FFmpeg  4.3
aacenc.c
Go to the documentation of this file.
1 /*
2  * AAC encoder
3  * Copyright (C) 2008 Konstantin Shishkov
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 /**
23  * @file
24  * AAC encoder
25  */
26 
27 /***********************************
28  * TODOs:
29  * add sane pulse detection
30  ***********************************/
31 
32 #include "libavutil/libm.h"
33 #include "libavutil/thread.h"
34 #include "libavutil/float_dsp.h"
35 #include "libavutil/opt.h"
36 #include "avcodec.h"
37 #include "put_bits.h"
38 #include "internal.h"
39 #include "mpeg4audio.h"
40 #include "kbdwin.h"
41 #include "sinewin.h"
42 #include "profiles.h"
43 
44 #include "aac.h"
45 #include "aactab.h"
46 #include "aacenc.h"
47 #include "aacenctab.h"
48 #include "aacenc_utils.h"
49 
50 #include "psymodel.h"
51 
53 
54 static void put_pce(PutBitContext *pb, AVCodecContext *avctx)
55 {
56  int i, j;
57  AACEncContext *s = avctx->priv_data;
58  AACPCEInfo *pce = &s->pce;
59  const int bitexact = avctx->flags & AV_CODEC_FLAG_BITEXACT;
60  const char *aux_data = bitexact ? "Lavc" : LIBAVCODEC_IDENT;
61 
62  put_bits(pb, 4, 0);
63 
64  put_bits(pb, 2, avctx->profile);
65  put_bits(pb, 4, s->samplerate_index);
66 
67  put_bits(pb, 4, pce->num_ele[0]); /* Front */
68  put_bits(pb, 4, pce->num_ele[1]); /* Side */
69  put_bits(pb, 4, pce->num_ele[2]); /* Back */
70  put_bits(pb, 2, pce->num_ele[3]); /* LFE */
71  put_bits(pb, 3, 0); /* Assoc data */
72  put_bits(pb, 4, 0); /* CCs */
73 
74  put_bits(pb, 1, 0); /* Stereo mixdown */
75  put_bits(pb, 1, 0); /* Mono mixdown */
76  put_bits(pb, 1, 0); /* Something else */
77 
78  for (i = 0; i < 4; i++) {
79  for (j = 0; j < pce->num_ele[i]; j++) {
80  if (i < 3)
81  put_bits(pb, 1, pce->pairing[i][j]);
82  put_bits(pb, 4, pce->index[i][j]);
83  }
84  }
85 
87  put_bits(pb, 8, strlen(aux_data));
88  avpriv_put_string(pb, aux_data, 0);
89 }
90 
91 /**
92  * Make AAC audio config object.
93  * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
94  */
96 {
97  PutBitContext pb;
98  AACEncContext *s = avctx->priv_data;
99  int channels = (!s->needs_pce)*(s->channels - (s->channels == 8 ? 1 : 0));
100  const int max_size = 32;
101 
102  avctx->extradata = av_mallocz(max_size);
103  if (!avctx->extradata)
104  return AVERROR(ENOMEM);
105 
106  init_put_bits(&pb, avctx->extradata, max_size);
107  put_bits(&pb, 5, s->profile+1); //profile
108  put_bits(&pb, 4, s->samplerate_index); //sample rate index
109  put_bits(&pb, 4, channels);
110  //GASpecificConfig
111  put_bits(&pb, 1, 0); //frame length - 1024 samples
112  put_bits(&pb, 1, 0); //does not depend on core coder
113  put_bits(&pb, 1, 0); //is not extension
114  if (s->needs_pce)
115  put_pce(&pb, avctx);
116 
117  //Explicitly Mark SBR absent
118  put_bits(&pb, 11, 0x2b7); //sync extension
119  put_bits(&pb, 5, AOT_SBR);
120  put_bits(&pb, 1, 0);
121  flush_put_bits(&pb);
122  avctx->extradata_size = put_bits_count(&pb) >> 3;
123 
124  return 0;
125 }
126 
128 {
131  memset(s->quantize_band_cost_cache, 0, sizeof(s->quantize_band_cost_cache));
133  }
134 }
135 
136 #define WINDOW_FUNC(type) \
137 static void apply_ ##type ##_window(AVFloatDSPContext *fdsp, \
138  SingleChannelElement *sce, \
139  const float *audio)
140 
141 WINDOW_FUNC(only_long)
142 {
143  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
144  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
145  float *out = sce->ret_buf;
146 
147  fdsp->vector_fmul (out, audio, lwindow, 1024);
148  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, pwindow, 1024);
149 }
150 
151 WINDOW_FUNC(long_start)
152 {
153  const float *lwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
154  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
155  float *out = sce->ret_buf;
156 
157  fdsp->vector_fmul(out, audio, lwindow, 1024);
158  memcpy(out + 1024, audio + 1024, sizeof(out[0]) * 448);
159  fdsp->vector_fmul_reverse(out + 1024 + 448, audio + 1024 + 448, swindow, 128);
160  memset(out + 1024 + 576, 0, sizeof(out[0]) * 448);
161 }
162 
163 WINDOW_FUNC(long_stop)
164 {
165  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
166  const float *swindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
167  float *out = sce->ret_buf;
168 
169  memset(out, 0, sizeof(out[0]) * 448);
170  fdsp->vector_fmul(out + 448, audio + 448, swindow, 128);
171  memcpy(out + 576, audio + 576, sizeof(out[0]) * 448);
172  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, lwindow, 1024);
173 }
174 
175 WINDOW_FUNC(eight_short)
176 {
177  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
178  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
179  const float *in = audio + 448;
180  float *out = sce->ret_buf;
181  int w;
182 
183  for (w = 0; w < 8; w++) {
184  fdsp->vector_fmul (out, in, w ? pwindow : swindow, 128);
185  out += 128;
186  in += 128;
187  fdsp->vector_fmul_reverse(out, in, swindow, 128);
188  out += 128;
189  }
190 }
191 
192 static void (*const apply_window[4])(AVFloatDSPContext *fdsp,
194  const float *audio) = {
195  [ONLY_LONG_SEQUENCE] = apply_only_long_window,
196  [LONG_START_SEQUENCE] = apply_long_start_window,
197  [EIGHT_SHORT_SEQUENCE] = apply_eight_short_window,
198  [LONG_STOP_SEQUENCE] = apply_long_stop_window
199 };
200 
202  float *audio)
203 {
204  int i;
205  const float *output = sce->ret_buf;
206 
207  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, audio);
208 
210  s->mdct1024.mdct_calc(&s->mdct1024, sce->coeffs, output);
211  else
212  for (i = 0; i < 1024; i += 128)
213  s->mdct128.mdct_calc(&s->mdct128, &sce->coeffs[i], output + i*2);
214  memcpy(audio, audio + 1024, sizeof(audio[0]) * 1024);
215  memcpy(sce->pcoeffs, sce->coeffs, sizeof(sce->pcoeffs));
216 }
217 
218 /**
219  * Encode ics_info element.
220  * @see Table 4.6 (syntax of ics_info)
221  */
223 {
224  int w;
225 
226  put_bits(&s->pb, 1, 0); // ics_reserved bit
227  put_bits(&s->pb, 2, info->window_sequence[0]);
228  put_bits(&s->pb, 1, info->use_kb_window[0]);
229  if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
230  put_bits(&s->pb, 6, info->max_sfb);
231  put_bits(&s->pb, 1, !!info->predictor_present);
232  } else {
233  put_bits(&s->pb, 4, info->max_sfb);
234  for (w = 1; w < 8; w++)
235  put_bits(&s->pb, 1, !info->group_len[w]);
236  }
237 }
238 
239 /**
240  * Encode MS data.
241  * @see 4.6.8.1 "Joint Coding - M/S Stereo"
242  */
244 {
245  int i, w;
246 
247  put_bits(pb, 2, cpe->ms_mode);
248  if (cpe->ms_mode == 1)
249  for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
250  for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
251  put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
252 }
253 
254 /**
255  * Produce integer coefficients from scalefactors provided by the model.
