FFmpeg  3.4.9
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/thread.h"
35 #include "libavutil/float_dsp.h"
36 #include "libavutil/opt.h"
37 #include "avcodec.h"
38 #include "put_bits.h"
39 #include "internal.h"
40 #include "mpeg4audio.h"
41 #include "kbdwin.h"
42 #include "sinewin.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 /**
55  * Make AAC audio config object.
56  * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
57  */
59 {
60  PutBitContext pb;
61  AACEncContext *s = avctx->priv_data;
62  int channels = s->channels - (s->channels == 8 ? 1 : 0);
63 
64  init_put_bits(&pb, avctx->extradata, avctx->extradata_size);
65  put_bits(&pb, 5, s->profile+1); //profile
66  put_bits(&pb, 4, s->samplerate_index); //sample rate index
67  put_bits(&pb, 4, channels);
68  //GASpecificConfig
69  put_bits(&pb, 1, 0); //frame length - 1024 samples
70  put_bits(&pb, 1, 0); //does not depend on core coder
71  put_bits(&pb, 1, 0); //is not extension
72 
73  //Explicitly Mark SBR absent
74  put_bits(&pb, 11, 0x2b7); //sync extension
75  put_bits(&pb, 5, AOT_SBR);
76  put_bits(&pb, 1, 0);
77  flush_put_bits(&pb);
78 }
79 
81 {
84  memset(s->quantize_band_cost_cache, 0, sizeof(s->quantize_band_cost_cache));
86  }
87 }
88 
89 #define WINDOW_FUNC(type) \
90 static void apply_ ##type ##_window(AVFloatDSPContext *fdsp, \
91  SingleChannelElement *sce, \
92  const float *audio)
93 
94 WINDOW_FUNC(only_long)
95 {
96  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
97  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
98  float *out = sce->ret_buf;
99 
100  fdsp->vector_fmul (out, audio, lwindow, 1024);
101  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, pwindow, 1024);
102 }
103 
104 WINDOW_FUNC(long_start)
105 {
106  const float *lwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024;
107  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
108  float *out = sce->ret_buf;
109 
110  fdsp->vector_fmul(out, audio, lwindow, 1024);
111  memcpy(out + 1024, audio + 1024, sizeof(out[0]) * 448);
112  fdsp->vector_fmul_reverse(out + 1024 + 448, audio + 1024 + 448, swindow, 128);
113  memset(out + 1024 + 576, 0, sizeof(out[0]) * 448);
114 }
115 
116 WINDOW_FUNC(long_stop)
117 {
118  const float *lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
119  const float *swindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
120  float *out = sce->ret_buf;
121 
122  memset(out, 0, sizeof(out[0]) * 448);
123  fdsp->vector_fmul(out + 448, audio + 448, swindow, 128);
124  memcpy(out + 576, audio + 576, sizeof(out[0]) * 448);
125  fdsp->vector_fmul_reverse(out + 1024, audio + 1024, lwindow, 1024);
126 }
127 
128 WINDOW_FUNC(eight_short)
129 {
130  const float *swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
131  const float *pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
132  const float *in = audio + 448;
133  float *out = sce->ret_buf;
134  int w;
135 
136  for (w = 0; w < 8; w++) {
137  fdsp->vector_fmul (out, in, w ? pwindow : swindow, 128);
138  out += 128;
139  in += 128;
140  fdsp->vector_fmul_reverse(out, in, swindow, 128);
141  out += 128;
142  }
143 }
144 
145 static void (*const apply_window[4])(AVFloatDSPContext *fdsp,
147  const float *audio) = {
148  [ONLY_LONG_SEQUENCE] = apply_only_long_window,
149  [LONG_START_SEQUENCE] = apply_long_start_window,
150  [EIGHT_SHORT_SEQUENCE] = apply_eight_short_window,
151  [LONG_STOP_SEQUENCE] = apply_long_stop_window
152 };
153 
155  float *audio)
156 {
157  int i;
158  const float *output = sce->ret_buf;
159 
160  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, audio);
161 
163  s->mdct1024.mdct_calc(&s->mdct1024, sce->coeffs, output);
164  else
165  for (i = 0; i < 1024; i += 128)
166  s->mdct128.mdct_calc(&s->mdct128, &sce->coeffs[i], output + i*2);
167  memcpy(audio, audio + 1024, sizeof(audio[0]) * 1024);
168  memcpy(sce->pcoeffs, sce->coeffs, sizeof(sce->pcoeffs));
169 }
170 
171 /**
172  * Encode ics_info element.
173  * @see Table 4.6 (syntax of ics_info)
174  */
176 {
177  int w;
178 
179  put_bits(&s->pb, 1, 0); // ics_reserved bit
180  put_bits(&s->pb, 2, info->window_sequence[0]);
181  put_bits(&s->pb, 1, info->use_kb_window[0]);
182  if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
183  put_bits(&s->pb, 6, info->max_sfb);
184  put_bits(&s->pb, 1, !!info->predictor_present);
185  } else {
186  put_bits(&s->pb, 4, info->max_sfb);
187  for (w = 1; w < 8; w++)
188  put_bits(&s->pb, 1, !info->group_len[w]);
189  }
190 }
191 
192 /**
193  * Encode MS data.
194  * @see 4.6.8.1 "Joint Coding - M/S Stereo"
195  */
197 {
198  int i, w;
199 
200  put_bits(pb, 2, cpe->ms_mode);
201  if (cpe->ms_mode == 1)
202  for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
203  for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
204  put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
205 }
206 
207 /**
208  * Produce integer coefficients from scalefactors provided by the model.
