FFmpeg
opusenc.c
Go to the documentation of this file.
1 /*
2  * Opus encoder
3  * Copyright (c) 2017 Rostislav Pehlivanov <atomnuker@gmail.com>
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 #include "opusenc.h"
23 #include "opus_pvq.h"
24 #include "opusenc_psy.h"
25 #include "opustab.h"
26 
27 #include "libavutil/float_dsp.h"
28 #include "libavutil/opt.h"
29 #include "internal.h"
30 #include "bytestream.h"
31 #include "audio_frame_queue.h"
32 
33 typedef struct OpusEncContext {
43 
46 
48 
49  int channels;
50 
53 
54  /* Actual energy the decoder will have */
56 
57  DECLARE_ALIGNED(32, float, scratch)[2048];
59 
61 {
62  uint8_t *bs = avctx->extradata;
63 
64  bytestream_put_buffer(&bs, "OpusHead", 8);
65  bytestream_put_byte (&bs, 0x1);
66  bytestream_put_byte (&bs, avctx->channels);
67  bytestream_put_le16 (&bs, avctx->initial_padding);
68  bytestream_put_le32 (&bs, avctx->sample_rate);
69  bytestream_put_le16 (&bs, 0x0);
70  bytestream_put_byte (&bs, 0x0); /* Default layout */
71 }
72 
73 static int opus_gen_toc(OpusEncContext *s, uint8_t *toc, int *size, int *fsize_needed)
74 {
75  int tmp = 0x0, extended_toc = 0;
76  static const int toc_cfg[][OPUS_MODE_NB][OPUS_BANDWITH_NB] = {
77  /* Silk Hybrid Celt Layer */
78  /* NB MB WB SWB FB NB MB WB SWB FB NB MB WB SWB FB Bandwidth */
79  { { 0, 0, 0, 0, 0 }, { 0, 0, 0, 0, 0 }, { 17, 0, 21, 25, 29 } }, /* 2.5 ms */
80  { { 0, 0, 0, 0, 0 }, { 0, 0, 0, 0, 0 }, { 18, 0, 22, 26, 30 } }, /* 5 ms */
81  { { 1, 5, 9, 0, 0 }, { 0, 0, 0, 13, 15 }, { 19, 0, 23, 27, 31 } }, /* 10 ms */
82  { { 2, 6, 10, 0, 0 }, { 0, 0, 0, 14, 16 }, { 20, 0, 24, 28, 32 } }, /* 20 ms */
83  { { 3, 7, 11, 0, 0 }, { 0, 0, 0, 0, 0 }, { 0, 0, 0, 0, 0 } }, /* 40 ms */
84  { { 4, 8, 12, 0, 0 }, { 0, 0, 0, 0, 0 }, { 0, 0, 0, 0, 0 } }, /* 60 ms */
85  };
86  int cfg = toc_cfg[s->packet.framesize][s->packet.mode][s->packet.bandwidth];
87  *fsize_needed = 0;
88  if (!cfg)
89  return 1;
90  if (s->packet.frames == 2) { /* 2 packets */
91  if (s->frame[0].framebits == s->frame[1].framebits) { /* same size */
92  tmp = 0x1;
93  } else { /* different size */
94  tmp = 0x2;
95  *fsize_needed = 1; /* put frame sizes in the packet */
96  }
97  } else if (s->packet.frames > 2) {
98  tmp = 0x3;
99  extended_toc = 1;
100  }
101  tmp |= (s->channels > 1) << 2; /* Stereo or mono */
102  tmp |= (cfg - 1) << 3; /* codec configuration */
103  *toc++ = tmp;
104  if (extended_toc) {
105  for (int i = 0; i < (s->packet.frames - 1); i++)
106  *fsize_needed |= (s->frame[i].framebits != s->frame[i + 1].framebits);
107  tmp = (*fsize_needed) << 7; /* vbr flag */
108  tmp |= (0) << 6; /* padding flag */
109  tmp |= s->packet.frames;
110  *toc++ = tmp;
111  }
112  *size = 1 + extended_toc;
113  return 0;
114 }
115 
117 {
118  AVFrame *cur = NULL;
119  const int subframesize = s->avctx->frame_size;
120  int subframes = OPUS_BLOCK_SIZE(s->packet.framesize) / subframesize;
121 
122  cur = ff_bufqueue_get(&s->bufqueue);
123 
124  for (int ch = 0; ch < f->channels; ch++) {
125  CeltBlock *b = &f->block[ch];
126  const void *input = cur->extended_data[ch];
127  size_t bps = av_get_bytes_per_sample(cur->format);
128  memcpy(b->overlap, input, bps*cur->nb_samples);
129  }
130 
131  av_frame_free(&cur);
132 
133  for (int sf = 0; sf < subframes; sf++) {
134  if (sf != (subframes - 1))
135  cur = ff_bufqueue_get(&s->bufqueue);
136  else