256  */
257 static void adjust_frame_information(ChannelElement *cpe, int chans)
258 {
259  int i, w, w2, g, ch;
260  int maxsfb, cmaxsfb;
261 
262  for (ch = 0; ch < chans; ch++) {
263  IndividualChannelStream *ics = &cpe->ch[ch].ics;
264  maxsfb = 0;
265  cpe->ch[ch].pulse.num_pulse = 0;
266  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
267  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
268  for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w*16+cmaxsfb-1]; cmaxsfb--)
269  ;
270  maxsfb = FFMAX(maxsfb, cmaxsfb);
271  }
272  }
273  ics->max_sfb = maxsfb;
274 
275  //adjust zero bands for window groups
276  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
277  for (g = 0; g < ics->max_sfb; g++) {
278  i = 1;
279  for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
280  if (!cpe->ch[ch].zeroes[w2*16 + g]) {
281  i = 0;
282  break;
283  }
284  }
285  cpe->ch[ch].zeroes[w*16 + g] = i;
286  }
287  }
288  }
289 
290  if (chans > 1 && cpe->common_window) {
291  IndividualChannelStream *ics0 = &cpe->ch[0].ics;
292  IndividualChannelStream *ics1 = &cpe->ch[1].ics;
293  int msc = 0;
294  ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
295  ics1->max_sfb = ics0->max_sfb;
296  for (w = 0; w < ics0->num_windows*16; w += 16)
297  for (i = 0; i < ics0->max_sfb; i++)
298  if (cpe->ms_mask[w+i])
299  msc++;
300  if (msc == 0 || ics0->max_sfb == 0)
301  cpe->ms_mode = 0;
302  else
303  cpe->ms_mode = msc < ics0->max_sfb * ics0->num_windows ? 1 : 2;
304  }
305 }
306 
308 {
309  int w, w2, g, i;
310  IndividualChannelStream *ics = &cpe->ch[0].ics;
311  if (!cpe->common_window)
312  return;
313  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
314  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
315  int start = (w+w2) * 128;
316  for (g = 0; g < ics->num_swb; g++) {
317  int p = -1 + 2 * (cpe->ch[1].band_type[w*16+g] - 14);
318  float scale = cpe->ch[0].is_ener[w*16+g];
319  if (!cpe->is_mask[w*16 + g]) {
320  start += ics->swb_sizes[g];
321  continue;
322  }
323  if (cpe->ms_mask[w*16 + g])
324  p *= -1;
325  for (i = 0; i < ics->swb_sizes[g]; i++) {
326  float sum = (cpe->ch[0].coeffs[start+i] + p*cpe->ch[1].coeffs[start+i])*scale;
327  cpe->ch[0].coeffs[start+i] = sum;
328  cpe->ch[1].coeffs[start+i] = 0.0f;
329  }
330  start += ics->swb_sizes[g];
331  }
332  }
333  }
334 }
335 
337 {
338  int w, w2, g, i;
339  IndividualChannelStream *ics = &cpe->ch[0].ics;
340  if (!cpe->common_window)
341  return;
342  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
343  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
344  int start = (w+w2) * 128;
345  for (g = 0; g < ics->num_swb; g++) {
346  /* ms_mask can be used for other purposes in PNS and I/S,
347  * so must not apply M/S if any band uses either, even if
348  * ms_mask is set.
349  */
350  if (!cpe->ms_mask[w*16 + g] || cpe->is_mask[w*16 + g]
351  || cpe->ch[0].band_type[w*16 + g] >= NOISE_BT
352  || cpe->ch[1].band_type[w*16 + g] >= NOISE_BT) {
353  start += ics->swb_sizes[g];
354  continue;
355  }
356  for (i = 0; i < ics->swb_sizes[g]; i++) {
357  float L = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) * 0.5f;
358  float R = L - cpe->ch[1].coeffs[start+i];
359  cpe->ch[0].coeffs[start+i] = L;
360  cpe->ch[1].coeffs[start+i] = R;
361  }
362  start += ics->swb_sizes[g];
363  }
364  }
365  }
366 }
367 
368 /**
369  * Encode scalefactor band coding type.
370  */
372 {
373  int w;
374 
377 
378  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
379  s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
380 }
381 
382 /**
383  * Encode scalefactors.
384  */
387 {
388  int diff, off_sf = sce->sf_idx[0], off_pns = sce->sf_idx[0] - NOISE_OFFSET;
389  int off_is = 0, noise_flag = 1;
390  int i, w;
391 
392  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
393  for (i = 0; i < sce->ics.max_sfb; i++) {
394  if (!sce->zeroes[w*16 + i]) {
395  if (sce->band_type[w*16 + i] == NOISE_BT) {
396  diff = sce->sf_idx[w*16 + i] - off_pns;
397  off_pns = sce->sf_idx[w*16 + i];
398  if (noise_flag-- > 0) {
399  put_bits(&s->pb, NOISE_PRE_BITS, diff + NOISE_PRE);
400  continue;
401  }
402  } else if (sce->band_type[w*16 + i] == INTENSITY_BT ||
403  sce->band_type[w*16 + i] == INTENSITY_BT2) {
404  diff = sce->sf_idx[w*16 + i] - off_is;
405  off_is = sce->sf_idx[w*16 + i];
406  } else {
407  diff = sce->sf_idx[w*16 + i] - off_sf;
408  off_sf = sce->sf_idx[w*16 + i];
409  }
410  diff += SCALE_DIFF_ZERO;
411  av_assert0(diff >= 0 && diff <= 120);
413  }
414  }
415  }
416 }
417 
418 /**
419  * Encode pulse data.
420  */
421 static void encode_pulses(AACEncContext *s, Pulse *pulse)
422 {
423  int i;
424 
425  put_bits(&s->pb, 1, !!pulse->num_pulse);
426  if (!pulse->num_pulse)
427  return;
428 
429  put_bits(&s->pb, 2, pulse->num_pulse - 1);
430  put_bits(&s->pb, 6, pulse->start);
431  for (i = 0; i < pulse->num_pulse; i++) {
432  put_bits(&s->pb, 5, pulse->pos[i]);
433  put_bits(&s->pb, 4, pulse->amp[i]);
434  }
435 }
436 
437 /**
438  * Encode spectral coefficients processed by psychoacoustic model.
439  */
441 {
442  int start, i, w, w2;
443 
444  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
445  start = 0;
446  for (i = 0; i < sce->ics.max_sfb; i++) {
447  if (sce->zeroes[w*16 + i]) {
448  start += sce->ics.swb_sizes[i];
449  continue;
450  }
451  for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
452  s->coder->quantize_and_encode_band(s, &s->pb,
453  &sce->coeffs[start + w2*128],
454  NULL, sce->ics.swb_sizes[i],
455  sce->sf_idx[w*16 + i],
456  sce->band_type[w*16 + i],
457  s->lambda,
458  sce->ics.window_clipping[w]);
459  }
460  start += sce->ics.swb_sizes[i];
461  }
462  }
463 }
464 
465 /**
466  * Downscale spectral coefficients for near-clipping windows to avoid artifacts
467  */
469 {
470  int start, i, j, w;
471 
472  if (sce->ics.clip_avoidance_factor < 1.0f) {
473  for (w = 0; w < sce->ics.num_windows; w++) {
474  start = 0;
475  for (i = 0; i < sce->ics.max_sfb; i++) {
476  float *swb_coeffs = &sce->coeffs[start + w*128];
477  for (j = 0; j < sce->ics.swb_sizes[i]; j++)
478  swb_coeffs[j] *= sce->ics.clip_avoidance_factor;
479  start += sce->ics.swb_sizes[i];
480  }
481  }
482  }
483 }
484 
485 /**
486  * Encode one channel of audio data.
487  */
490  int common_window)
491 {
492  put_bits(&s->pb, 8, sce->sf_idx[0]);
493  if (!common_window) {
494  put_ics_info(s, &sce->ics);
495  if (s->coder->encode_main_pred)
496  s->coder->encode_main_pred(s, sce);
497  if (s->coder->encode_ltp_info)
498  s->coder->encode_ltp_info(s, sce, 0);
499  }
500  encode_band_info(s, sce);
501  encode_scale_factors(avctx, s, sce);
502  encode_pulses(s, &sce->pulse);
503  put_bits(&s->pb, 1, !!sce->tns.present);
504  if (s->coder->encode_tns_info)
505  s->coder->encode_tns_info(s, sce);
506  put_bits(&s->pb, 1, 0); //ssr
507  encode_spectral_coeffs(s, sce);
508  return 0;
509 }
510 
511 /**
512  * Write some auxiliary information about the created AAC file.