209  */
210 static void adjust_frame_information(ChannelElement *cpe, int chans)
211 {
212  int i, w, w2, g, ch;
213  int maxsfb, cmaxsfb;
214 
215  for (ch = 0; ch < chans; ch++) {
216  IndividualChannelStream *ics = &cpe->ch[ch].ics;
217  maxsfb = 0;
218  cpe->ch[ch].pulse.num_pulse = 0;
219  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
220  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
221  for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w*16+cmaxsfb-1]; cmaxsfb--)
222  ;
223  maxsfb = FFMAX(maxsfb, cmaxsfb);
224  }
225  }
226  ics->max_sfb = maxsfb;
227 
228  //adjust zero bands for window groups
229  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
230  for (g = 0; g < ics->max_sfb; g++) {
231  i = 1;
232  for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
233  if (!cpe->ch[ch].zeroes[w2*16 + g]) {
234  i = 0;
235  break;
236  }
237  }
238  cpe->ch[ch].zeroes[w*16 + g] = i;
239  }
240  }
241  }
242 
243  if (chans > 1 && cpe->common_window) {
244  IndividualChannelStream *ics0 = &cpe->ch[0].ics;
245  IndividualChannelStream *ics1 = &cpe->ch[1].ics;
246  int msc = 0;
247  ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
248  ics1->max_sfb = ics0->max_sfb;
249  for (w = 0; w < ics0->num_windows*16; w += 16)
250  for (i = 0; i < ics0->max_sfb; i++)
251  if (cpe->ms_mask[w+i])
252  msc++;
253  if (msc == 0 || ics0->max_sfb == 0)
254  cpe->ms_mode = 0;
255  else
256  cpe->ms_mode = msc < ics0->max_sfb * ics0->num_windows ? 1 : 2;
257  }
258 }
259 
261 {
262  int w, w2, g, i;
263  IndividualChannelStream *ics = &cpe->ch[0].ics;
264  if (!cpe->common_window)
265  return;
266  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
267  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
268  int start = (w+w2) * 128;
269  for (g = 0; g < ics->num_swb; g++) {
270  int p = -1 + 2 * (cpe->ch[1].band_type[w*16+g] - 14);
271  float scale = cpe->ch[0].is_ener[w*16+g];
272  if (!cpe->is_mask[w*16 + g]) {
273  start += ics->swb_sizes[g];
274  continue;
275  }
276  if (cpe->ms_mask[w*16 + g])
277  p *= -1;
278  for (i = 0; i < ics->swb_sizes[g]; i++) {
279  float sum = (cpe->ch[0].coeffs[start+i] + p*cpe->ch[1].coeffs[start+i])*scale;
280  cpe->ch[0].coeffs[start+i] = sum;
281  cpe->ch[1].coeffs[start+i] = 0.0f;
282  }
283  start += ics->swb_sizes[g];
284  }
285  }
286  }
287 }
288 
290 {
291  int w, w2, g, i;
292  IndividualChannelStream *ics = &cpe->ch[0].ics;
293  if (!cpe->common_window)
294  return;
295  for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
296  for (w2 = 0; w2 < ics->group_len[w]; w2++) {
297  int start = (w+w2) * 128;
298  for (g = 0; g < ics->num_swb; g++) {
299  /* ms_mask can be used for other purposes in PNS and I/S,
300  * so must not apply M/S if any band uses either, even if
301  * ms_mask is set.
302  */
303  if (!cpe->ms_mask[w*16 + g] || cpe->is_mask[w*16 + g]
304  || cpe->ch[0].band_type[w*16 + g] >= NOISE_BT
305  || cpe->ch[1].band_type[w*16 + g] >= NOISE_BT) {
306  start += ics->swb_sizes[g];
307  continue;
308  }
309  for (i = 0; i < ics->swb_sizes[g]; i++) {
310  float L = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) * 0.5f;
311  float R = L - cpe->ch[1].coeffs[start+i];
312  cpe->ch[0].coeffs[start+i] = L;
313  cpe->ch[1].coeffs[start+i] = R;
314  }
315  start += ics->swb_sizes[g];
316  }
317  }
318  }
319 }
320 
321 /**
322  * Encode scalefactor band coding type.
323  */
325 {
326  int w;
327 
330 
331  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
332  s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
333 }
334 
335 /**
336  * Encode scalefactors.
337  */
340 {
341  int diff, off_sf = sce->sf_idx[0], off_pns = sce->sf_idx[0] - NOISE_OFFSET;
342  int off_is = 0, noise_flag = 1;
343  int i, w;
344 
345  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
346  for (i = 0; i < sce->ics.max_sfb; i++) {
347  if (!sce->zeroes[w*16 + i]) {
348  if (sce->band_type[w*16 + i] == NOISE_BT) {
349  diff = sce->sf_idx[w*16 + i] - off_pns;
350  off_pns = sce->sf_idx[w*16 + i];
351  if (noise_flag-- > 0) {
352  put_bits(&s->pb, NOISE_PRE_BITS, diff + NOISE_PRE);
353  continue;
354  }
355  } else if (sce->band_type[w*16 + i] == INTENSITY_BT ||
356  sce->band_type[w*16 + i] == INTENSITY_BT2) {
357  diff = sce->sf_idx[w*16 + i] - off_is;
358  off_is = sce->sf_idx[w*16 + i];
359  } else {
360  diff = sce->sf_idx[w*16 + i] - off_sf;
361  off_sf = sce->sf_idx[w*16 + i];
362  }
363  diff += SCALE_DIFF_ZERO;
364  av_assert0(diff >= 0 && diff <= 120);
366  }
367  }
368  }
369 }
370 
371 /**
372  * Encode pulse data.
373  */
374 static void encode_pulses(AACEncContext *s, Pulse *pulse)
375 {
376  int i;
377 
378  put_bits(&s->pb, 1, !!pulse->num_pulse);
379  if (!pulse->num_pulse)
380  return;
381 
382  put_bits(&s->pb, 2, pulse->num_pulse - 1);
383  put_bits(&s->pb, 6, pulse->start);
384  for (i = 0; i < pulse->num_pulse; i++) {
385  put_bits(&s->pb, 5, pulse->pos[i]);
386  put_bits(&s->pb, 4, pulse->amp[i]);
387  }
388 }
389 
390 /**
391  * Encode spectral coefficients processed by psychoacoustic model.
392  */
394 {
395  int start, i, w, w2;
396 
397  for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
398  start = 0;
399  for (i = 0; i < sce->ics.max_sfb; i++) {
400  if (sce->zeroes[w*16 + i]) {
401  start += sce->ics.swb_sizes[i];
402  continue;
403  }
404  for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
405  s->coder->quantize_and_encode_band(s, &s->pb,
406  &sce->coeffs[start + w2*128],
407  NULL, sce->ics.swb_sizes[i],
408  sce->sf_idx[w*16 + i],
409  sce->band_type[w*16 + i],
410  s->lambda,
411  sce->ics.window_clipping[w]);
412  }
413  start += sce->ics.swb_sizes[i];
414  }
415  }
416 }
417 
418 /**
419  * Downscale spectral coefficients for near-clipping windows to avoid artifacts
420  */
422 {
423  int start, i, j, w;
424 
425  if (sce->ics.clip_avoidance_factor < 1.0f) {
426  for (w = 0; w < sce->ics.num_windows; w++) {
427  start = 0;
428  for (i = 0; i < sce->ics.max_sfb; i++) {
429  float *swb_coeffs = &sce->coeffs[start + w*128];
430  for (j = 0; j < sce->ics.swb_sizes[i]; j++)
431  swb_coeffs[j] *= sce->ics.clip_avoidance_factor;
432  start += sce->ics.swb_sizes[i];
433  }
434  }
435  }
436 }
437 
438 /**
439  * Encode one channel of audio data.