137  cur = ff_bufqueue_peek(&s->bufqueue, 0);
138 
139  for (int ch = 0; ch < f->channels; ch++) {
140  CeltBlock *b = &f->block[ch];
141  const void *input = cur->extended_data[ch];
142  const size_t bps = av_get_bytes_per_sample(cur->format);
143  const size_t left = (subframesize - cur->nb_samples)*bps;
144  const size_t len = FFMIN(subframesize, cur->nb_samples)*bps;
145  memcpy(&b->samples[sf*subframesize], input, len);
146  memset(&b->samples[cur->nb_samples], 0, left);
147  }
148 
149  /* Last frame isn't popped off and freed yet - we need it for overlap */
150  if (sf != (subframes - 1))
151  av_frame_free(&cur);
152  }
153 }
154 
155 /* Apply the pre emphasis filter */
157 {
158  const int subframesize = s->avctx->frame_size;
159  const int subframes = OPUS_BLOCK_SIZE(s->packet.framesize) / subframesize;
160 
161  /* Filter overlap */
162  for (int ch = 0; ch < f->channels; ch++) {
163  CeltBlock *b = &f->block[ch];
164  float m = b->emph_coeff;
165  for (int i = 0; i < CELT_OVERLAP; i++) {
166  float sample = b->overlap[i];
167  b->overlap[i] = sample - m;
168  m = sample * CELT_EMPH_COEFF;
169  }
170  b->emph_coeff = m;
171  }
172 
173  /* Filter the samples but do not update the last subframe's coeff - overlap ^^^ */
174  for (int sf = 0; sf < subframes; sf++) {
175  for (int ch = 0; ch < f->channels; ch++) {
176  CeltBlock *b = &f->block[ch];
177  float m = b->emph_coeff;
178  for (int i = 0; i < subframesize; i++) {
179  float sample = b->samples[sf*subframesize + i];
180  b->samples[sf*subframesize + i] = sample - m;
181  m = sample * CELT_EMPH_COEFF;
182  }
183  if (sf != (subframes - 1))
184  b->emph_coeff = m;
185  }
186  }
187 }
188 
189 /* Create the window and do the mdct */
191 {
192  float *win = s->scratch, *temp = s->scratch + 1920;
193 
194  if (f->transient) {
195  for (int ch = 0; ch < f->channels; ch++) {
196  CeltBlock *b = &f->block[ch];
197  float *src1 = b->overlap;
198  for (int t = 0; t < f->blocks; t++) {
199  float *src2 = &b->samples[CELT_OVERLAP*t];
200  s->dsp->vector_fmul(win, src1, ff_celt_window, 128);
201  s->dsp->vector_fmul_reverse(&win[CELT_OVERLAP], src2,
202  ff_celt_window - 8, 128);
203  src1 = src2;
204  s->mdct[0]->mdct(s->mdct[0], b->coeffs + t, win, f->blocks);
205  }
206  }
207  } else {
208  int blk_len = OPUS_BLOCK_SIZE(f->size), wlen = OPUS_BLOCK_SIZE(f->size + 1);
209  int rwin = blk_len - CELT_OVERLAP, lap_dst = (wlen - blk_len - CELT_OVERLAP) >> 1;
210  memset(win, 0, wlen*sizeof(float));
211  for (int ch = 0; ch < f->channels; ch++) {
212  CeltBlock *b = &f->block[ch];
213 
214  /* Overlap */
215  s->dsp->vector_fmul(temp, b->overlap, ff_celt_window, 128);
216  memcpy(win + lap_dst, temp, CELT_OVERLAP*sizeof(float));
217 
218  /* Samples, flat top window */
219  memcpy(&win[lap_dst + CELT_OVERLAP], b->samples, rwin*sizeof(float));
220 
221  /* Samples, windowed */
222  s->dsp->vector_fmul_reverse(temp, b->samples + rwin,
223  ff_celt_window - 8, 128);
224  memcpy(win + lap_dst + blk_len, temp, CELT_OVERLAP*sizeof(float));
225 
226  s->mdct[f->size]->mdct(s->mdct[f->size], b->coeffs, win, 1);
227  }
228  }
229 
230  for (int ch = 0; ch < f->channels; ch++) {
231  CeltBlock *block = &f->block[ch];
232  for (int i = 0; i < CELT_MAX_BANDS; i++) {
233  float ener = 0.0f;
234  int band_offset = ff_celt_freq_bands[i] << f->size;
235  int band_size = ff_celt_freq_range[i] << f->size;
236  float *coeffs = &block->coeffs[band_offset];
