefactor data are intensity stereo positions (out of phase).
Definition: aac.h:89
@ RESERVED_BT
Band types following are encoded differently from others.
Definition: aac.h:87
@ NOISE_BT
Spectral data are scaled white noise not coded in the bitstream.
Definition: aac.h:88
@ TYPE_CPE
Definition: aac.h:58
@ TYPE_FIL
Definition: aac.h:63
@ TYPE_LFE
Definition: aac.h:60
@ TYPE_END
Definition: aac.h:64
#define CLIP_AVOIDANCE_FACTOR
Definition: aac.h:54
#define TNS_MAX_ORDER
Definition: aac.h:51
#define NOISE_PRE_BITS
length of preamble
Definition: aac.h:158
#define SCALE_DIFF_ZERO
codebook index corresponding to zero scalefactor indices difference
Definition: aac.h:153
#define NOISE_OFFSET
subtracted from global gain, used as offset for the preamble
Definition: aac.h:159
const AACCoefficientsEncoder ff_aac_coders[AAC_CODER_NB]
Definition: aaccoder.c:897
static void put_bitstream_info(AACEncContext *s, const char *name)
Write some auxiliary information about the created AAC file.
Definition: aacenc.c:511
static const AVCodecDefault aac_encode_defaults[]
Definition: aacenc.c:1129
static void avoid_clipping(AACEncContext *s, SingleChannelElement *sce)
Downscale spectral coefficients for near-clipping windows to avoid artifacts.
Definition: aacenc.c:465
static void apply_mid_side_stereo(ChannelElement *cpe)
Definition: aacenc.c:333
static void adjust_frame_information(ChannelElement *cpe, int chans)
Produce integer coefficients from scalefactors provided by the model.
Definition: aacenc.c:254
static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce, int common_window)
Encode one channel of audio data.
Definition: aacenc.c:485
static av_cold int dsp_init(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:916
static const AVOption aacenc_options[]
Definition: aacenc.c:1106
static int put_audio_specific_config(AVCodecContext *avctx)
Make AAC audio config object.
Definition: aacenc.c:92
static void(*const apply_window[4])(AVFloatDSPContext *fdsp, SingleChannelElement *sce, const float *audio)
Definition: aacenc.c:189
static void encode_pulses(AACEncContext *s, Pulse *pulse)
Encode pulse data.
Definition: aacenc.c:418
#define WINDOW_FUNC(type)
Definition: aacenc.c:133
static const AVClass aacenc_class
Definition: aacenc.c:1122
static av_cold int aac_encode_end(AVCodecContext *avctx)
Definition: aacenc.c:897
static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
Encode scalefactor band coding type.
Definition: aacenc.c:368
static void apply_window_and_mdct(AACEncContext *s, SingleChannelElement *sce, float *audio)
Definition: aacenc.c:198
static av_cold int aac_encode_init(AVCodecContext *avctx)
Definition: aacenc.c:948
static void apply_intensity_stereo(ChannelElement *cpe)
Definition: aacenc.c:304
static void put_pce(PutBitContext *pb, AVCodecContext *avctx)
Definition: aacenc.c:51
static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
Encode MS data.
Definition: aacenc.c:240
static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
Encode ics_info element.
Definition: aacenc.c:219
#define AACENC_FLAGS
Definition: aacenc.c:1105
static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
Encode spectral coefficients processed by psychoacoustic model.
Definition: aacenc.c:437
static av_cold int alloc_buffers(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:935
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: aacenc.c:554
static void copy_input_samples(AACEncContext *s, const AVFrame *frame)
Definition: aacenc.c:532
static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce)
Encode scalefactors.
Definition: aacenc.c:382
void ff_quantize_band_cost_cache_init(struct AACEncContext *s)
Definition: aacenc.c:124
AVCodec ff_aac_encoder
Definition: aacenc.c:1134
void ff_aac_coder_init_mips(AACEncContext *c)
static const AACPCEInfo aac_pce_configs[]
List of PCE (Program Configuration Element) for the channel layouts listed in channel_layout....
Definition: aacenc.h:139
@ AAC_CODER_ANMR
Definition: aacenc.h:38
@ AAC_CODER_FAST
Definition: aacenc.h:40
@ AAC_CODER_NB
Definition: aacenc.h:42
@ AAC_CODER_TWOLOOP
Definition: aacenc.h:39
void ff_aac_dsp_init_x86(AACEncContext *s)
AAC encoder utilities.
#define WARN_IF(cond,...)
Definition: aacenc_utils.h:274
static void quantize_bands(int *out, const float *in, const float *scaled, int size, int is_signed, int maxval, const float Q34, const float rounding)
Definition: aacenc_utils.h:65
#define ERROR_IF(cond,...)