440  */
443  int common_window)
444 {
445  put_bits(&s->pb, 8, sce->sf_idx[0]);
446  if (!common_window) {
447  put_ics_info(s, &sce->ics);
448  if (s->coder->encode_main_pred)
449  s->coder->encode_main_pred(s, sce);
450  if (s->coder->encode_ltp_info)
451  s->coder->encode_ltp_info(s, sce, 0);
452  }
453  encode_band_info(s, sce);
454  encode_scale_factors(avctx, s, sce);
455  encode_pulses(s, &sce->pulse);
456  put_bits(&s->pb, 1, !!sce->tns.present);
457  if (s->coder->encode_tns_info)
458  s->coder->encode_tns_info(s, sce);
459  put_bits(&s->pb, 1, 0); //ssr
460  encode_spectral_coeffs(s, sce);
461  return 0;
462 }
463 
464 /**
465  * Write some auxiliary information about the created AAC file.
466  */
467 static void put_bitstream_info(AACEncContext *s, const char *name)
468 {
469  int i, namelen, padbits;
470 
471  namelen = strlen(name) + 2;
472  put_bits(&s->pb, 3, TYPE_FIL);
473  put_bits(&s->pb, 4, FFMIN(namelen, 15));
474  if (namelen >= 15)
475  put_bits(&s->pb, 8, namelen - 14);
476  put_bits(&s->pb, 4, 0); //extension type - filler
477  padbits = -put_bits_count(&s->pb) & 7;
479  for (i = 0; i < namelen - 2; i++)
480  put_bits(&s->pb, 8, name[i]);
481  put_bits(&s->pb, 12 - padbits, 0);
482 }
483 
484 /*
485  * Copy input samples.
486  * Channels are reordered from libavcodec's default order to AAC order.
487  */
489 {
490  int ch;
491  int end = 2048 + (frame ? frame->nb_samples : 0);
492  const uint8_t *channel_map = aac_chan_maps[s->channels - 1];
493 
494  /* copy and remap input samples */
495  for (ch = 0; ch < s->channels; ch++) {
496  /* copy last 1024 samples of previous frame to the start of the current frame */
497  memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0]));
498 
499  /* copy new samples and zero any remaining samples */
500  if (frame) {
501  memcpy(&s->planar_samples[ch][2048],
502  frame->extended_data[channel_map[ch]],
503  frame->nb_samples * sizeof(s->planar_samples[0][0]));
504  }
505  memset(&s->planar_samples[ch][end], 0,
506  (3072 - end) * sizeof(s->planar_samples[0][0]));
507  }
508 }
509 
510 static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
511  const AVFrame *frame, int *got_packet_ptr)
512 {
513  AACEncContext *s = avctx->priv_data;
514  float **samples = s->planar_samples, *samples2, *la, *overlap;
515  ChannelElement *cpe;
518  int i, its, ch, w, chans, tag, start_ch, ret, frame_bits;
519  int target_bits, rate_bits, too_many_bits, too_few_bits;
520  int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
521  int chan_el_counter[4];
523 
524  /* add current frame to queue */
525  if (frame) {
526  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
527  return ret;
528  } else {
529  if (!s->afq.remaining_samples || (!s->afq.frame_alloc && !s->afq.frame_count))
530  return 0;
531  }
532 
533  copy_input_samples(s, frame);
534  if (s->psypp)
536 
537  if (!avctx->frame_number)
538  return 0;
539 
540  start_ch = 0;
541  for (i = 0; i < s->chan_map[0]; i++) {
542  FFPsyWindowInfo* wi = windows + start_ch;
543  tag = s->chan_map[i+1];
544  chans = tag == TYPE_CPE ? 2 : 1;
545  cpe = &s->cpe[i];
546  for (ch = 0; ch < chans; ch++) {
547  int k;
548  float clip_avoidance_factor;
549  sce = &cpe->ch[ch];
550  ics = &sce->ics;
551  s->cur_channel = start_ch + ch;
552  overlap = &samples[s->cur_channel][0];
553  samples2 = overlap + 1024;
554  la = samples2 + (448+64);
555  if (!frame)
556  la = NULL;
557  if (tag == TYPE_LFE) {
558  wi[ch].window_type[0] = wi[ch].window_type[1] = ONLY_LONG_SEQUENCE;
559  wi[ch].window_shape = 0;
560  wi[ch].num_windows = 1;
561  wi[ch].grouping[0] = 1;
562  wi[ch].clipping[0] = 0;
563 
564  /* Only the lowest 12 coefficients are used in a LFE channel.
565  * The expression below results in only the bottom 8 coefficients
566  * being used for 11.025kHz to 16kHz sample rates.
567  */
568  ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
569  } else {
570  wi[ch] = s->psy.model->window(&s->psy, samples2, la, s->cur_channel,
571  ics->window_sequence[0]);
572  }
573  ics->window_sequence[1] = ics->window_sequence[0];
574  ics->window_sequence[0] = wi[ch].window_type[0];
575  ics->use_kb_window[1] = ics->use_kb_window[0];
576  ics->use_kb_window[0] = wi[ch].window_shape;
577  ics->num_windows = wi[ch].num_windows;
578  ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
579  ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
580  ics->max_sfb = FFMIN(ics->max_sfb, ics->num_swb);
581  ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
587 
588  for (w = 0; w < ics->num_windows; w++)
589  ics->group_len[w] = wi[ch].grouping[w];
590 
591  /* Calculate input sample maximums and evaluate clipping risk */
592  clip_avoidance_factor = 0.0f;
593  for (w = 0; w < ics->num_windows; w++) {
594  const float *wbuf = overlap + w * 128;
595  const int wlen = 2048 / ics->num_windows;
596  float max = 0;
597  int j;
598  /* mdct input is 2 * output */
599  for (j = 0; j < wlen; j++)
600  max = FFMAX(max, fabsf(wbuf[j]));
601  wi[ch].clipping[w] = max;
602  }
603  for (w = 0; w < ics->num_windows; w++) {
604  if (wi[ch].clipping[w] > CLIP_AVOIDANCE_FACTOR) {
605  ics->window_clipping[w] = 1;
606  clip_avoidance_factor = FFMAX(clip_avoidance_factor, wi[ch].clipping[w]);
607  } else {
608  ics->window_clipping[w] = 0;
609  }
610  }
611  if (clip_avoidance_factor > CLIP_AVOIDANCE_FACTOR) {
612  ics->clip_avoidance_factor = CLIP_AVOIDANCE_FACTOR / clip_avoidance_factor;
613  } else {
614  ics->clip_avoidance_factor = 1.0f;
615  }
616 
617  apply_window_and_mdct(s, sce, overlap);
618 
619  if (s->options.ltp && s->coder->update_ltp) {
620  s->coder->update_ltp(s, sce);