237 
238  for (int j = 0; j < band_size; j++)
239  ener += coeffs[j]*coeffs[j];
240 
241  block->lin_energy[i] = sqrtf(ener) + FLT_EPSILON;
242  ener = 1.0f/block->lin_energy[i];
243 
244  for (int j = 0; j < band_size; j++)
245  coeffs[j] *= ener;
246 
247  block->energy[i] = log2f(block->lin_energy[i]) - ff_celt_mean_energy[i];
248 
249  /* CELT_ENERGY_SILENCE is what the decoder uses and its not -infinity */
250  block->energy[i] = FFMAX(block->energy[i], CELT_ENERGY_SILENCE);
251  }
252  }
253 }
254 
256 {
257  int tf_select = 0, diff = 0, tf_changed = 0, tf_select_needed;
258  int bits = f->transient ? 2 : 4;
259 
260  tf_select_needed = ((f->size && (opus_rc_tell(rc) + bits + 1) <= f->framebits));
261 
262  for (int i = f->start_band; i < f->end_band; i++) {
263  if ((opus_rc_tell(rc) + bits + tf_select_needed) <= f->framebits) {
264  const int tbit = (diff ^ 1) == f->tf_change[i];
265  ff_opus_rc_enc_log(rc, tbit, bits);
266  diff ^= tbit;
267  tf_changed |= diff;
268  }
269  bits = f->transient ? 4 : 5;
270  }
271 
272  if (tf_select_needed && ff_celt_tf_select[f->size][f->transient][0][tf_changed] !=
273  ff_celt_tf_select[f->size][f->transient][1][tf_changed]) {
274  ff_opus_rc_enc_log(rc, f->tf_select, 1);
275  tf_select = f->tf_select;
276  }
277 
278  for (int i = f->start_band; i < f->end_band; i++)
279  f->tf_change[i] = ff_celt_tf_select[f->size][f->transient][tf_select][f->tf_change[i]];
280 }
281 
283 {
284  float gain = f->pf_gain;
285  int txval, octave = f->pf_octave, period = f->pf_period, tapset = f->pf_tapset;
286 
287  ff_opus_rc_enc_log(rc, f->pfilter, 1);
288  if (!f->pfilter)
289  return;
290 
291  /* Octave */
292  txval = FFMIN(octave, 6);
293  ff_opus_rc_enc_uint(rc, txval, 6);
294  octave = txval;
295  /* Period */
296  txval = av_clip(period - (16 << octave) + 1, 0, (1 << (4 + octave)) - 1);
297  ff_opus_rc_put_raw(rc, period, 4 + octave);
298  period = txval + (16 << octave) - 1;
299  /* Gain */
300  txval = FFMIN(((int)(gain / 0.09375f)) - 1, 7);
301  ff_opus_rc_put_raw(rc, txval, 3);
302  gain = 0.09375f * (txval + 1);
303  /* Tapset */
304  if ((opus_rc_tell(rc) + 2) <= f->framebits)
306  else
307  tapset = 0;
308  /* Finally create the coeffs */
309  for (int i = 0; i < 2; i++) {
310  CeltBlock *block = &f->block[i];
311 
312  block->pf_period_new = FFMAX(period, CELT_POSTFILTER_MINPERIOD);
313  block->pf_gains_new[0] = gain * ff_celt_postfilter_taps[tapset][0];
314  block->pf_gains_new[1] = gain * ff_celt_postfilter_taps[tapset][1];
315  block->pf_gains_new[2] = gain * ff_celt_postfilter_taps[tapset][2];
316  }
317 }
318 
320  float last_energy[][CELT_MAX_BANDS], int intra)
321 {
322  float alpha, beta, prev[2] = { 0, 0 };
323  const uint8_t *pmod = ff_celt_coarse_energy_dist[f->size][intra];
324 
325  /* Inter is really just differential coding */
326  if (opus_rc_tell(rc) + 3 <= f->framebits)
327  ff_opus_rc_enc_log(rc, intra, 3);
328  else
329  intra = 0;
330 
331  if (intra) {
332  alpha = 0.0f;
333  beta = 1.0f - (4915.0f/32768.0f);
334  } else {
335  alpha = ff_celt_alpha_coef[f->size];
336  beta = ff_celt_beta_coef[f->size];
337  }
338 
339  for (int i = f->start_band; i < f->end_band; i++) {
340  for (int ch = 0; ch < f->channels; ch++) {
341  CeltBlock *block = &f->block[ch];
342  const int left = f->framebits - opus_rc_tell(rc);
343  const float last = FFMAX(-9.0f, last_energy[ch][i]);