Definition: aacenc_utils.h:268
static void abs_pow34_v(float *out, const float *in, const int size)
Definition: aacenc_utils.h:40
const uint8_t *const ff_aac_swb_size_1024[]
Definition: aacenctab.c:99
const int ff_aac_swb_size_128_len
Definition: aacenctab.c:107
const uint8_t *const ff_aac_swb_size_128[]
Definition: aacenctab.c:91
const int ff_aac_swb_size_1024_len
Definition: aacenctab.c:108
AAC encoder data.
static const uint8_t aac_chan_maps[AAC_MAX_CHANNELS][AAC_MAX_CHANNELS]
Table to remap channels from libavcodec's default order to AAC order.
Definition: aacenctab.h:72
static const int64_t aac_normal_chan_layouts[7]
Definition: aacenctab.h:47
#define AAC_MAX_CHANNELS
Definition: aacenctab.h:39
static const uint8_t aac_chan_configs[AAC_MAX_CHANNELS][6]
default channel configurations
Definition: aacenctab.h:58
static const int aacenc_profiles[]
Definition: aacenctab.h:132
static const int mpeg4audio_sample_rates[16]
Definition: aacenctab.h:85
const uint32_t ff_aac_scalefactor_code[121]
Definition: aactab.c:92
const uint8_t ff_tns_max_bands_1024[]
Definition: aactab.c:1413
const uint16_t *const ff_swb_offset_128[]
Definition: aactab.c:1387
const uint16_t *const ff_swb_offset_1024[]
Definition: aactab.c:1355
const uint8_t ff_aac_scalefactor_bits[121]
Definition: aactab.c:111
void ff_aac_tableinit(void)
Definition: aactab.c:3347
const uint8_t ff_aac_num_swb_1024[]
Definition: aactab.c:64
const uint8_t ff_aac_num_swb_128[]
Definition: aactab.c:80
const uint8_t ff_tns_max_bands_128[]
Definition: aactab.c:1425
AAC data declarations.
float ff_aac_kbd_long_1024[1024]
void ff_aac_float_common_init(void)
float ff_aac_kbd_short_128[128]
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:925
static const AVCodecDefault defaults[]
Definition: amfenc_h264.c:361
channels
Definition: aptx.h:33
#define L(x)
Definition: vp56_arith.h:36
#define av_cold
Definition: attributes.h:88
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(const int16_t *) pi >> 8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(const int32_t *) pi >> 24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) #define SET_CONV_FUNC_GROUP(ofmt, ifmt) static void set_generic_function(AudioConvert *ac) { } void ff_audio_convert_free(AudioConvert **ac) { if(! *ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);} AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, int sample_rate, int apply_map) { AudioConvert *ac;int in_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) return NULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method !=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt) > 2) { ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc) { av_free(ac);return NULL;} return ac;} in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar) { ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar ? ac->channels :1;} else if(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;else ac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);return ac;} int ff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in) { int use_generic=1;int len=in->nb_samples;int p;if(ac->dc) { av_log(ac->avr, AV_LOG_TRACE, "%d samples - audio_convert: %s to %s (dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));return ff_convert_dither(ac-> in
uint8_t
av_cold void ff_af_queue_init(AVCodecContext *avctx, AudioFrameQueue *afq)
Initialize AudioFrameQueue.
void ff_af_queue_close(AudioFrameQueue *afq)
Close AudioFrameQueue.
void ff_af_queue_remove(AudioFrameQueue *afq, int nb_samples, int64_t *pts, int64_t *duration)
Remove frame(s) from the queue.
int ff_af_queue_add(AudioFrameQueue *afq, const AVFrame *f)
Add a frame to the queue.
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
Libavcodec external API header.
#define FF_COMPLIANCE_EXPERIMENTAL
Allow nonstandardized experimental things.
Definition: avcodec.h:1606
#define FF_PROFILE_UNKNOWN
Definition: avcodec.h:1859
#define FF_PROFILE_MPEG2_AAC_LOW
Definition: avcodec.h:1870
#define FF_PROFILE_AAC_MAIN
Definition: avcodec.h:1862
#define FF_PROFILE_AAC_LOW
Definition: avcodec.h:1863
#define FF_PROFILE_AAC_LTP
Definition: avcodec.h:1865
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:31
void ff_put_string(PutBitContext *pb, const char *string, int terminate_string)
Put the string string in the bitstream.