621  apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
622  s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
623  }
624 
625  for (k = 0; k < 1024; k++) {
626  if (!(fabs(cpe->ch[ch].coeffs[k]) < 1E16)) { // Ensure headroom for energy calculation
627  av_log(avctx, AV_LOG_ERROR, "Input contains (near) NaN/+-Inf\n");
628  return AVERROR(EINVAL);
629  }
630  }
631  avoid_clipping(s, sce);
632  }
633  start_ch += chans;
634  }
635  if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels, 0)) < 0)
636  return ret;
637  frame_bits = its = 0;
638  do {
639  init_put_bits(&s->pb, avpkt->data, avpkt->size);
640 
641  if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & AV_CODEC_FLAG_BITEXACT))
643  start_ch = 0;
644  target_bits = 0;
645  memset(chan_el_counter, 0, sizeof(chan_el_counter));
646  for (i = 0; i < s->chan_map[0]; i++) {
647  FFPsyWindowInfo* wi = windows + start_ch;
648  const float *coeffs[2];
649  tag = s->chan_map[i+1];
650  chans = tag == TYPE_CPE ? 2 : 1;
651  cpe = &s->cpe[i];
652  cpe->common_window = 0;
653  memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
654  memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
655  put_bits(&s->pb, 3, tag);
656  put_bits(&s->pb, 4, chan_el_counter[tag]++);
657  for (ch = 0; ch < chans; ch++) {
658  sce = &cpe->ch[ch];
659  coeffs[ch] = sce->coeffs;
660  sce->ics.predictor_present = 0;
661  sce->ics.ltp.present = 0;
662  memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
663  memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
664  memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
665  for (w = 0; w < 128; w++)
666  if (sce->band_type[w] > RESERVED_BT)
667  sce->band_type[w] = 0;
668  }
669  s->psy.bitres.alloc = -1;
671  s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
672  if (s->psy.bitres.alloc > 0) {
673  /* Lambda unused here on purpose, we need to take psy's unscaled allocation */
674  target_bits += s->psy.bitres.alloc
675  * (s->lambda / (avctx->global_quality ? avctx->global_quality : 120));
676  s->psy.bitres.alloc /= chans;
677  }
678  s->cur_type = tag;
679  for (ch = 0; ch < chans; ch++) {
680  s->cur_channel = start_ch + ch;
681  if (s->options.pns && s->coder->mark_pns)
682  s->coder->mark_pns(s, avctx, &cpe->ch[ch]);
683  s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
684  }
685  if (chans > 1
686  && wi[0].window_type[0] == wi[1].window_type[0]
687  && wi[0].window_shape == wi[1].window_shape) {
688 
689  cpe->common_window = 1;
690  for (w = 0; w < wi[0].num_windows; w++) {
691  if (wi[0].grouping[w] != wi[1].grouping[w]) {
692  cpe->common_window = 0;
693  break;
694  }
695  }
696  }
697  for (ch = 0; ch < chans; ch++) { /* TNS and PNS */
698  sce = &cpe->ch[ch];
699  s->cur_channel = start_ch + ch;
700  if (s->options.tns && s->coder->search_for_tns)
701  s->coder->search_for_tns(s, sce);
702  if (s->options.tns && s->coder->apply_tns_filt)
703  s->coder->apply_tns_filt(s, sce);
704  if (sce->tns.present)
705  tns_mode = 1;
706  if (s->options.pns && s->coder->search_for_pns)
707  s->coder->search_for_pns(s, avctx, sce);
708  }
709  s->cur_channel = start_ch;
710  if (s->options.intensity_stereo) { /* Intensity Stereo */
711  if (s->coder->search_for_is)
712  s->coder->search_for_is(s, avctx, cpe);
713  if (cpe->is_mode) is_mode = 1;
715  }
716  if (s->options.pred) { /* Prediction */
717  for (ch = 0; ch < chans; ch++) {
718  sce = &cpe->ch[ch];
719  s->cur_channel = start_ch + ch;
720  if (s->options.pred && s->coder->search_for_pred)
721  s->coder->search_for_pred(s, sce);
722  if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
723  }
724  if (s->coder->adjust_common_pred)
725  s->coder->adjust_common_pred(s, cpe);
726  for (ch = 0; ch < chans; ch++) {
727  sce = &cpe->ch[ch];
728  s->cur_channel = start_ch + ch;
729  if (s->options.pred && s->coder->apply_main_pred)
730  s->coder->apply_main_pred(s, sce);
731  }
732  s->cur_channel = start_ch;
733  }
734  if (s->options.mid_side) { /* Mid/Side stereo */
735  if (s->options.mid_side == -1 && s->coder->search_for_ms)
736  s->coder->search_for_ms(s, cpe);
737  else if (cpe->common_window)
738  memset(cpe->ms_mask, 1, sizeof(cpe->ms_mask));
740  }
741  adjust_frame_information(cpe, chans);
742  if (s->options.ltp) { /* LTP */
743  for (ch = 0; ch < chans; ch++) {
744  sce = &cpe->ch[ch];
745  s->cur_channel = start_ch + ch;
746  if (s->coder->search_for_ltp)
747  s->coder->search_for_ltp(s, sce, cpe->common_window);
748  if (sce->ics.ltp.present) pred_mode = 1;
749  }
750  s->cur_channel = start_ch;
751  if (s->coder->adjust_common_ltp)
752  s->coder->adjust_common_ltp(s, cpe);
753  }
754  if (chans == 2) {
755  put_bits(&s->pb, 1, cpe->common_window);
756  if (cpe->common_window) {
757  put_ics_info(s, &cpe->ch[0].ics);
758  if (s->coder->encode_main_pred)
759  s->coder->encode_main_pred(s, &cpe->ch[0]);
760  if (s->coder->encode_ltp_info)
761  s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
762  encode_ms_info(&s->pb, cpe);
763  if (cpe->ms_mode) ms_mode = 1;
764  }
765  }
766  for (ch = 0; ch < chans; ch++) {
767  s->cur_channel = start_ch + ch;
768  encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
769  }
770  start_ch += chans;
771  }
772 
773  if (avctx->flags & AV_CODEC_FLAG_QSCALE) {
774  /* When using a constant Q-scale, don't mess with lambda */
775  break;
776  }
777 
778  /* rate control stuff
779  * allow between the nominal bitrate, and what psy's bit reservoir says to target
780  * but drift towards the nominal bitrate always
781  */
782  frame_bits = put_bits_count(&s->pb);
783  rate_bits = avctx->bit_rate * 1024 / avctx->sample_rate;