513  */
514 static void put_bitstream_info(AACEncContext *s, const char *name)
515 {
516  int i, namelen, padbits;
517 
518  namelen = strlen(name) + 2;
519  put_bits(&s->pb, 3, TYPE_FIL);
520  put_bits(&s->pb, 4, FFMIN(namelen, 15));
521  if (namelen >= 15)
522  put_bits(&s->pb, 8, namelen - 14);
523  put_bits(&s->pb, 4, 0); //extension type - filler
524  padbits = -put_bits_count(&s->pb) & 7;
526  for (i = 0; i < namelen - 2; i++)
527  put_bits(&s->pb, 8, name[i]);
528  put_bits(&s->pb, 12 - padbits, 0);
529 }
530 
531 /*
532  * Copy input samples.
533  * Channels are reordered from libavcodec's default order to AAC order.
534  */
536 {
537  int ch;
538  int end = 2048 + (frame ? frame->nb_samples : 0);
539  const uint8_t *channel_map = s->reorder_map;
540 
541  /* copy and remap input samples */
542  for (ch = 0; ch < s->channels; ch++) {
543  /* copy last 1024 samples of previous frame to the start of the current frame */
544  memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0]));
545 
546  /* copy new samples and zero any remaining samples */
547  if (frame) {
548  memcpy(&s->planar_samples[ch][2048],
549  frame->extended_data[channel_map[ch]],
550  frame->nb_samples * sizeof(s->planar_samples[0][0]));
551  }
552  memset(&s->planar_samples[ch][end], 0,
553  (3072 - end) * sizeof(s->planar_samples[0][0]));
554  }
555 }
556 
557 static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
558  const AVFrame *frame, int *got_packet_ptr)
559 {
560  AACEncContext *s = avctx->priv_data;
561  float **samples = s->planar_samples, *samples2, *la, *overlap;
562  ChannelElement *cpe;
565  int i, its, ch, w, chans, tag, start_ch, ret, frame_bits;
566  int target_bits, rate_bits, too_many_bits, too_few_bits;
567  int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
568  int chan_el_counter[4];
570 
571  /* add current frame to queue */
572  if (frame) {
573  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
574  return ret;
575  } else {
576  if (!s->afq.remaining_samples || (!s->afq.frame_alloc && !s->afq.frame_count))
577  return 0;
578  }
579 
580  copy_input_samples(s, frame);
581  if (s->psypp)
583 
584  if (!avctx->frame_number)
585  return 0;
586 
587  start_ch = 0;
588  for (i = 0; i < s->chan_map[0]; i++) {
589  FFPsyWindowInfo* wi = windows + start_ch;
590  tag = s->chan_map[i+1];
591  chans = tag == TYPE_CPE ? 2 : 1;
592  cpe = &s->cpe[i];
593  for (ch = 0; ch < chans; ch++) {
594  int k;
595  float clip_avoidance_factor;
596  sce = &cpe->ch[ch];
597  ics = &sce->ics;
598  s->cur_channel = start_ch + ch;
599  overlap = &samples[s->cur_channel][0];
600  samples2 = overlap + 1024;
601  la = samples2 + (448+64);
602  if (!frame)
603  la = NULL;
604  if (tag == TYPE_LFE) {
605  wi[ch].window_type[0] = wi[ch].window_type[1] = ONLY_LONG_SEQUENCE;
606  wi[ch].window_shape = 0;
607  wi[ch].num_windows = 1;
608  wi[ch].grouping[0] = 1;
609  wi[ch].clipping[0] = 0;
610 
611  /* Only the lowest 12 coefficients are used in a LFE channel.
612  * The expression below results in only the bottom 8 coefficients
613  * being used for 11.025kHz to 16kHz sample rates.
614  */
615  ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
616  } else {
617  wi[ch] = s->psy.model->window(&s->psy, samples2, la, s->cur_channel,
618  ics->window_sequence[0]);
619  }
620  ics->window_sequence[1] = ics->window_sequence[0];
621  ics->window_sequence[0] = wi[ch].window_type[0];
622  ics->use_kb_window[1] = ics->use_kb_window[0];
623  ics->use_kb_window[0] = wi[ch].window_shape;
624  ics->num_windows = wi[ch].num_windows;
625  ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
626  ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
627  ics->max_sfb = FFMIN(ics->max_sfb, ics->num_swb);
628  ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
634 
635  for (w = 0; w < ics->num_windows; w++)
636  ics->group_len[w] = wi[ch].grouping[w];
637 
638  /* Calculate input sample maximums and evaluate clipping risk */
639  clip_avoidance_factor = 0.0f;
640  for (w = 0; w < ics->num_windows; w++) {
641  const float *wbuf = overlap + w * 128;
642  const int wlen = 2048 / ics->num_windows;
643  float max = 0;
644  int j;
645  /* mdct input is 2 * output */
646  for (j = 0; j < wlen; j++)
647  max = FFMAX(max, fabsf(wbuf[j]));
648  wi[ch].clipping[w] = max;
649  }
650  for (w = 0; w < ics->num_windows; w++) {
651  if (wi[ch].clipping[w] > CLIP_AVOIDANCE_FACTOR) {
652  ics->window_clipping[w] = 1;
653  clip_avoidance_factor = FFMAX(clip_avoidance_factor, wi[ch].clipping[w]);
654  } else {
655  ics->window_clipping[w] = 0;
656  }
657  }
658  if (clip_avoidance_factor > CLIP_AVOIDANCE_FACTOR) {
659  ics->clip_avoidance_factor = CLIP_AVOIDANCE_FACTOR / clip_avoidance_factor;
660  } else {
661  ics->clip_avoidance_factor = 1.0f;
662  }
663 
664  apply_window_and_mdct(s, sce, overlap);
665 
666  if (s->options.ltp && s->coder->update_ltp) {
667  s->coder->update_ltp(s, sce);
668  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
669  s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
670  }
671 
672  for (k = 0; k < 1024; k++) {
673  if (!(fabs(cpe->ch[ch].coeffs[k]) < 1E16)) { // Ensure headroom for energy calculation
674  av_log(avctx, AV_LOG_ERROR, "Input contains (near) NaN/+-Inf\n");
675  return AVERROR(EINVAL);
676  }
677  }
678  avoid_clipping(s, sce);
679  }
680  start_ch += chans;
681  }
682  if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels, 0)) < 0)
683  return ret;
684  frame_bits = its = 0;
685  do {
686  init_put_bits(&s->pb, avpkt->data, avpkt->size);
687 
688  if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & AV_CODEC_FLAG_BITEXACT))
690  start_ch = 0;
691  target_bits = 0;
692  memset(chan_el_counter, 0, sizeof(chan_el_counter));
693  for (i = 0; i < s->chan_map[0]; i++) {
694  FFPsyWindowInfo* wi = windows + start_ch;
695  const float *coeffs[2];
696  tag = s->chan_map[i+1];
697  chans = tag == TYPE_CPE ? 2 : 1;
698  cpe = &s->cpe[i];
699  cpe->common_window = 0;
700  memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
701  memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
702  put_bits(&s->pb, 3, tag);
703  put_bits(&s->pb, 4, chan_el_counter[tag]++);
704  for (ch = 0; ch < chans; ch++) {
705  sce = &cpe->ch[ch];
706  coeffs[ch] = sce->coeffs;
707  sce->ics.predictor_present = 0;
708  sce->ics.ltp.present = 0;
709  memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
710  memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
711  memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
712  for (w = 0; w < 128; w++)
713  if (sce->band_type[w] > RESERVED_BT)
714  sce->band_type[w] = 0;
715  }
716  s->psy.bitres.alloc = -1;
718  s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
719  if (s->psy.bitres.alloc > 0) {
720  /* Lambda unused here on purpose, we need to take psy's unscaled allocation */
721  target_bits += s->psy.bitres.alloc
722  * (s->lambda / (avctx->global_quality ? avctx->global_quality : 120));
723  s->psy.bitres.alloc /= chans;
724  }
725  s->cur_type = tag;
726  for (ch = 0; ch < chans; ch++) {
727  s->cur_channel = start_ch + ch;
728  if (s->options.pns && s->coder->mark_pns)
729  s->coder->mark_pns(s, avctx, &cpe->ch[ch]);
730  s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