344  float diff = block->energy[i] - prev[ch] - last*alpha;
345  int q_en = lrintf(diff);
346  if (left >= 15) {
347  ff_opus_rc_enc_laplace(rc, &q_en, pmod[i << 1] << 7, pmod[(i << 1) + 1] << 6);
348  } else if (left >= 2) {
349  q_en = av_clip(q_en, -1, 1);
350  ff_opus_rc_enc_cdf(rc, 2*q_en + 3*(q_en < 0), ff_celt_model_energy_small);
351  } else if (left >= 1) {
352  q_en = av_clip(q_en, -1, 0);
353  ff_opus_rc_enc_log(rc, (q_en & 1), 1);
354  } else q_en = -1;
355 
356  block->error_energy[i] = q_en - diff;
357  prev[ch] += beta * q_en;
358  }
359  }
360 }
361 
363  float last_energy[][CELT_MAX_BANDS])
364 {
365  uint32_t inter, intra;
367 
368  exp_quant_coarse(rc, f, last_energy, 1);
369  intra = OPUS_RC_CHECKPOINT_BITS(rc);
370 
372 
373  exp_quant_coarse(rc, f, last_energy, 0);
374  inter = OPUS_RC_CHECKPOINT_BITS(rc);
375 
376  if (inter > intra) { /* Unlikely */
378  exp_quant_coarse(rc, f, last_energy, 1);
379  }
380 }
381 
383 {
384  for (int i = f->start_band; i < f->end_band; i++) {
385  if (!f->fine_bits[i])
386  continue;
387  for (int ch = 0; ch < f->channels; ch++) {
388  CeltBlock *block = &f->block[ch];
389  int quant, lim = (1 << f->fine_bits[i]);
390  float offset, diff = 0.5f - block->error_energy[i];
391  quant = av_clip(floor(diff*lim), 0, lim - 1);
392  ff_opus_rc_put_raw(rc, quant, f->fine_bits[i]);
393  offset = 0.5f - ((quant + 0.5f) * (1 << (14 - f->fine_bits[i])) / 16384.0f);
394  block->error_energy[i] -= offset;
395  }
396  }
397 }
398 
400 {
401  for (int priority = 0; priority < 2; priority++) {
402  for (int i = f->start_band; i < f->end_band && (f->framebits - opus_rc_tell(rc)) >= f->channels; i++) {
403  if (f->fine_priority[i] != priority || f->fine_bits[i] >= CELT_MAX_FINE_BITS)
404  continue;
405  for (int ch = 0; ch < f->channels; ch++) {
406  CeltBlock *block = &f->block[ch];
407  const float err = block->error_energy[i];
408  const float offset = 0.5f * (1 << (14 - f->fine_bits[i] - 1)) / 16384.0f;
409  const int sign = FFABS(err + offset) < FFABS(err - offset);
410  ff_opus_rc_put_raw(rc, sign, 1);
411  block->error_energy[i] -= offset*(1 - 2*sign);
412  }
413  }
414  }
415 }
416 
418  CeltFrame *f, int index)
419 {
421 
422  ff_opus_psy_celt_frame_init(&s->psyctx, f, index);
423 
425 
426  if (f->silence) {
427  if (f->framebits >= 16)
428  ff_opus_rc_enc_log(rc, 1, 15); /* Silence (if using explicit singalling) */
429  for (int ch = 0; ch < s->channels; ch++)
430  memset(s->last_quantized_energy[ch], 0.0f, sizeof(float)*CELT_MAX_BANDS);
431  return;
432  }
433 
434  /* Filters */
436  if (f->pfilter) {
437  ff_opus_rc_enc_log(rc, 0, 15);
439  }
440 
441  /* Transform */
442  celt_frame_mdct(s, f);
443 
444  /* Need to handle transient/non-transient switches at any point during analysis */
445  while (ff_opus_psy_celt_frame_process(&s->psyctx, f, index))
446  celt_frame_mdct(s, f);
447 
449 
450  /* Silence */
451  ff_opus_rc_enc_log(rc, 0, 15);
452 
453  /* Pitch filter */
454  if (!f->start_band && opus_rc_tell(rc) + 16 <= f->framebits)
456 
457  /* Transient flag */
458  if (f->size && opus_rc_tell(rc) + 3 <= f->framebits)
459  ff_opus_rc_enc_log(rc, f->transient, 3);
460 
461  /* Main encoding */
462  celt_quant_coarse (f, rc, s->last_quantized_energy);
463  celt_enc_tf (f, rc);
464  ff_celt_bitalloc (f, rc, 1);
465  celt_quant_fine (f, rc);
466  ff_celt_quant_bands(f, rc);
467 