784  rate_bits = FFMIN(rate_bits, 6144 * s->channels - 3);
785  too_many_bits = FFMAX(target_bits, rate_bits);
786  too_many_bits = FFMIN(too_many_bits, 6144 * s->channels - 3);
787  too_few_bits = FFMIN(FFMAX(rate_bits - rate_bits/4, target_bits), too_many_bits);
788 
789  /* When using ABR, be strict (but only for increasing) */
790  too_few_bits = too_few_bits - too_few_bits/8;
791  too_many_bits = too_many_bits + too_many_bits/2;
792 
793  if ( its == 0 /* for steady-state Q-scale tracking */
794  || (its < 5 && (frame_bits < too_few_bits || frame_bits > too_many_bits))
795  || frame_bits >= 6144 * s->channels - 3 )
796  {
797  float ratio = ((float)rate_bits) / frame_bits;
798 
799  if (frame_bits >= too_few_bits && frame_bits <= too_many_bits) {
800  /*
801  * This path is for steady-state Q-scale tracking
802  * When frame bits fall within the stable range, we still need to adjust
803  * lambda to maintain it like so in a stable fashion (large jumps in lambda
804  * create artifacts and should be avoided), but slowly
805  */
806  ratio = sqrtf(sqrtf(ratio));
807  ratio = av_clipf(ratio, 0.9f, 1.1f);
808  } else {
809  /* Not so fast though */
810  ratio = sqrtf(ratio);
811  }
812  s->lambda = av_clipf(s->lambda * ratio, FLT_EPSILON, 65536.f);
813 
814  /* Keep iterating if we must reduce and lambda is in the sky */
815  if (ratio > 0.9f && ratio < 1.1f) {
816  break;
817  } else {
818  if (is_mode || ms_mode || tns_mode || pred_mode) {
819  for (i = 0; i < s->chan_map[0]; i++) {
820  // Must restore coeffs
821  chans = tag == TYPE_CPE ? 2 : 1;
822  cpe = &s->cpe[i];
823  for (ch = 0; ch < chans; ch++)
824  memcpy(cpe->ch[ch].coeffs, cpe->ch[ch].pcoeffs, sizeof(cpe->ch[ch].coeffs));
825  }
826  }
827  its++;
828  }
829  } else {
830  break;
831  }
832  } while (1);
833 
834  if (s->options.ltp && s->coder->ltp_insert_new_frame)
836 
837  put_bits(&s->pb, 3, TYPE_END);
838  flush_put_bits(&s->pb);
839 
841 
842  s->lambda_sum += s->lambda;
843  s->lambda_count++;
844 
845  ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
846  &avpkt->duration);
847 
848  avpkt->size = put_bits_count(&s->pb) >> 3;
849  *got_packet_ptr = 1;
850  return 0;
851 }
852 
854 {
855  AACEncContext *s = avctx->priv_data;
856 
857  av_log(avctx, AV_LOG_INFO, "Qavg: %.3f\n", s->lambda_count ? s->lambda_sum / s->lambda_count : NAN);
858 
859  ff_mdct_end(&s->mdct1024);
860  ff_mdct_end(&s->mdct128);
861  ff_psy_end(&s->psy);
862  ff_lpc_end(&s->lpc);
863  if (s->psypp)
865  av_freep(&s->buffer.samples);
866  av_freep(&s->cpe);
867  av_freep(&s->fdsp);
868  ff_af_queue_close(&s->afq);
869  return 0;
870 }
871 
873 {
874  int ret = 0;
875 
877  if (!s->fdsp)
878  return AVERROR(ENOMEM);
879 
880  // window init
885 
886  if ((ret = ff_mdct_init(&s->mdct1024, 11, 0, 32768.0)) < 0)
887  return ret;
888  if ((ret = ff_mdct_init(&s->mdct128, 8, 0, 32768.0)) < 0)
889  return ret;
890 
891  return 0;
892 }
893 
895 {
896  int ch;
897  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->buffer.samples, s->channels, 3 * 1024 * sizeof(s->buffer.samples[0]), alloc_fail);
898  FF_ALLOCZ_ARRAY_OR_GOTO(avctx, s->cpe, s->chan_map[0], sizeof(ChannelElement), alloc_fail);
899  FF_ALLOCZ_OR_GOTO(avctx, avctx->extradata, 5 + AV_INPUT_BUFFER_PADDING_SIZE, alloc_fail);
900 
901  for(ch = 0; ch < s->channels; ch++)
902  s->planar_samples[ch] = s->buffer.samples + 3 * 1024 * ch;
903 
904  return 0;
905 alloc_fail:
906  return AVERROR(ENOMEM);
907 }
908 
910 {
912 }
913 
915 {
916  AACEncContext *s = avctx->priv_data;
917  int i, ret = 0;
918  const uint8_t *sizes[2];
919  uint8_t grouping[AAC_MAX_CHANNELS];
920  int lengths[2];
921 
922  /* Constants */
923  s->last_frame_pb_count = 0;
924  avctx->extradata_size = 5;
925  avctx->frame_size = 1024;
926  avctx->initial_padding = 1024;
927  s->lambda = avctx->global_quality > 0 ? avctx->global_quality : 120;
928 
929  /* Channel map and unspecified bitrate guessing */
930  s->channels = avctx->channels;
931  ERROR_IF(s->channels > AAC_MAX_CHANNELS || s->channels == 7,
932  "Unsupported number of channels: %d\n", s->channels);
934  if (!avctx->bit_rate) {
935  for (i = 1; i <= s->chan_map[0]; i++) {
936  avctx->bit_rate += s->chan_map[i] == TYPE_CPE ? 128000 : /* Pair */
937  s->chan_map[i] == TYPE_LFE ? 16000 : /* LFE */
938  69000 ; /* SCE */
939  }
940  }
941 
942  /* Samplerate */
943  for (i = 0; i < 16; i++)
945  break;
946  s->samplerate_index = i;
947  ERROR_IF(s->samplerate_index == 16 ||
950  "Unsupported sample rate %d\n", avctx->sample_rate);
951 
952  /* Bitrate limiting */
953  WARN_IF(1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * s->channels,
954  "Too many bits %f > %d per frame requested, clamping to max\n",
955  1024.0 * avctx->bit_rate / avctx->sample_rate,
956  6144 * s->channels);
957  avctx->bit_rate = (int64_t)FFMIN(6144 * s->channels / 1024.0 * avctx->sample_rate,
958  avctx->bit_rate);
959 
960  /* Profile and option setting */
962  avctx->profile;
963  for (i = 0; i < FF_ARRAY_ELEMS(aacenc_profiles); i++)
964  if (avctx->profile == aacenc_profiles[i])
965  break;
966  if (avctx->profile == FF_PROFILE_MPEG2_AAC_LOW) {
967  avctx->profile = FF_PROFILE_AAC_LOW;
968  ERROR_IF(s->options.pred,
969  "Main prediction unavailable in the \"mpeg2_aac_low\" profile\n");
970  ERROR_IF(s->options.ltp,
971  "LTP prediction unavailable in the \"mpeg2_aac_low\" profile\n");
972  WARN_IF(s->options.pns,
973  "PNS unavailable in the \"mpeg2_aac_low\" profile, turning off\n");
974  s->options.pns = 0;
975  } else if (avctx->profile == FF_PROFILE_AAC_LTP) {
976  s->options.ltp = 1;
977  ERROR_IF(s->options.pred,