731  }
732  if (chans > 1
733  && wi[0].window_type[0] == wi[1].window_type[0]
734  && wi[0].window_shape == wi[1].window_shape) {
735 
736  cpe->common_window = 1;
737  for (w = 0; w < wi[0].num_windows; w++) {
738  if (wi[0].grouping[w] != wi[1].grouping[w]) {
739  cpe->common_window = 0;
740  break;
741  }
742  }
743  }
744  for (ch = 0; ch < chans; ch++) { /* TNS and PNS */
745  sce = &cpe->ch[ch];
746  s->cur_channel = start_ch + ch;
747  if (s->options.tns && s->coder->search_for_tns)
748  s->coder->search_for_tns(s, sce);
749  if (s->options.tns && s->coder->apply_tns_filt)
750  s->coder->apply_tns_filt(s, sce);
751  if (sce->tns.present)
752  tns_mode = 1;
753  if (s->options.pns && s->coder->search_for_pns)
754  s->coder->search_for_pns(s, avctx, sce);
755  }
756  s->cur_channel = start_ch;
757  if (s->options.intensity_stereo) { /* Intensity Stereo */
758  if (s->coder->search_for_is)
759  s->coder->search_for_is(s, avctx, cpe);
760  if (cpe->is_mode) is_mode = 1;
762  }
763  if (s->options.pred) { /* Prediction */
764  for (ch = 0; ch < chans; ch++) {
765  sce = &cpe->ch[ch];
766  s->cur_channel = start_ch + ch;
767  if (s->options.pred && s->coder->search_for_pred)
768  s->coder->search_for_pred(s, sce);
769  if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
770  }
771  if (s->coder->adjust_common_pred)
772  s->coder->adjust_common_pred(s, cpe);
773  for (ch = 0; ch < chans; ch++) {
774  sce = &cpe->ch[ch];
775  s->cur_channel = start_ch + ch;
776  if (s->options.pred && s->coder->apply_main_pred)
777  s->coder->apply_main_pred(s, sce);
778  }
779  s->cur_channel = start_ch;
780  }
781  if (s->options.mid_side) { /* Mid/Side stereo */
782  if (s->options.mid_side == -1 && s->coder->search_for_ms)
783  s->coder->search_for_ms(s, cpe);
784  else if (cpe->common_window)
785  memset(cpe->ms_mask, 1, sizeof(cpe->ms_mask));
787  }
788  adjust_frame_information(cpe, chans);
789  if (s->options.ltp) { /* LTP */
790  for (ch = 0; ch < chans; ch++) {
791  sce = &cpe->ch[ch];
792  s->cur_channel = start_ch + ch;
793  if (s->coder->search_for_ltp)
794  s->coder->search_for_ltp(s, sce, cpe->common_window);
795  if (sce->ics.ltp.present) pred_mode = 1;
796  }
797  s->cur_channel = start_ch;
798  if (s->coder->adjust_common_ltp)
799  s->coder->adjust_common_ltp(s, cpe);
800  }
801  if (chans == 2) {
802  put_bits(&s->pb, 1, cpe->common_window);
803  if (cpe->common_window) {
804  put_ics_info(s, &cpe->ch[0].ics);
805  if (s->coder->encode_main_pred)
806  s->coder->encode_main_pred(s, &cpe->ch[0]);
807  if (s->coder->encode_ltp_info)
808  s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
809  encode_ms_info(&s->pb, cpe);
810  if (cpe->ms_mode) ms_mode = 1;
811  }
812  }
813  for (ch = 0; ch < chans; ch++) {
814  s->cur_channel = start_ch + ch;
815  encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
816  }
817  start_ch += chans;
818  }
819 
820  if (avctx->flags & AV_CODEC_FLAG_QSCALE) {
821  /* When using a constant Q-scale, don't mess with lambda */
822  break;
823  }
824 
825  /* rate control stuff
826  * allow between the nominal bitrate, and what psy's bit reservoir says to target
827  * but drift towards the nominal bitrate always
828  */
829  frame_bits = put_bits_count(&s->pb);
830  rate_bits = avctx->bit_rate * 1024 / avctx->sample_rate;
831  rate_bits = FFMIN(rate_bits, 6144 * s->channels - 3);
832  too_many_bits = FFMAX(target_bits, rate_bits);
833  too_many_bits = FFMIN(too_many_bits, 6144 * s->channels - 3);
834  too_few_bits = FFMIN(FFMAX(rate_bits - rate_bits/4, target_bits), too_many_bits);
835 
836  /* When using ABR, be strict (but only for increasing) */
837  too_few_bits = too_few_bits - too_few_bits/8;
838  too_many_bits = too_many_bits + too_many_bits/2;
839 
840  if ( its == 0 /* for steady-state Q-scale tracking */
841  || (its < 5 && (frame_bits < too_few_bits || frame_bits > too_many_bits))
842  || frame_bits >= 6144 * s->channels - 3 )
843  {
844  float ratio = ((float)rate_bits) / frame_bits;
845 
846  if (frame_bits >= too_few_bits && frame_bits <= too_many_bits) {
847  /*
848  * This path is for steady-state Q-scale tracking
849  * When frame bits fall within the stable range, we still need to adjust
850  * lambda to maintain it like so in a stable fashion (large jumps in lambda
851  * create artifacts and should be avoided), but slowly
852  */
853  ratio = sqrtf(sqrtf(ratio));
854  ratio = av_clipf(ratio, 0.9f, 1.1f);
855  } else {
856  /* Not so fast though */
857  ratio = sqrtf(ratio);
858  }
859  s->lambda = FFMIN(s->lambda * ratio, 65536.f);
860 
861  /* Keep iterating if we must reduce and lambda is in the sky */
862  if (ratio > 0.9f && ratio < 1.1f) {
863  break;
864  } else {
865  if (is_mode || ms_mode || tns_mode || pred_mode) {
866  for (i = 0; i < s->chan_map[0]; i++) {
867  // Must restore coeffs
868  chans = tag == TYPE_CPE ? 2 : 1;
869  cpe = &s->cpe[i];
870  for (ch = 0; ch < chans; ch++)
871  memcpy(cpe->ch[ch].coeffs, cpe->ch[ch].pcoeffs, sizeof(cpe->ch[ch].coeffs));
872  }
873  }
874  its++;
875  }
876  } else {
877  break;
878  }
879  } while (1);
880 
881  if (s->options.ltp && s->coder->ltp_insert_new_frame)
883 
884  put_bits(&s->pb, 3, TYPE_END);
885  flush_put_bits(&s->pb);
886 
888 
889  s->lambda_sum += s->lambda;
890  s->lambda_count++;
891 
892  ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
893  &avpkt->duration);
894 
895  avpkt->size = put_bits_count(&s->pb) >> 3;
896  *got_packet_ptr = 1;
897  return 0;
898 }
899 
901 {
902  AACEncContext *s = avctx->priv_data;
903 
904  av_log(avctx, AV_LOG_INFO, "Qavg: %.3f\n", s->lambda_sum / s->lambda_count);
905 
906  ff_mdct_end(&s->mdct1024);
907  ff_mdct_end(&s->mdct128);
908  ff_psy_end(&s->psy);
909  ff_lpc_end(&s->lpc);
910  if (s->psypp)
912  av_freep(&s->buffer.samples);
913  av_freep(&s->cpe);
914  av_freep(&s->fdsp);
915  ff_af_queue_close(&s->afq);
916  return 0;
917 }
918 
920 {
921  int ret = 0;
922 
924  if (!s->fdsp)
925  return AVERROR(ENOMEM);
926 
927  // window init
932 
933  if ((ret = ff_mdct_init(&s->mdct1024, 11, 0, 32768.0)) < 0)
934  return ret;
935  if ((ret = ff_mdct_init(&s->mdct128, 8, 0, 32768.0)) < 0)
936  return ret;
937 
938  return 0;
939 }
940 
942 {
943  int ch;
944  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->buffer.samples, s->channels, 3 * 1024 * sizeof(s->buffer.samples[0]), alloc_fail);
945  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->cpe, s->chan_map[0], sizeof(ChannelElement), alloc_fail);
946 
947  for(ch = 0; ch < s->channels; ch++)
948  s->planar_samples[ch] = s->buffer.samples + 3 * 1024 * ch;
949 
950  return 0;
951 alloc_fail:
952  return AVERROR(ENOMEM);
953 }
954 
956 {
958 }
959 
961 {
962  AACEncContext *s = avctx->priv_data;
963  int i, ret = 0;
964  const uint8_t *sizes[2];
965  uint8_t grouping[AAC_MAX_CHANNELS];
966  int lengths[2];
967 
968  /* Constants */
969  s->last_frame_pb_count = 0;
970  avctx->frame_size = 1024;
971  avctx->initial_padding = 1024;
972  s->lambda = avctx->global_quality > 0 ? avctx->global_quality : 120;
973 
974  /* Channel map and unspecified bitrate guessing */
975  s->channels = avctx->channels;
976 
977  s->needs_pce = 1;