Definition: bitstream.c:59
#define s(width, name)
Definition: cbs_vp9.c:257
#define FFMIN(a, b)
Definition: common.h:105
#define FFMAX(a, b)
Definition: common.h:103
#define av_clipf
Definition: common.h:170
#define HAVE_MIPSDSP
Definition: config.h:81
#define ARCH_X86
Definition: config.h:39
#define NULL
Definition: coverity.c:32
long long int64_t
Definition: coverity.c:34
static __device__ float fabsf(float a)
Definition: cuda_runtime.h:181
static __device__ float fabs(float a)
Definition: cuda_runtime.h:182
#define max(a, b)
Definition: cuda_runtime.h:33
static AVFrame * frame
int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
Check AVPacket size and/or allocate data.
Definition: encode.c:33
const OptionDef options[]
#define ff_mdct_init
Definition: fft.h:161
#define ff_mdct_end
Definition: fft.h:162
@ AV_OPT_TYPE_CONST
Definition: opt.h:234
@ AV_OPT_TYPE_INT
Definition: opt.h:225
@ AV_OPT_TYPE_BOOL
Definition: opt.h:242
void av_get_channel_layout_string(char *buf, int buf_size, int nb_channels, uint64_t channel_layout)
Return a description of a channel layout.
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:333
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: codec.h:77
#define AV_CODEC_FLAG_QSCALE
Use fixed qscale.
Definition: avcodec.h:275
#define AV_CODEC_CAP_SMALL_LAST_FRAME
Codec can be fed a final frame with a smaller size.
Definition: codec.h:82
@ AV_CODEC_ID_AAC
Definition: codec_id.h:426
#define AVERROR(e)
Definition: error.h:43
#define AV_LOG_INFO
Standard information.
Definition: log.h:205
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:194
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:235
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:237
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:69
@ AV_SAMPLE_FMT_NONE
Definition: samplefmt.h:59
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
for(j=16;j >0;--j)
#define R
Definition: huffyuvdsp.h:34
static const int sizes[][2]
Definition: img2dec.c:54
int i
Definition: input.c:407
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:218
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:41
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: internal.h:49
#define LIBAVCODEC_IDENT
Definition: version.h:42
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:135
common internal API header
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:117
#define FF_ALLOCZ_TYPED_ARRAY(p, nelem)
Definition: internal.h:103
Replacements for frequently missing libm functions.
uint8_t w
Definition: llviddspenc.c:39
av_cold void ff_lpc_end(LPCContext *s)
Uninitialize LPCContext.
Definition: lpc.c:325
av_cold int ff_lpc_init(LPCContext *s, int blocksize, int max_order, enum FFLPCType lpc_type)
Initialize LPCContext.
Definition: lpc.c:303
@ FF_LPC_TYPE_LEVINSON
Levinson-Durbin recursion.
Definition: lpc.h:47
#define NAN
Definition: mathematics.h:64
uint32_t tag
Definition: movenc.c:1611
const int avpriv_mpeg4audio_sample_rates[16]
Definition: mpeg4audio.c:62
@ AOT_SBR
Y Spectral Band Replication.
Definition: mpeg4audio.h:94
AVOptions.
#define FF_AAC_PROFILE_OPTS
Definition: profiles.h:28
av_cold struct FFPsyPreprocessContext * ff_psy_preprocess_init(AVCodecContext *avctx)
psychoacoustic model audio preprocessing initialization
Definition: psymodel.c:103
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx)
Cleanup audio preprocessing module.
Definition: psymodel.c:152
av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, int num_lens, const uint8_t **bands, const int *num_bands, int num_groups, const uint8_t *group_map)
Initialize psychoacoustic model.
Definition: psymodel.c:31
av_cold void ff_psy_end(FFPsyContext *ctx)
Cleanup model context at the end.
Definition: psymodel.c:83
void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx, float **audio, int channels)
Preprocess several channel in audio frame in order to compress it better.
Definition: psymodel.c:139
bitstream writer API
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:57
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:76
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:110
static void align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: put_bits.h:386
const char * name
Definition: qsvenc.c:46
typedef void(RENAME(mix_any_func_type))
#define FF_ARRAY_ELEMS(a)
AAC encoder context.
Definition: aacenc.h:378
int num_ele[4]
front, side, back, lfe
Definition: aacenc.h:97
int index[4][8]
front, side, back, lfe
Definition: aacenc.h:99
int pairing[3][8]
front, side, back
Definition: aacenc.h:98
int64_t layout
Definition: aacenc.h:96
Describe the class of an AVClass context structure.
Definition: log.h:67
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
main external API structure.
Definition: avcodec.h:536
int global_quality
Global quality for codecs which cannot change it per frame.
Definition: avcodec.h:602
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:1601
int64_t bit_rate
the average bitrate
Definition: avcodec.h:586
int profile
profile
Definition: avcodec.h:1858
int initial_padding
Audio only.