468  /* Anticollapse bit */
469  if (f->anticollapse_needed)
470  ff_opus_rc_put_raw(rc, f->anticollapse, 1);
471 
472  /* Final per-band energy adjustments from leftover bits */
473  celt_quant_final(s, rc, f);
474 
475  for (int ch = 0; ch < f->channels; ch++) {
476  CeltBlock *block = &f->block[ch];
477  for (int i = 0; i < CELT_MAX_BANDS; i++)
478  s->last_quantized_energy[ch][i] = block->energy[i] + block->error_energy[i];
479  }
480 }
481 
482 static inline int write_opuslacing(uint8_t *dst, int v)
483 {
484  dst[0] = FFMIN(v - FFALIGN(v - 255, 4), v);
485  dst[1] = v - dst[0] >> 2;
486  return 1 + (v >= 252);
487 }
488 
490 {
491  int offset, fsize_needed;
492 
493  /* Write toc */
494  opus_gen_toc(s, avpkt->data, &offset, &fsize_needed);
495 
496  /* Frame sizes if needed */
497  if (fsize_needed) {
498  for (int i = 0; i < s->packet.frames - 1; i++) {
499  offset += write_opuslacing(avpkt->data + offset,
500  s->frame[i].framebits >> 3);
501  }
502  }
503 
504  /* Packets */
505  for (int i = 0; i < s->packet.frames; i++) {
506  ff_opus_rc_enc_end(&s->rc[i], avpkt->data + offset,
507  s->frame[i].framebits >> 3);
508  offset += s->frame[i].framebits >> 3;
509  }
510 
511  avpkt->size = offset;
512 }
513 
514 /* Used as overlap for the first frame and padding for the last encoded packet */
516 {
517  AVFrame *f = av_frame_alloc();
518  if (!f)
519  return NULL;
520  f->format = s->avctx->sample_fmt;
521  f->nb_samples = s->avctx->frame_size;
522  f->channel_layout = s->avctx->channel_layout;
523  if (av_frame_get_buffer(f, 4)) {
524  av_frame_free(&f);
525  return NULL;
526  }
527  for (int i = 0; i < s->channels; i++) {
528  size_t bps = av_get_bytes_per_sample(f->format);
529  memset(f->extended_data[i], 0, bps*f->nb_samples);
530  }
531  return f;
532 }
533 
534 static int opus_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
535  const AVFrame *frame, int *got_packet_ptr)
536 {
537  OpusEncContext *s = avctx->priv_data;
538  int ret, frame_size, alloc_size = 0;
539 
540  if (frame) { /* Add new frame to queue */
541  if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
542  return ret;
543  ff_bufqueue_add(avctx, &s->bufqueue, av_frame_clone(frame));
544  } else {
545  ff_opus_psy_signal_eof(&s->psyctx);
546  if (!s->afq.remaining_samples || !avctx->frame_number)
547  return 0; /* We've been flushed and there's nothing left to encode */
548  }
549 
550  /* Run the psychoacoustic system */
551  if (ff_opus_psy_process(&s->psyctx, &s->packet))
552  return 0;
553 
554  frame_size = OPUS_BLOCK_SIZE(s->packet.framesize);
555 
556  if (!frame) {
557  /* This can go negative, that's not a problem, we only pad if positive */
558  int pad_empty = s->packet.frames*(frame_size/s->avctx->frame_size) - s->bufqueue.available + 1;
559  /* Pad with empty 2.5 ms frames to whatever framesize was decided,
560  * this should only happen at the very last flush frame. The frames
561  * allocated here will be freed (because they have no other references)
562  * after they get used by celt_frame_setup_input() */
563  for (int i = 0; i < pad_empty; i++) {
564  AVFrame *empty = spawn_empty_frame(s);
565  if (!empty)
566  return AVERROR(ENOMEM);
567  ff_bufqueue_add(avctx, &s->bufqueue, empty);
568  }
569  }
570 
571  for (int i = 0; i < s->packet.frames; i++) {
572  celt_encode_frame(s, &s->rc[i], &s->frame[i], i);
573  alloc_size += s->frame[i].framebits >> 3;