978  "Main prediction unavailable in the \"aac_ltp\" profile\n");
979  } else if (avctx->profile == FF_PROFILE_AAC_MAIN) {
980  s->options.pred = 1;
981  ERROR_IF(s->options.ltp,
982  "LTP prediction unavailable in the \"aac_main\" profile\n");
983  } else if (s->options.ltp) {
984  avctx->profile = FF_PROFILE_AAC_LTP;
985  WARN_IF(1,
986  "Chainging profile to \"aac_ltp\"\n");
987  ERROR_IF(s->options.pred,
988  "Main prediction unavailable in the \"aac_ltp\" profile\n");
989  } else if (s->options.pred) {
990  avctx->profile = FF_PROFILE_AAC_MAIN;
991  WARN_IF(1,
992  "Chainging profile to \"aac_main\"\n");
993  ERROR_IF(s->options.ltp,
994  "LTP prediction unavailable in the \"aac_main\" profile\n");
995  }
996  s->profile = avctx->profile;
997 
998  /* Coder limitations */
999  s->coder = &ff_aac_coders[s->options.coder];
1000  if (s->options.coder == AAC_CODER_ANMR) {
1002  "The ANMR coder is considered experimental, add -strict -2 to enable!\n");
1003  s->options.intensity_stereo = 0;
1004  s->options.pns = 0;
1005  }
1007  "The LPT profile requires experimental compliance, add -strict -2 to enable!\n");
1008 
1009  /* M/S introduces horrible artifacts with multichannel files, this is temporary */
1010  if (s->channels > 3)
1011  s->options.mid_side = 0;
1012 
1013  if ((ret = dsp_init(avctx, s)) < 0)
1014  goto fail;
1015 
1016  if ((ret = alloc_buffers(avctx, s)) < 0)
1017  goto fail;
1018 
1020 
1021  sizes[0] = ff_aac_swb_size_1024[s->samplerate_index];
1022  sizes[1] = ff_aac_swb_size_128[s->samplerate_index];
1023  lengths[0] = ff_aac_num_swb_1024[s->samplerate_index];
1024  lengths[1] = ff_aac_num_swb_128[s->samplerate_index];
1025  for (i = 0; i < s->chan_map[0]; i++)
1026  grouping[i] = s->chan_map[i + 1] == TYPE_CPE;
1027  if ((ret = ff_psy_init(&s->psy, avctx, 2, sizes, lengths,
1028  s->chan_map[0], grouping)) < 0)
1029  goto fail;
1030  s->psypp = ff_psy_preprocess_init(avctx);
1032  s->random_state = 0x1f2e3d4c;
1033 
1034  s->abs_pow34 = abs_pow34_v;
1036 
1037  if (ARCH_X86)
1039 
1040  if (HAVE_MIPSDSP)
1042 
1044  return AVERROR_UNKNOWN;
1045 
1046  ff_af_queue_init(avctx, &s->afq);
1047 
1048  return 0;
1049 fail:
1050  aac_encode_end(avctx);
1051  return ret;
1052 }
1053 
1054 #define AACENC_FLAGS AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
1055 static const AVOption aacenc_options[] = {
1056  {"aac_coder", "Coding algorithm", offsetof(AACEncContext, options.coder), AV_OPT_TYPE_INT, {.i64 = AAC_CODER_TWOLOOP}, 0, AAC_CODER_NB-1, AACENC_FLAGS, "coder"},
1057  {"anmr", "ANMR method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_ANMR}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1058  {"twoloop", "Two loop searching method", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_TWOLOOP}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1059  {"fast", "Constant quantizer", 0, AV_OPT_TYPE_CONST, {.i64 = AAC_CODER_FAST}, INT_MIN, INT_MAX, AACENC_FLAGS, "coder"},
1060  {"aac_ms", "Force M/S stereo coding", offsetof(AACEncContext, options.mid_side), AV_OPT_TYPE_BOOL, {.i64 = -1}, -1, 1, AACENC_FLAGS},
1061  {"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1062  {"aac_pns", "Perceptual noise substitution", offsetof(AACEncContext, options.pns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1063  {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_BOOL, {.i64 = 1}, -1, 1, AACENC_FLAGS},
1064  {"aac_ltp", "Long term prediction", offsetof(AACEncContext, options.ltp), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1065  {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
1066  {NULL}
1067 };
1068 
1069 static const AVClass aacenc_class = {
1070  "AAC encoder",
1074 };
1075 
1077  { "b", "0" },
1078  { NULL }
1079 };
1080 
1082  .name = "aac",
1083  .long_name = NULL_IF_CONFIG_SMALL("AAC (Advanced Audio Coding)"),
1084  .type = AVMEDIA_TYPE_AUDIO,
1085  .id = AV_CODEC_ID_AAC,
1086  .priv_data_size = sizeof(AACEncContext),
1087  .init = aac_encode_init,
1088  .encode2 = aac_encode_frame,
1089  .close = aac_encode_end,
1090  .defaults = aac_encode_defaults,
1091  .supported_samplerates = mpeg4audio_sample_rates,
1092  .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
1094  .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
1096  .priv_class = &aacenc_class,
1097 };
const char * name
Definition: avisynth_c.h:775
#define FF_COMPLIANCE_EXPERIMENTAL
Allow nonstandardized experimental things.
Definition: avcodec.h:2986
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:80
#define NULL
Definition: coverity.c:32
const char * s
Definition: avisynth_c.h:768
const AACCoefficientsEncoder * coder
Definition: aacenc.h:113
Band types following are encoded differently from others.
Definition: aac.h:86
static const uint8_t aac_chan_configs[AAC_MAX_CHANNELS][6]
default channel configurations
Definition: aacenctab.h:47
uint8_t use_kb_window[2]
If set, use Kaiser-Bessel window, otherwise use a sine window.
Definition: aac.h:177
int coder
Definition: aacenc.h:44
This structure describes decoded (raw) audio or video data.
Definition: frame.h:201
#define FF_ALLOCZ_ARRAY_OR_GOTO(ctx, p, nelem, elsize, label)
Definition: internal.h:160
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:76
AVOption.
Definition: opt.h:246
enum RawDataBlockType cur_type
channel group type cur_channel belongs to
Definition: aacenc.h:120
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:127
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:1069
static void put_bits(Jpeg2000EncoderContext *s, int val, int n)
put n times val bit
Definition: j2kenc.c:206
av_cold void ff_kbd_window_init(float *window, float alpha, int n)
Generate a Kaiser-Bessel Derived Window.