978  for (i = 0; i < FF_ARRAY_ELEMS(aac_normal_chan_layouts); i++) {
979  if (avctx->channel_layout == aac_normal_chan_layouts[i]) {
980  s->needs_pce = s->options.pce;
981  break;
982  }
983  }
984 
985  if (s->needs_pce) {
986  char buf[64];
987  for (i = 0; i < FF_ARRAY_ELEMS(aac_pce_configs); i++)
988  if (avctx->channel_layout == aac_pce_configs[i].layout)
989  break;
990  av_get_channel_layout_string(buf, sizeof(buf), -1, avctx->channel_layout);
991  ERROR_IF(i == FF_ARRAY_ELEMS(aac_pce_configs), "Unsupported channel layout \"%s\"\n", buf);
992  av_log(avctx, AV_LOG_INFO, "Using a PCE to encode channel layout \"%s\"\n", buf);
993  s->pce = aac_pce_configs[i];
994  s->reorder_map = s->pce.reorder_map;
995  s->chan_map = s->pce.config_map;
996  } else {
997  s->reorder_map = aac_chan_maps[s->channels - 1];
998  s->chan_map = aac_chan_configs[s->channels - 1];
999  }
1000 
1001  if (!avctx->bit_rate) {
1002  for (i = 1; i <= s->chan_map[0]; i++) {
1003  avctx->bit_rate += s->chan_map[i] == TYPE_CPE ? 128000 : /* Pair */
1004  s->chan_map[i] == TYPE_LFE ? 16000 : /* LFE */
1005  69000 ; /* SCE */
1006  }
1007  }
1008 
1009  /* Samplerate */
1010  for (i = 0; i < 16; i++)
1012  break;
1013  s->samplerate_index = i;
1014  ERROR_IF(s->samplerate_index == 16 ||
1017  "Unsupported sample rate %d\n", avctx->sample_rate);
1018 
1019  /* Bitrate limiting */
1020  WARN_IF(1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * s->channels,
1021  "Too many bits %f > %d per frame requested, clamping to max\n",
1022  1024.0 * avctx->bit_rate / avctx->sample_rate,
1023  6144 * s->channels);
1024  avctx->bit_rate = (int64_t)FFMIN(6144 * s->channels / 1024.0 * avctx->sample_rate,
1025  avctx->bit_rate);
1026 
1027  /* Profile and option setting */
1028  avctx->profile = avctx->profile == FF_PROFILE_UNKNOWN ? FF_PROFILE_AAC_LOW :
1029  avctx->profile;
1030  for (i = 0; i < FF_ARRAY_ELEMS(aacenc_profiles); i++)
1031  if (avctx->profile == aacenc_profiles[i])
1032  break;
1033  if (avctx->profile == FF_PROFILE_MPEG2_AAC_LOW) {
1034  avctx->profile = FF_PROFILE_AAC_LOW;
1035  ERROR_IF(s->options.pred,
1036  "Main prediction unavailable in the \"mpeg2_aac_low\" profile\n");
1037  ERROR_IF(s->options.ltp,
1038  "LTP prediction unavailable in the \"mpeg2_aac_low\" profile\n");
1039  WARN_IF(s->options.pns,
1040  "PNS unavailable in the \"mpeg2_aac_low\" profile, turning off\n");
1041  s->options.pns = 0;
1042  } else if (avctx->profile == FF_PROFILE_AAC_LTP) {
1043  s->options.ltp = 1;
1044  ERROR_IF(s->options.pred,
1045  "Main prediction unavailable in the \"aac_ltp\" profile\n");
1046  } else if (avctx->profile == FF_PROFILE_AAC_MAIN) {
1047  s->options.pred = 1;
1048  ERROR_IF(s->options.ltp,
1049  "LTP prediction unavailable in the \"aac_main\" profile\n");
1050  } else if (s->options.ltp) {
1051  avctx->profile = FF_PROFILE_AAC_LTP;
1052  WARN_IF(1,
1053  "Chainging profile to \"aac_ltp\"\n");
1054  ERROR_IF(s->options.pred,
1055  "Main prediction unavailable in the \"aac_ltp\" profile\n");
1056  } else if (s->options.pred) {
1057  avctx->profile = FF_PROFILE_AAC_MAIN;
1058  WARN_IF(1,
1059  "Chainging profile to \"aac_main\"\n");
1060  ERROR_IF(s->options.ltp,
1061  "LTP prediction unavailable in the \"aac_main\" profile\n");
1062  }
1063  s->profile = avctx->profile;
1064 
1065  /* Coder limitations */
1066  s->coder = &ff_aac_coders[s->options.coder];
1067  if (s->options.coder == AAC_CODER_ANMR) {
1069  "The ANMR coder is considered experimental, add -strict -2 to enable!\n");
1070  s->options.intensity_stereo = 0;
1071  s->options.pns = 0;
1072  }
1074  "The LPT profile requires experimental compliance, add -strict -2 to enable!\n");
1075 
1076  /* M/S introduces horrible artifacts with multichannel files, this is temporary */
1077  if (s->channels > 3)
1078  s->options.mid_side = 0;
1079 
1080  if ((ret = dsp_init(avctx, s)) < 0)
1081  goto fail;
1082 
1083  if ((ret = alloc_buffers(avctx, s)) < 0)
1084  goto fail;
1085 
1086  if ((ret = put_audio_specific_config(avctx)))
1087  goto fail;
1088 
1089  sizes[0] = ff_aac_swb_size_1024[s->samplerate_index];
1090  sizes[1] = ff_aac_swb_size_128[s->samplerate_index];
1091  lengths[0] = ff_aac_num_swb_1024[s->samplerate_index];
1092  lengths[1] = ff_aac_num_swb_128[s->samplerate_index];
1093  for (i = 0; i < s->chan_map[0]; i++)
1094  grouping[i] = s->chan_map[i + 1] == TYPE_CPE;
1095  if ((ret = ff_psy_init(&s->psy, avctx, 2, sizes, lengths,
1096  s->chan_map[0], grouping)) < 0)
1097  goto fail;
1098  s->psypp = ff_psy_preprocess_init(avctx);
1100  s->random_state = 0x1f2e3d4c;
1101 
1102  s->abs_pow34 = abs_pow34_v;
1104 
1105  if (ARCH_X86)
1107 
1108  if (HAVE_MIPSDSP)
1110 
1112  return AVERROR_UNKNOWN;
1113 
1114  ff_af_queue_init(avctx, &s->afq);
1115 
1116  return 0;
1117 fail:
1118  aac_encode_end(avctx);
1119  return ret;
1120 }
1121 
1122 #define AACENC_FLAGS AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
1123 static const AVOption aacenc_options[] = {
1124  {"aac_coder", "Coding algorithm", offsetof(AACEncContext, options.coder), AV_OPT_TYPE_INT, {.i64 = AAC_CODER_FAST}, 0, AAC_CODER_NB-1, AACENC_FLAGS, "coder"},
1125  {"anmr", "ANMR method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_ANMR}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1126  {"twoloop", "Two loop searching method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_TWOLOOP}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1127  {"fast", "Default fast search", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_FAST}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1128  {"aac_ms", "Force M/S stereo coding", offsetof(AACEncContext, options.mid_side), AV_OPT_TYPE_BOOL, {.i64 = -1}, -1, 1, AACENC_FLAGS},
1129  {"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1130  {"aac_pns", "Perceptual noise substitution", offsetof(AACEncContext, options.pns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1131  {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1132  {"aac_ltp", "Long term prediction", offsetof(AACEncContext, options.ltp), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1133  {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1134  {"aac_pce", "Forces the use of PCEs", offsetof(AACEncContext, options.pce), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1136  {NULL}
1137 };
1138 
1139 static const AVClass aacenc_class = {
1140  .class_name = "AAC encoder",
1141  .item_name = av_default_item_name,
1142  .option = aacenc_options,
1143  .version = LIBAVUTIL_VERSION_INT,
1144 };
1145 
1147  { "b", "0" },
1148  { NULL }
1149 };
1150 
1152  .name = "aac",
1153  .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
1154  .type = AVMEDIA_TYPE_AUDIO,
1155  .id = AV_CODEC_ID_AAC,
1156  .priv_data_size = sizeof(AACEncContext),
1157  .init = aac_encode_init,
1158  .encode2 = aac_encode_frame,
1159  .close = aac_encode_end,
1160  .defaults = aac_encode_defaults,
1161  .supported_samplerates = mpeg4audio_sample_rates,
1162  .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
1164  .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
1166  .priv_class = &aacenc_class,
1167 };
struct FFPsyContext::@114 bitres
#define FF_COMPLIANCE_EXPERIMENTAL
Allow nonstandardized experimental things.