574  }
575 
576  /* Worst case toc + the frame lengths if needed */
577  alloc_size += 2 + s->packet.frames*2;
578 
579  if ((ret = ff_alloc_packet2(avctx, avpkt, alloc_size, 0)) < 0)
580  return ret;
581 
582  /* Assemble packet */
583  opus_packet_assembler(s, avpkt);
584 
585  /* Update the psychoacoustic system */
586  ff_opus_psy_postencode_update(&s->psyctx, s->frame, s->rc);
587 
588  /* Remove samples from queue and skip if needed */
589  ff_af_queue_remove(&s->afq, s->packet.frames*frame_size, &avpkt->pts, &avpkt->duration);
590  if (s->packet.frames*frame_size > avpkt->duration) {
592  if (!side)
593  return AVERROR(ENOMEM);
594  AV_WL32(&side[4], s->packet.frames*frame_size - avpkt->duration + 120);
595  }
596 
597  *got_packet_ptr = 1;
598 
599  return 0;
600 }
601 
603 {
604  OpusEncContext *s = avctx->priv_data;
605 
606  for (int i = 0; i < CELT_BLOCK_NB; i++)
607  ff_mdct15_uninit(&s->mdct[i]);
608 
609  ff_celt_pvq_uninit(&s->pvq);
610  av_freep(&s->dsp);
611  av_freep(&s->frame);
612  av_freep(&s->rc);
613  ff_af_queue_close(&s->afq);
614  ff_opus_psy_end(&s->psyctx);
615  ff_bufqueue_discard_all(&s->bufqueue);
616  av_freep(&avctx->extradata);
617 
618  return 0;
619 }
620 
622 {
623  int ret, max_frames;
624  OpusEncContext *s = avctx->priv_data;
625 
626  s->avctx = avctx;
627  s->channels = avctx->channels;
628 
629  /* Opus allows us to change the framesize on each packet (and each packet may
630  * have multiple frames in it) but we can't change the codec's frame size on
631  * runtime, so fix it to the lowest possible number of samples and use a queue
632  * to accumulate AVFrames until we have enough to encode whatever the encoder
633  * decides is the best */
634  avctx->frame_size = 120;
635  /* Initial padding will change if SILK is ever supported */
636  avctx->initial_padding = 120;
637 
638  if (!avctx->bit_rate) {
639  int coupled = ff_opus_default_coupled_streams[s->channels - 1];
640  avctx->bit_rate = coupled*(96000) + (s->channels - coupled*2)*(48000);
641  } else if (avctx->bit_rate < 6000 || avctx->bit_rate > 255000 * s->channels) {
642  int64_t clipped_rate = av_clip(avctx->bit_rate, 6000, 255000 * s->channels);
643  av_log(avctx, AV_LOG_ERROR, "Unsupported bitrate %"PRId64" kbps, clipping to %"PRId64" kbps\n",
644  avctx->bit_rate/1000, clipped_rate/1000);
645  avctx->bit_rate = clipped_rate;
646  }
647 
648  /* Extradata */
649  avctx->extradata_size = 19;
651  if (!avctx->extradata)
652  return AVERROR(ENOMEM);
653  opus_write_extradata(avctx);
654 
655  ff_af_queue_init(avctx, &s->afq);
656 
657  if ((ret = ff_celt_pvq_init(&s->pvq, 1)) < 0)
658  return ret;
659 
660  if (!(s->dsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT)))
661  return AVERROR(ENOMEM);
662 
663  /* I have no idea why a base scaling factor of 68 works, could be the twiddles */
664  for (int i = 0; i < CELT_BLOCK_NB; i++)
665  if ((ret = ff_mdct15_init(&s->mdct[i], 0, i + 3, 68 << (CELT_BLOCK_NB - 1 - i))))
666  return AVERROR(ENOMEM);
667 
668  /* Zero out previous energy (matters for inter first frame) */
669  for (int ch = 0; ch < s->channels; ch++)
670  memset(s->last_quantized_energy[ch], 0.0f, sizeof(float)*CELT_MAX_BANDS);
671 
672  /* Allocate an empty frame to use as overlap for the first frame of audio */
673  ff_bufqueue_add(avctx, &s->bufqueue, spawn_empty_frame(s));
674  if (!ff_bufqueue_peek(&s->bufqueue, 0))