Definition: kbdwin.c:26
int64_t bit_rate
the average bitrate
Definition: avcodec.h:1826
#define LIBAVUTIL_VERSION_INT
Definition: version.h:86
#define SCALE_DIFF_ZERO
codebook index corresponding to zero scalefactor indices difference
Definition: aac.h:152
uint8_t window_clipping[8]
set if a certain window is near clipping
Definition: aac.h:191
Definition: aac.h:63
const char * g
Definition: vf_curves.c:112
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
Definition: aac.h:57
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx)
Cleanup audio preprocessing module.
Definition: psymodel.c:152
#define WARN_IF(cond,...)
Definition: aacenc_utils.h:274
int size
Definition: avcodec.h:1680
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:175
void(* search_for_tns)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:74
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:3270
int lambda_count
count(lambda), for Qvg reporting
Definition: aacenc.h:119
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:116
#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:3266
AVCodec.
Definition: avcodec.h:3739
static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
Encode spectral coefficients processed by psychoacoustic model.
Definition: aacenc.c:393
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
void(* apply_tns_filt)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:68
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:56
INTFLOAT pcoeffs[1024]
coefficients for IMDCT, pristine
Definition: aac.h:261
const uint16_t * swb_offset
table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular wind...
Definition: aac.h:181
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: avcodec.h:1027
#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:97
#define FF_CODEC_CAP_INIT_THREADSAFE
The codec does not modify any global variables in the init function, allowing to call the init functi...
Definition: internal.h:40
AAC encoder context.
Definition: aacenc.h:95
uint8_t
#define av_cold
Definition: attributes.h:82
void(* search_for_pns)(struct AACEncContext *s, AVCodecContext *avctx, SingleChannelElement *sce)
Definition: aacenc.h:72
AVOptions.
int intensity_stereo
Definition: aacenc.h:50
#define WINDOW_FUNC(type)
Definition: aacenc.c:89
void(* update_ltp)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:69
LPCContext lpc
used by TNS
Definition: aacenc.h:105
void ff_aac_coder_init_mips(AACEncContext *c)
SingleChannelElement ch[2]
Definition: aac.h:284
int samplerate_index
MPEG-4 samplerate index.
Definition: aacenc.h:106
Definition: aac.h:59
av_cold void ff_af_queue_init(AVCodecContext *avctx, AudioFrameQueue *afq)
Initialize AudioFrameQueue.
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: avcodec.h:1697
const uint8_t * chan_map
channel configuration map
Definition: aacenc.h:108
TemporalNoiseShaping tns
Definition: aac.h:250
const uint8_t ff_aac_scalefactor_bits[121]
Definition: aactab.c:92
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1876
AudioFrameQueue afq
Definition: aacenc.h:122
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:61
#define FF_PROFILE_AAC_LTP
Definition: avcodec.h:3273
uint8_t * data
Definition: avcodec.h:1679
const uint8_t * ff_aac_swb_size_128[]
Definition: aacenctab.c:91
uint32_t tag
Definition: movenc.c:1414
Scalefactor data are intensity stereo positions (in phase).
Definition: aac.h:89
int profile
copied from avctx
Definition: aacenc.h:104
#define AVOnce
Definition: thread.h:157
const OptionDef options[]
Definition: ffserver.c:3948
static void adjust_frame_information(ChannelElement *cpe, int chans)
Produce integer coefficients from scalefactors provided by the model.
Definition: aacenc.c:210
#define av_log(a,...)
float * planar_samples[8]
saved preprocessed input
Definition: aacenc.h:102
static const AVOption aacenc_options[]
Definition: aacenc.c:1055
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:127
void(* encode_ltp_info)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:63
#define ARCH_X86
Definition: config.h:38
av_default_item_name
static const int sizes[][2]
Definition: img2dec.c:51
#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:3278
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:181
float is_ener[128]
Intensity stereo pos (used by encoder)
Definition: aac.h:259
int initial_padding
Audio only.
Definition: avcodec.h:3451
float ff_aac_kbd_long_1024[1024]
Definition: aactab.c:38
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1856
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:74
int amp[4]
Definition: aac.h:228
const char * name
Name of the codec implementation.
Definition: avcodec.h:3746
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:488
uint8_t max_sfb
number of scalefactor bands per group
Definition: aac.h:175
void(* adjust_common_ltp)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:66
struct AACEncContext::@42 buffer
#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:73
#define fail()
Definition: checkasm.h:109
static int put_bits_count(PutBitContext *s)
Definition: put_bits.h:85
#define AACENC_FLAGS
Definition: aacenc.c:1054
INTFLOAT ret_buf[2048]
PCM output buffer.
Definition: aac.h:264
void(* set_special_band_scalefactors)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:71
enum WindowSequence window_sequence[2]
Definition: aac.h:176
INTFLOAT ltp_state[3072]
time signal for LTP
Definition: aac.h:265
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:929
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:876
#define AV_CODEC_CAP_SMALL_LAST_FRAME
Codec can be fed a final frame with a smaller size.
Definition: avcodec.h:1032
int cur_channel
current channel for coder context
Definition: aacenc.h:114
int last_frame_pb_count
number of bits for the previous frame
Definition: aacenc.h:117
#define FFMIN(a, b)
Definition: common.h:96
static void apply_intensity_stereo(ChannelElement *cpe)
Definition: aacenc.c:260
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: aacenc.c:510
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:130
#define FF_PROFILE_AAC_LOW
Definition: avcodec.h:3271
static const AVCodecDefault aac_encode_defaults[]
Definition: aacenc.c:1076
#define FF_PROFILE_UNKNOWN
Definition: avcodec.h:3267
int pos[4]
Definition: aac.h:227
int channels
channel count
Definition: aacenc.h:107
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:100
PutBitContext pb
Definition: aacenc.h:98
static void(*const apply_window[4])(AVFloatDSPContext *fdsp, SingleChannelElement *sce, const float *audio)
Definition: aacenc.c:145
#define L(x)
Definition: vp56_arith.h:36
if(ret< 0)
Definition: vf_mcdeint.c:279
AVFloatDSPContext * fdsp
Definition: aacenc.h:101
int mid_side
Definition: aacenc.h:49
#define FF_ARRAY_ELEMS(a)
static av_cold int aac_encode_end(AVCodecContext *avctx)
Definition: aacenc.c:853
void(* search_for_is)(struct AACEncContext *s, AVCodecContext *avctx, ChannelElement *cpe)
Definition: aacenc.h:77
void ff_aac_dsp_init_x86(AACEncContext *s)
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:2543
void(* search_for_ltp)(struct AACEncContext *s, SingleChannelElement *sce, int common_window)
Definition: aacenc.h:75
#define AV_LOG_INFO
Standard information.