Definition: avcodec.h:1594
float, planar
Definition: samplefmt.h:69
void ff_af_queue_remove(AudioFrameQueue *afq, int nb_samples, int64_t *pts, int64_t *duration)
Remove frame(s) from the queue.
void ff_quantize_band_cost_cache_init(struct AACEncContext *s)
Definition: aacenc.c:127
#define NULL
Definition: coverity.c:32
const AACCoefficientsEncoder * coder
Definition: aacenc.h:397
Band types following are encoded differently from others.
Definition: aac.h:86
static const uint8_t aac_chan_configs[AAC_MAX_CHANNELS][6]
default channel configurations
Definition: aacenctab.h:58
uint8_t use_kb_window[2]
If set, use Kaiser-Bessel window, otherwise use a sine window.
Definition: aac.h:177
int coder
Definition: aacenc.h:44
This structure describes decoded (raw) audio or video data.
Definition: frame.h:300
#define FF_ALLOCZ_ARRAY_OR_GOTO(ctx, p, nelem, elsize, label)
Definition: internal.h:167
if(ret< 0)
Definition: vf_mcdeint.c:279
int grouping[8]
window grouping (for e.g. AAC)
Definition: psymodel.h:81
void(* search_for_ms)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:77
AVOption.
Definition: opt.h:246
enum RawDataBlockType cur_type
channel group type cur_channel belongs to
Definition: aacenc.h:404
uint8_t ** bands
scalefactor band sizes for possible frame sizes
Definition: psymodel.h:98
Definition: aac.h:224
AACQuantizeBandCostCacheEntry quantize_band_cost_cache[256][128]
memoization area for quantize_band_cost
Definition: aacenc.h:411
static void abs_pow34_v(float *out, const float *in, const int size)
Definition: aacenc_utils.h:40
static const AVClass aacenc_class
Definition: aacenc.c:1139
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:208
av_cold void ff_kbd_window_init(float *window, float alpha, int n)
Generate a Kaiser-Bessel Derived Window.
Definition: kbdwin.c:26
int64_t bit_rate
the average bitrate
Definition: avcodec.h:576
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
#define SCALE_DIFF_ZERO
codebook index corresponding to zero scalefactor indices difference
Definition: aac.h:152
uint8_t window_clipping[8]
set if a certain window is near clipping
Definition: aac.h:191
Definition: aac.h:63
const char * g
Definition: vf_curves.c:115
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
Definition: aac.h:57
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx)
Cleanup audio preprocessing module.
Definition: psymodel.c:152
#define WARN_IF(cond,...)
Definition: aacenc_utils.h:274
int size
Definition: packet.h:356
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:235
const int ff_aac_swb_size_1024_len
Definition: aacenctab.c:108
void avpriv_align_put_bits(PutBitContext *s)
Pad the bitstream with zeros up to the next byte boundary.
Definition: bitstream.c:48
static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
Encode ics_info element.
Definition: aacenc.c:222
void(* search_for_tns)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:75
int common_window
Set if channels share a common &#39;IndividualChannelStream&#39; in bitstream.
Definition: aac.h:278
int alloc
number of bits allocated by the psy, or -1 if no allocation was done
Definition: psymodel.h:105
const uint8_t * ff_aac_swb_size_1024[]
Definition: aacenctab.c:99
#define FF_PROFILE_AAC_MAIN
Definition: avcodec.h:1863
int lambda_count
count(lambda), for Qvg reporting
Definition: aacenc.h:403
av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, int num_lens, const uint8_t **bands, const int *num_bands, int num_groups, const uint8_t *group_map)
Initialize psychoacoustic model.
Definition: psymodel.c:31
uint8_t ms_mask[128]
Set if mid/side stereo is used for each scalefactor window band.
Definition: aac.h:281
float lambda
Definition: aacenc.h:400
#define NOISE_PRE
preamble for NOISE_BT, put in bitstream with the first noise band
Definition: aac.h:156
int profile
profile
Definition: avcodec.h:1859
AVCodec.
Definition: codec.h:190
static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
Encode spectral coefficients processed by psychoacoustic model.
Definition: aacenc.c:440
Spectral data are scaled white noise not coded in the bitstream.
Definition: aac.h:87
int * num_bands
number of scalefactor bands for possible frame sizes
Definition: psymodel.h:99
static AVOnce aac_table_init
Definition: aacenc.c:52
static int put_audio_specific_config(AVCodecContext *avctx)
Make AAC audio config object.
Definition: aacenc.c:95
struct AACEncContext::@6 buffer
void(* apply_tns_filt)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:69
const uint8_t ff_aac_num_swb_128[]
Definition: aactab.c:61
void(* search_for_quantizers)(AVCodecContext *avctx, struct AACEncContext *s, SingleChannelElement *sce, const float lambda)
Definition: aacenc.h:57
INTFLOAT pcoeffs[1024]
coefficients for IMDCT, pristine
Definition: aac.h:261
const uint16_t * swb_offset
table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular wind...
Definition: aac.h:181
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_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:75
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
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:32
AACEncOptions options
encoding options
Definition: aacenc.h:378
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:40
AAC encoder context.
Definition: aacenc.h:376
int num_ele[4]
front, side, back, lfe
Definition: aacenc.h:95
uint8_t
#define av_cold
Definition: attributes.h:88
void(* search_for_pns)(struct AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
Definition: aacenc.h:73
AVOptions.
int intensity_stereo
Definition: aacenc.h:51
#define WINDOW_FUNC(type)
Definition: aacenc.c:136
void(* update_ltp)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:70
LPCContext lpc
used by TNS
Definition: aacenc.h:388
void ff_aac_coder_init_mips(AACEncContext *c)
SingleChannelElement ch[2]
Definition: aac.h:284
int samplerate_index
MPEG-4 samplerate index.
Definition: aacenc.h:389
#define f(width, name)
Definition: cbs_vp9.c:255
Definition: aac.h:59
av_cold void ff_af_queue_init(AVCodecContext *avctx, AudioFrameQueue *afq)
Initialize AudioFrameQueue.
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: packet.h:373
const uint8_t * chan_map
channel configuration map
Definition: aacenc.h:392
TemporalNoiseShaping tns
Definition: aac.h:250
const uint8_t ff_aac_scalefactor_bits[121]
Definition: aactab.c:92
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:627
AudioFrameQueue afq
Definition: aacenc.h:406
const AACCoefficientsEncoder ff_aac_coders[AAC_CODER_NB]
Definition: aaccoder.c:897
static AVFrame * frame
static const uint8_t aac_chan_maps[AAC_MAX_CHANNELS][AAC_MAX_CHANNELS]
Table to remap channels from libavcodec&#39;s default order to AAC order.
Definition: aacenctab.h:72
#define FF_PROFILE_AAC_LTP
Definition: avcodec.h:1866
uint8_t * data
Definition: packet.h:355
static void put_pce(PutBitContext *pb, AVCodecContext *avctx)
Definition: aacenc.c:54
const uint8_t * ff_aac_swb_size_128[]
Definition: aacenctab.c:91
uint32_t tag
Definition: movenc.c:1532
Scalefactor data are intensity stereo positions (in phase).
Definition: aac.h:89
#define max(a, b)
Definition: cuda_runtime.h:33
int profile
copied from avctx
Definition: aacenc.h:386
channels
Definition: aptx.h:33
#define AVOnce
Definition: thread.h:172
uint8_t reorder_map[16]
maps channels from lavc to aac order
Definition: aacenc.h:99
static void adjust_frame_information(ChannelElement *cpe, int chans)
Produce integer coefficients from scalefactors provided by the model.
Definition: aacenc.c:257
#define av_log(a,...)
static const AVOption aacenc_options[]
Definition: aacenc.c:1123
int64_t layout
Definition: aacenc.h:94
const uint8_t * reorder_map
lavc to aac reorder map
Definition: aacenc.h:391
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:269
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
#define R
Definition: huffyuvdsp.h:34
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:135
void(* encode_ltp_info)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:64
#define ARCH_X86
Definition: config.h:38
static const int sizes[][2]
Definition: img2dec.c:53
#define AVERROR(e)
Definition: error.h:43
const uint8_t ff_aac_num_swb_1024[]
Definition: aactab.c:45
#define FF_PROFILE_MPEG2_AAC_LOW
Definition: avcodec.h:1871
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:186
#define FF_AAC_PROFILE_OPTS
Definition: profiles.h:28
float is_ener[128]
Intensity stereo pos (used by encoder)
Definition: aac.h:259
int initial_padding
Audio only.