675  return AVERROR(ENOMEM);
676 
677  if ((ret = ff_opus_psy_init(&s->psyctx, s->avctx, &s->bufqueue, &s->options)))
678  return ret;
679 
680  /* Frame structs and range coder buffers */
681  max_frames = ceilf(FFMIN(s->options.max_delay_ms, 120.0f)/2.5f);
682  s->frame = av_malloc(max_frames*sizeof(CeltFrame));
683  if (!s->frame)
684  return AVERROR(ENOMEM);
685  s->rc = av_malloc(max_frames*sizeof(OpusRangeCoder));
686  if (!s->rc)
687  return AVERROR(ENOMEM);
688 
689  for (int i = 0; i < max_frames; i++) {
690  s->frame[i].dsp = s->dsp;
691  s->frame[i].avctx = s->avctx;
692  s->frame[i].seed = 0;
693  s->frame[i].pvq = s->pvq;
694  s->frame[i].apply_phase_inv = 1;
695  s->frame[i].block[0].emph_coeff = s->frame[i].block[1].emph_coeff = 0.0f;
696  }
697 
698  return 0;
699 }
700 
701 #define OPUSENC_FLAGS AV_OPT_FLAG_ENCODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM
702 static const AVOption opusenc_options[] = {
703  { "opus_delay", "Maximum delay in milliseconds", offsetof(OpusEncContext, options.max_delay_ms), AV_OPT_TYPE_FLOAT, { .dbl = OPUS_MAX_LOOKAHEAD }, 2.5f, OPUS_MAX_LOOKAHEAD, OPUSENC_FLAGS, "max_delay_ms" },
704  { NULL },
705 };
706 
707 static const AVClass opusenc_class = {
708  .class_name = "Opus encoder",
709  .item_name = av_default_item_name,
710  .option = opusenc_options,
711  .version = LIBAVUTIL_VERSION_INT,
712 };
713 
715  { "b", "0" },
716  { "compression_level", "10" },
717  { NULL },
718 };
719 
721  .name = "opus",
722  .long_name = NULL_IF_CONFIG_SMALL("Opus"),
723  .type = AVMEDIA_TYPE_AUDIO,
724  .id = AV_CODEC_ID_OPUS,
725  .defaults = opusenc_defaults,
726  .priv_class = &opusenc_class,
727  .priv_data_size = sizeof(OpusEncContext),
729  .encode2 = opus_encode_frame,
730  .close = opus_encode_end,
733  .supported_samplerates = (const int []){ 48000, 0 },
734  .channel_layouts = (const uint64_t []){ AV_CH_LAYOUT_MONO,
735  AV_CH_LAYOUT_STEREO, 0 },
736  .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
738 };
AVCodecContext::frame_size
int frame_size
Number of samples per channel in an audio frame.
Definition: avcodec.h:2245
AVCodec
AVCodec.
Definition: avcodec.h:3481
AV_SAMPLE_FMT_FLTP
@ AV_SAMPLE_FMT_FLTP
float, planar
Definition: samplefmt.h:69
ff_celt_postfilter_taps
const float ff_celt_postfilter_taps[3][3]
Definition: opustab.c:1093
ff_opus_rc_enc_cdf
void ff_opus_rc_enc_cdf(OpusRangeCoder *rc, int val, const uint16_t *cdf)
Definition: opus_rc.c:109
FF_CODEC_CAP_INIT_THREADSAFE
#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
OpusEncContext::av_class
AVClass * av_class
Definition: opusenc.c:34
ff_opus_psy_process
int ff_opus_psy_process(OpusPsyContext *s, OpusPacketInfo *p)
Definition: opusenc_psy.c:223
spawn_empty_frame
static AVFrame * spawn_empty_frame(OpusEncContext *s)
Definition: opusenc.c:515
init
static av_cold int init(AVCodecContext *avctx)
Definition: avrndec.c:35
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
opt.h
ff_celt_freq_bands
const uint8_t ff_celt_freq_bands[]
Definition: opustab.c:763
AV_WL32
#define AV_WL32(p, v)
Definition: intreadwrite.h:426
ff_af_queue_remove
void ff_af_queue_remove(AudioFrameQueue *afq, int nb_samples, int64_t *pts, int64_t *duration)
Remove frame(s) from the queue.
Definition: audio_frame_queue.c:75
av_frame_get_buffer
int av_frame_get_buffer(AVFrame *frame, int align)
Allocate new buffer(s) for audio or video data.