Definition: avcodec.h:2062
int sample_rate
samples per second
Definition: avcodec.h:1196
int frame_number
Frame counter, set by libavcodec.
Definition: avcodec.h:1227
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:616
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:637
int channels
number of audio channels
Definition: avcodec.h:1197
int extradata_size
Definition: avcodec.h:638
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:1247
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:1216
void * priv_data
Definition: avcodec.h:563
AVCodec.
Definition: codec.h:197
const char * name
Name of the codec implementation.
Definition: codec.h:204
This structure describes decoded (raw) audio or video data.
Definition: frame.h:318
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:384
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:365
AVOption.
Definition: opt.h:248
This structure stores compressed data.
Definition: packet.h:346
int size
Definition: packet.h:370
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: packet.h:387
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: packet.h:362
uint8_t * data
Definition: packet.h:369
channel element - generic struct for SCE/CPE/CCE/LFE
Definition: aac.h:276
uint8_t is_mask[128]
Set if intensity stereo is used (used by encoder)
Definition: aac.h:283
int ms_mode
Signals mid/side stereo flags coding mode (used by encoder)
Definition: aac.h:280
int common_window
Set if channels share a common 'IndividualChannelStream' in bitstream.
Definition: aac.h:279
uint8_t is_mode
Set if any bands have been encoded using intensity stereo (used by encoder)
Definition: aac.h:281
uint8_t ms_mask[128]
Set if mid/side stereo is used for each scalefactor window band.
Definition: aac.h:282
SingleChannelElement ch[2]
Definition: aac.h:285
windowing related information
Definition: psymodel.h:77
int num_windows
number of windows in a frame
Definition: psymodel.h:80
int grouping[8]
window grouping (for e.g. AAC)
Definition: psymodel.h:81
float clipping[8]
maximum absolute normalized intensity in the given window for clip avoidance
Definition: psymodel.h:82
int window_shape
window shape (sine/KBD/whatever)
Definition: psymodel.h:79
int window_type[3]
window type (short/long/transitional, etc.) - current, previous and next
Definition: psymodel.h:78
Individual Channel Stream.
Definition: aac.h:175
uint8_t max_sfb
number of scalefactor bands per group
Definition: aac.h:176
uint8_t group_len[8]
Definition: aac.h:180
int num_swb
number of scalefactor window bands
Definition: aac.h:184
LongTermPrediction ltp
Definition: aac.h:181
uint8_t use_kb_window[2]
If set, use Kaiser-Bessel window, otherwise use a sine window.
Definition: aac.h:178
uint8_t prediction_used[41]
Definition: aac.h:191
float clip_avoidance_factor
set if any window is near clipping to the necessary atennuation factor to avoid it
Definition: aac.h:193
const uint8_t * swb_sizes
table of scalefactor band sizes for a particular window
Definition: aac.h:183
enum WindowSequence window_sequence[2]
Definition: aac.h:177
uint8_t window_clipping[8]
set if a certain window is near clipping
Definition: aac.h:192
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:182
int8_t present
Definition: aac.h:165
int8_t used[MAX_LTP_LONG_SFB]
Definition: aac.h:169
Definition: aac.h:225
int pos[4]
Definition: aac.h:228
int start
Definition: aac.h:227
int amp[4]
Definition: aac.h:229
int num_pulse
Definition: aac.h:226
Single Channel Element - used for both SCE and LFE elements.
Definition: aac.h:249
uint8_t zeroes[128]
band is not coded (used by encoder)
Definition: aac.h:258
enum BandType band_type[128]
band types
Definition: aac.h:253
float is_ener[128]
Intensity stereo pos (used by encoder)
Definition: aac.h:260
INTFLOAT pcoeffs[1024]
coefficients for IMDCT, pristine
Definition: aac.h:262
INTFLOAT ret_buf[2048]
PCM output buffer.
Definition: aac.h:265
TemporalNoiseShaping tns
Definition: aac.h:251
INTFLOAT coeffs[1024]
coefficients for IMDCT, maybe processed
Definition: aac.h:263
INTFLOAT ltp_state[3072]
time signal for LTP
Definition: aac.h:266
AAC_FLOAT lcoeffs[1024]
MDCT of LTP coefficients (used by encoder)
Definition: aac.h:267
IndividualChannelStream ics
Definition: aac.h:250
int sf_idx[128]
scalefactor indices (used by encoder)
Definition: aac.h:257
Temporal Noise Shaping.
Definition: aac.h:199
#define av_freep(p)
#define av_log(a,...)
FILE * out
Definition: movenc.c:54
const char * g
Definition: vf_curves.c:117
if(ret< 0)
Definition: vf_mcdeint.c:282
static av_always_inline int diff(const uint32_t a, const uint32_t b)