Definition: avcodec.h:2060
static const AACPCEInfo aac_pce_configs[]
List of PCE (Program Configuration Element) for the channel layouts listed in channel_layout.h.
Definition: aacenc.h:137
float ff_aac_kbd_long_1024[1024]
Definition: aactab.c:38
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:606
void(* mdct_calc)(struct FFTContext *s, FFTSample *output, const FFTSample *input)
Definition: fft.h:109
static const int mpeg4audio_sample_rates[16]
Definition: aacenctab.h:85
int amp[4]
Definition: aac.h:228
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:239
const char * name
Name of the codec implementation.
Definition: codec.h:197
int num_windows
number of windows in a frame
Definition: psymodel.h:80
static void copy_input_samples(AACEncContext *s, const AVFrame *frame)
Definition: aacenc.c:535
uint8_t max_sfb
number of scalefactor bands per group
Definition: aac.h:175
static const AVCodecDefault defaults[]
Definition: amfenc_h264.c:361
void(* adjust_common_ltp)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:67
#define ff_mdct_init
Definition: fft.h:169
Definition: aac.h:62
int num_swb
number of scalefactor window bands
Definition: aac.h:183
int ff_af_queue_add(AudioFrameQueue *afq, const AVFrame *f)
Add a frame to the queue.
#define FFMAX(a, b)
Definition: common.h:94
void(* mark_pns)(struct AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
Definition: aacenc.h:74
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(const uint8_t *) pi - 0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(const int16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(const int16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(const int32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(const int32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(const int64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0f/(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0/(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float *) pi *(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double *) pi *(UINT64_C(1)<< 63))) #define FMT_PAIR_FUNC(out, in) static conv_func_type *const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB *AV_SAMPLE_FMT_NB]={ FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64), };static void cpy1(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, len);} static void cpy2(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 2 *len);} static void cpy4(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 4 *len);} static void cpy8(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 8 *len);} AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, const int *ch_map, int flags) { AudioConvert *ctx;conv_func_type *f=fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt)+AV_SAMPLE_FMT_NB *av_get_packed_sample_fmt(in_fmt)];if(!f) return NULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) return NULL;if(channels==1){ in_fmt=av_get_planar_sample_fmt(in_fmt);out_fmt=av_get_planar_sample_fmt(out_fmt);} ctx->channels=channels;ctx->conv_f=f;ctx->ch_map=ch_map;if(in_fmt==AV_SAMPLE_FMT_U8||in_fmt==AV_SAMPLE_FMT_U8P) memset(ctx->silence, 0x80, sizeof(ctx->silence));if(out_fmt==in_fmt &&!ch_map) { switch(av_get_bytes_per_sample(in_fmt)){ case 1:ctx->simd_f=cpy1;break;case 2:ctx->simd_f=cpy2;break;case 4:ctx->simd_f=cpy4;break;case 8:ctx->simd_f=cpy8;break;} } if(HAVE_X86ASM &&HAVE_MMX) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels);if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels);if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels);return ctx;} void swri_audio_convert_free(AudioConvert **ctx) { av_freep(ctx);} int swri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, int len) { int ch;int off=0;const int os=(out->planar ? 1 :out->ch_count) *out->bps;unsigned misaligned=0;av_assert0(ctx->channels==out->ch_count);if(ctx->in_simd_align_mask) { int planes=in->planar ? in->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) in->ch[ch];misaligned|=m &ctx->in_simd_align_mask;} if(ctx->out_simd_align_mask) { int planes=out->planar ? out->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) out->ch[ch];misaligned|=m &ctx->out_simd_align_mask;} if(ctx->simd_f &&!ctx->ch_map &&!misaligned){ off=len &~15;av_assert1(off >=0);av_assert1(off<=len);av_assert2(ctx->channels==SWR_CH_MAX||!in->ch[ctx->channels]);if(off >0){ if(out->planar==in->planar){ int planes=out->planar ? out->ch_count :1;for(ch=0;ch< planes;ch++){ ctx->simd_f(out-> ch ch
Definition: audioconvert.c:56
#define fail()
Definition: checkasm.h:123
int index[4][8]
front, side, back, lfe
Definition: aacenc.h:97
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:1237
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:67
#define AACENC_FLAGS
Definition: aacenc.c:1122
INTFLOAT ret_buf[2048]
PCM output buffer.
Definition: aac.h:264
void(* set_special_band_scalefactors)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:72
const char * name
Definition: qsvenc.c:46
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:333
av_cold void ff_lpc_end(LPCContext *s)
Uninitialize LPCContext.
Definition: lpc.c:322
#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:80
int cur_channel
current channel for coder context
Definition: aacenc.h:398
int last_frame_pb_count
number of bits for the previous frame
Definition: aacenc.h:401
#define FFMIN(a, b)
Definition: common.h:96
static void apply_intensity_stereo(ChannelElement *cpe)
Definition: aacenc.c:307
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: aacenc.c:557
void(* quant_bands)(int *out, const float *in, const float *scaled, int size, int is_signed, int maxval, const float Q34, const float rounding)
Definition: aacenc.h:414
uint8_t w
Definition: llviddspenc.c:38
#define FF_PROFILE_AAC_LOW
Definition: avcodec.h:1864
static const AVCodecDefault aac_encode_defaults[]
Definition: aacenc.c:1146
#define FF_PROFILE_UNKNOWN
Definition: avcodec.h:1860
int pos[4]
Definition: aac.h:227
int channels
channel count
Definition: aacenc.h:390
#define s(width, name)
Definition: cbs_vp9.c:257
AAC definitions and structures.
const uint8_t ff_tns_max_bands_1024[]
Definition: aactab.c:1394
static void quantize_bands(int *out, const float *in, const float *scaled, int size, int is_signed, int maxval, const float Q34, const float rounding)
Definition: aacenc_utils.h:65
FFTContext mdct128
short (128 samples) frame transform context
Definition: aacenc.h:381
PutBitContext pb
Definition: aacenc.h:379
static void(*const apply_window[4])(AVFloatDSPContext *fdsp, SingleChannelElement *sce, const float *audio)
Definition: aacenc.c:192
#define L(x)
Definition: vp56_arith.h:36
AVFloatDSPContext * fdsp
Definition: aacenc.h:382
int mid_side
Definition: aacenc.h:50
#define FF_ARRAY_ELEMS(a)
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.
static av_cold int aac_encode_end(AVCodecContext *avctx)
Definition: aacenc.c:900
void(* search_for_is)(struct AACEncContext *s, AVCodecContext *avctx, ChannelElement *cpe)
Definition: aacenc.h:78
void ff_aac_dsp_init_x86(AACEncContext *s)
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:1206
void(* search_for_ltp)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:76
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
#define AV_ONCE_INIT
Definition: thread.h:173
#define CLIP_AVOIDANCE_FACTOR
Definition: aac.h:53
Libavcodec external API header.
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
typedef void(RENAME(mix_any_func_type))
Temporal Noise Shaping.
Definition: aac.h:198
int sample_rate
samples per second
Definition: avcodec.h:1186
float ff_aac_kbd_short_128[128]
Definition: aactab.c:39
static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
Encode MS data.
Definition: aacenc.c:243
void(* ltp_insert_new_frame)(struct AACEncContext *s)
Definition: aacenc.h:71
void(* search_for_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:79
main external API structure.
Definition: avcodec.h:526
int pairing[3][8]
front, side, back
Definition: aacenc.h:96
int bits
number of bits used in the bitresevoir
Definition: psymodel.h:104
#define NOISE_PRE_BITS
length of preamble
Definition: aac.h:157
Levinson-Durbin recursion.
Definition: lpc.h:47
void(* apply_main_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:68
IndividualChannelStream ics
Definition: aac.h:249
int extradata_size
Definition: avcodec.h:628
uint8_t group_len[8]
Definition: aac.h:179
Replacements for frequently missing libm functions.
float lambda_sum
sum(lambda), for Qvg reporting
Definition: aacenc.h:402
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)\", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));return ff_convert_dither(ac-> in
Describe the class of an AVClass context structure.