Definition: log.h:187
#define AV_ONCE_INIT
Definition: thread.h:158
#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))
static void put_audio_specific_config(AVCodecContext *avctx)
Make AAC audio config object.
Definition: aacenc.c:58
Temporal Noise Shaping.
Definition: aac.h:198
int sample_rate
samples per second
Definition: avcodec.h:2523
float ff_aac_kbd_short_128[128]
Definition: aactab.c:39
static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
Encode MS data.
Definition: aacenc.c:196
void(* ltp_insert_new_frame)(struct AACEncContext *s)
Definition: aacenc.h:70
void(* search_for_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:78
main external API structure.
Definition: avcodec.h:1761
Definition: vf_geq.c:47
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:67
IndividualChannelStream ics
Definition: aac.h:249
int extradata_size
Definition: avcodec.h:1877
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:118
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:467
const int ff_aac_swb_size_128_len
Definition: aacenctab.c:107
void(* encode_main_pred)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:64
void(* adjust_common_pred)(struct AACEncContext *s, ChannelElement *cpe)
Definition: aacenc.h:65
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:374
uint16_t quantize_band_cost_cache_generation
Definition: aacenc.h:126
static av_cold void aac_encode_init_tables(void)
Definition: aacenc.c:909
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:111
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(const uint8_t *) pi - 0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(const int16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(const int16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(const int32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(const int32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(const int64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0f/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float *) pi *(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double *) pi *(INT64_C(1)<< 63))) #define FMT_PAIR_FUNC(out, in) static conv_func_type *const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB *AV_SAMPLE_FMT_NB]={ FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64), };static void cpy1(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, len);} static void cpy2(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 2 *len);} static void cpy4(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 4 *len);} static void cpy8(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 8 *len);} AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, const int *ch_map, int flags) { AudioConvert *ctx;conv_func_type *f=fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt)+AV_SAMPLE_FMT_NB *av_get_packed_sample_fmt(in_fmt)];if(!f) return NULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) return NULL;if(channels==1){ in_fmt=av_get_planar_sample_fmt(in_fmt);out_fmt=av_get_planar_sample_fmt(out_fmt);} ctx->channels=channels;ctx->conv_f=f;ctx->ch_map=ch_map;if(in_fmt==AV_SAMPLE_FMT_U8||in_fmt==AV_SAMPLE_FMT_U8P) memset(ctx->silence, 0x80, sizeof(ctx->silence));if(out_fmt==in_fmt &&!ch_map) { switch(av_get_bytes_per_sample(in_fmt)){ case 1:ctx->simd_f=cpy1;break;case 2:ctx->simd_f=cpy2;break;case 4:ctx->simd_f=cpy4;break;case 8:ctx->simd_f=cpy8;break;} } if(HAVE_X86ASM &&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
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:894
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
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:112
#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:1842
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:1081
struct FFPsyContext::@104 bitres
uint8_t is_mode
Set if any bands have been encoded using intensity stereo (used by encoder)
Definition: aac.h:280
INTFLOAT coeffs[1024]
coefficients for IMDCT, maybe processed
Definition: aac.h:262
const int avpriv_mpeg4audio_sample_rates[16]
Definition: mpeg4audio.c: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:60
Scalefactor data are intensity stereo positions (out of phase).
Definition: aac.h:88
Y Spectral Band Replication.
Definition: mpeg4audio.h:75
float * samples
Definition: aacenc.h:135
uint8_t prediction_used[41]
Definition: aac.h:190
static av_cold int aac_encode_init(AVCodecContext *avctx)
Definition: aacenc.c:914
common internal api header.
static void flush_put_bits(PutBitContext *s)
Pad the end of the output stream with zeros.
Definition: put_bits.h:101
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
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:338
ChannelElement * cpe
channel elements
Definition: aacenc.h:110
Individual Channel Stream.
Definition: aac.h:174
float clip_avoidance_factor
set if any window is near clipping to the necessary atennuation factor to avoid it ...
Definition: aac.h:192
static void init_put_bits(PutBitContext *s, uint8_t *buffer, int buffer_size)
Initialize the PutBitContext s.
Definition: put_bits.h:48
#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
#define AV_INPUT_BUFFER_PADDING_SIZE
Required number of additionally allocated bytes at the end of the input bitstream for decoding...
Definition: avcodec.h:777
channel element - generic struct for SCE/CPE/CCE/LFE
Definition: aac.h:275
void * priv_data
Definition: avcodec.h:1803
int start
Definition: aac.h:226
FFTContext mdct1024
long (1024 samples) frame transform context
Definition: aacenc.h:99
int random_state
Definition: aacenc.h:115
static av_always_inline int diff(const uint32_t a, const uint32_t b)
#define NAN
Definition: math.h:28
static const int16_t coeffs[]
int channels
number of audio channels
Definition: avcodec.h:2524
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:160
static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
Encode scalefactor band coding type.
Definition: aacenc.c:324
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:289
#define HAVE_MIPSDSP
Definition: config.h:79
void ff_af_queue_close(AudioFrameQueue *afq)
Close AudioFrameQueue.
enum BandType band_type[128]
band types
Definition: aac.h:252
#define LIBAVCODEC_IDENT
Definition: version.h:42
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:701
int frame_number
Frame counter, set by libavcodec.
Definition: avcodec.h:2554
FILE * out
Definition: movenc.c:54
#define av_freep(p)
void(* encode_tns_info)(struct AACEncContext *s, SingleChannelElement *sce)
Definition: aacenc.h:62
void INT64 start
Definition: avisynth_c.h:690
static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s, SingleChannelElement *sce, int common_window)
Encode one channel of audio data.
Definition: aacenc.c:441
static const AVCodecDefault defaults[]
Definition: dcaenc.c:1275
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:154
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:121
void(* abs_pow34)(float *out, const float *in, const int size)
Definition: aacenc.h:129
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:248
AAC data declarations.
av_cold void ff_psy_end(FFPsyContext *ctx)
Cleanup model context at the end.
Definition: psymodel.c:83
This structure stores compressed data.
Definition: avcodec.h:1656
static void avoid_clipping(AACEncContext *s, SingleChannelElement *sce)
Downscale spectral coefficients for near-clipping windows to avoid artifacts.
Definition: aacenc.c:421
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:267
int strict_std_compliance
strictly follow the standard (MPEG-4, ...).
Definition: avcodec.h:2981
static av_cold int dsp_init(AVCodecContext *avctx, AACEncContext *s)
Definition: aacenc.c:872
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: avcodec.h:1672
for(j=16;j >0;--j)
int pred
Definition: aacenc.h:48
#define FF_ALLOCZ_OR_GOTO(ctx, p, size, label)
Definition: internal.h:142
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:58
bitstream writer API