Definition: log.h:67
static void put_bitstream_info(AACEncContext *s, const char *name)
Write some auxiliary information about the created AAC file.
Definition: aacenc.c:514
const int ff_aac_swb_size_128_len
Definition: aacenctab.c:107
void(* encode_main_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:65
void(* adjust_common_pred)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:66
int window_shape
window shape (sine/KBD/whatever)
Definition: psymodel.h:79
static void encode_pulses(AACEncContext *s, Pulse *pulse)
Encode pulse data.
Definition: aacenc.c:421
uint16_t quantize_band_cost_cache_generation
Definition: aacenc.h:410
static av_cold void aac_encode_init_tables(void)
Definition: aacenc.c:955
const uint8_t * swb_sizes
table of scalefactor band sizes for a particular window
Definition: aac.h:182
#define TNS_MAX_ORDER
Definition: aac.h:50
FFPsyContext psy
Definition: aacenc.h:395
const uint32_t ff_aac_scalefactor_code[121]
Definition: aactab.c:73
LongTermPrediction ltp
Definition: aac.h:180
static av_cold int alloc_buffers(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:941
const struct FFPsyModel * model
encoder-specific model functions
Definition: psymodel.h:91
av_cold int ff_lpc_init(LPCContext *s, int blocksize, int max_order, enum FFLPCType lpc_type)
Initialize LPCContext.
Definition: lpc.c:300
#define AAC_MAX_CHANNELS
Definition: aacenctab.h:39
int needs_pce
flag for non-standard layout
Definition: aacenc.h:387
FFPsyWindowInfo(* window)(FFPsyContext *ctx, const float *audio, const float *la, int channel, int prev_type)
Suggest window sequence for channel.
Definition: psymodel.h:129
int ms_mode
Signals mid/side stereo flags coding mode (used by encoder)
Definition: aac.h:279
AAC encoder data.
const uint8_t ff_tns_max_bands_128[]
Definition: aactab.c:1406
struct FFPsyPreprocessContext * psypp
Definition: aacenc.h:396
#define NOISE_OFFSET
subtracted from global gain, used as offset for the preamble
Definition: aac.h:158
int global_quality
Global quality for codecs which cannot change it per frame.
Definition: avcodec.h:592
uint8_t zeroes[128]
band is not coded (used by encoder)
Definition: aac.h:257
int sf_idx[128]
scalefactor indices (used by encoder)
Definition: aac.h:256
AVCodec ff_aac_encoder
Definition: aacenc.c:1151
uint8_t is_mode
Set if any bands have been encoded using intensity stereo (used by encoder)
Definition: aac.h:280
INTFLOAT coeffs[1024]
coefficients for IMDCT, maybe processed
Definition: aac.h:262
const int avpriv_mpeg4audio_sample_rates[16]
Definition: mpeg4audio.c:62
void(* quantize_and_encode_band)(struct AACEncContext *s, PutBitContext *pb, const float *in, float *out, int size, int scale_idx, int cb, const float lambda, int rtz)
Definition: aacenc.h:61
Scalefactor data are intensity stereo positions (out of phase).
Definition: aac.h:88
Y Spectral Band Replication.
Definition: mpeg4audio.h:94
const OptionDef options[]
Definition: ffmpeg_opt.c:3388
float * samples
Definition: aacenc.h:419
uint8_t prediction_used[41]
Definition: aac.h:190
static av_cold int aac_encode_init(AVCodecContext *avctx)
Definition: aacenc.c:960
common internal api header.
AACPCEInfo pce
PCE data, if needed.
Definition: aacenc.h:383
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:101
AAC encoder utilities.
Single Channel Element - used for both SCE and LFE elements.
Definition: aac.h:248
windowing related information
Definition: psymodel.h:77
#define ff_mdct_end
Definition: fft.h:170
av_cold struct FFPsyPreprocessContext * ff_psy_preprocess_init(AVCodecContext *avctx)
psychoacoustic model audio preprocessing initialization
Definition: psymodel.c:103
const uint16_t *const ff_swb_offset_1024[]
Definition: aactab.c:1336
uint8_t config_map[16]
configs the encoder&#39;s channel specific settings
Definition: aacenc.h:98
void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx, float **audio, int channels)
Preprocess several channel in audio frame in order to compress it better.
Definition: psymodel.c:139
static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce)
Encode scalefactors.
Definition: aacenc.c:385
float * planar_samples[16]
saved preprocessed input
Definition: aacenc.h:384
ChannelElement * cpe
channel elements
Definition: aacenc.h:394
Individual Channel Stream.
Definition: aac.h:174
float clip_avoidance_factor
set if any window is near clipping to the necessary atennuation factor to avoid it ...
Definition: aac.h:192
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:48
#define AVERROR_UNKNOWN
Unknown error, typically from an external library.
Definition: error.h:71
#define ERROR_IF(cond,...)
Definition: aacenc_utils.h:268
static void ff_aac_tableinit(void)
Definition: aactab.h:45
channel element - generic struct for SCE/CPE/CCE/LFE
Definition: aac.h:275
void * priv_data
Definition: avcodec.h:553
int start
Definition: aac.h:226
FFTContext mdct1024
long (1024 samples) frame transform context
Definition: aacenc.h:380
int random_state
Definition: aacenc.h:399
static av_always_inline int diff(const uint32_t a, const uint32_t b)
int channels
number of audio channels
Definition: avcodec.h:1187
int num_pulse
Definition: aac.h:225
AAC_FLOAT lcoeffs[1024]
MDCT of LTP coefficients (used by encoder)
Definition: aac.h:266
static int ff_thread_once(char *control, void(*routine)(void))
Definition: thread.h:175
static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
Encode scalefactor band coding type.
Definition: aacenc.c:371
void(* analyze)(FFPsyContext *ctx, int channel, const float **coeffs, const FFPsyWindowInfo *wi)
Perform psychoacoustic analysis and set band info (threshold, energy) for a group of channels...
Definition: psymodel.h:139
static void apply_mid_side_stereo(ChannelElement *cpe)
Definition: aacenc.c:336
static const int64_t aac_normal_chan_layouts[7]
Definition: aacenctab.h:47
#define HAVE_MIPSDSP
Definition: config.h:80
void ff_af_queue_close(AudioFrameQueue *afq)
Close AudioFrameQueue.
enum BandType band_type[128]
band types
Definition: aac.h:252
#define LIBAVCODEC_IDENT
Definition: version.h:42
void avpriv_put_string(PutBitContext *pb, const char *string, int terminate_string)
Put the string string in the bitstream.
Definition: bitstream.c:53
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:731
int frame_number
Frame counter, set by libavcodec.
Definition: avcodec.h:1217
FILE * out
Definition: movenc.c:54
#define av_freep(p)
void(* encode_tns_info)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:63
static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce, int common_window)
Encode one channel of audio data.
Definition: aacenc.c:488
int8_t used[MAX_LTP_LONG_SFB]
Definition: aac.h:168
static void apply_window_and_mdct(AACEncContext *s, SingleChannelElement *sce, float *audio)
Definition: aacenc.c:201
const uint16_t *const ff_swb_offset_128[]
Definition: aactab.c:1368
int8_t present
Definition: aac.h:164
uint8_t is_mask[128]
Set if intensity stereo is used (used by encoder)
Definition: aac.h:282
static const int aacenc_profiles[]
Definition: aacenctab.h:132
void(* abs_pow34)(float *out, const float *in, const int size)
Definition: aacenc.h:413
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:347
AAC data declarations.
av_cold void ff_psy_end(FFPsyContext *ctx)
Cleanup model context at the end.
Definition: psymodel.c:83
This structure stores compressed data.
Definition: packet.h:332
static void avoid_clipping(AACEncContext *s, SingleChannelElement *sce)
Downscale spectral coefficients for near-clipping windows to avoid artifacts.
Definition: aacenc.c:468
int window_type[3]
window type (short/long/transitional, etc.) - current, previous and next
Definition: psymodel.h:78
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:366
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:1589
static av_cold int dsp_init(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:919
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: packet.h:348
for(j=16;j >0;--j)
int pred
Definition: aacenc.h:49
void AAC_RENAME() ff_init_ff_sine_windows(int index)
initialize the specified entry of ff_sine_windows
float clipping[8]
maximum absolute normalized intensity in the given window for clip avoidance
Definition: psymodel.h:82
void(* encode_window_bands_info)(struct AACEncContext *s, SingleChannelElement *sce, int win, int group_len, const float lambda)
Definition: aacenc.h:59
bitstream writer API