Definition: frame.c:324
AVCodecContext::sample_rate
int sample_rate
samples per second
Definition: avcodec.h:2225
ff_af_queue_close
void ff_af_queue_close(AudioFrameQueue *afq)
Close AudioFrameQueue.
Definition: audio_frame_queue.c:36
log2f
#define log2f(x)
Definition: libm.h:409
ff_opus_psy_celt_frame_init
void ff_opus_psy_celt_frame_init(OpusPsyContext *s, CeltFrame *f, int index)
Definition: opusenc_psy.c:254
ff_celt_window
const float * ff_celt_window
Definition: opustab.c:1130
AV_CH_LAYOUT_MONO
#define AV_CH_LAYOUT_MONO
Definition: channel_layout.h:85
ff_opus_rc_enc_uint
void ff_opus_rc_enc_uint(OpusRangeCoder *rc, uint32_t val, uint32_t size)
CELT: write a uniformly distributed integer.
Definition: opus_rc.c:204
ch
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 &&1) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels);if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels);if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels);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
ff_af_queue_init
av_cold void ff_af_queue_init(AVCodecContext *avctx, AudioFrameQueue *afq)
Initialize AudioFrameQueue.
Definition: audio_frame_queue.c:28
av_frame_free
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:202
ff_celt_model_tapset
const uint16_t ff_celt_model_tapset[]
Definition: opustab.c:753
opusenc_options
static const AVOption opusenc_options[]
Definition: opusenc.c:702
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:295
tmp
static uint8_t tmp[11]
Definition: aes_ctr.c:26
internal.h
AVPacket::data
uint8_t * data
Definition: avcodec.h:1477
opusenc_defaults
static const AVCodecDefault opusenc_defaults[]
Definition: opusenc.c:714
AVOption
AVOption.
Definition: opt.h:246
b
#define b
Definition: input.c:41
OPUS_RC_CHECKPOINT_SPAWN
#define OPUS_RC_CHECKPOINT_SPAWN(rc)
Definition: opus_rc.h:116
opus_packet_assembler
static void opus_packet_assembler(OpusEncContext *s, AVPacket *avpkt)
Definition: opusenc.c:489
ff_mdct15_init
av_cold int ff_mdct15_init(MDCT15Context **ps, int inverse, int N, double scale)
Definition: mdct15.c:247
AVPacket::duration
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: avcodec.h:1495
OpusEncContext::options
OpusEncOptions options
Definition: opusenc.c:35
opus_encode_end
static av_cold int opus_encode_end(AVCodecContext *avctx)
Definition: opusenc.c:602
ff_celt_pvq_init
int av_cold ff_celt_pvq_init(CeltPVQ **pvq, int encode)
Definition: opus_pvq.c:897
exp_quant_coarse
static void exp_quant_coarse(OpusRangeCoder *rc, CeltFrame *f, float last_energy[][CELT_MAX_BANDS], int intra)
Definition: opusenc.c:319
ff_bufqueue_get
static AVFrame * ff_bufqueue_get(struct FFBufQueue *queue)
Get the first buffer from the queue and remove it.
Definition: bufferqueue.h:98
opus_rc_tell
static av_always_inline uint32_t opus_rc_tell(const OpusRangeCoder *rc)
CELT: estimate bits of entropy that have thus far been consumed for the current CELT frame,...
Definition: opus_rc.h:61
ff_celt_coarse_energy_dist
const uint8_t ff_celt_coarse_energy_dist[4][2][42]
Definition: opustab.c:803
av_malloc
#define av_malloc(s)
Definition: tableprint_vlc.h:31
CeltBlock
Definition: opus_celt.h:65
win
static float win(SuperEqualizerContext *s, float n, int N)
Definition: af_superequalizer.c:119
CeltPVQ
Definition: opus_pvq.h:35
OPUSENC_FLAGS
#define OPUSENC_FLAGS
Definition: opusenc.c:701
audio_frame_queue.h
AVCodecContext::initial_padding
int initial_padding
Audio only.
Definition: avcodec.h:3096
opus_encode_frame
static int opus_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
Definition: opusenc.c:534
AVCodecContext::flags
int flags
AV_CODEC_FLAG_*.
Definition: avcodec.h:1645
OpusEncContext::frame
CeltFrame * frame
Definition: opusenc.c:51
ff_opus_psy_end
av_cold int ff_opus_psy_end(OpusPsyContext *s)
Definition: opusenc_psy.c:593
ff_opus_encoder
AVCodec ff_opus_encoder
Definition: opusenc.c:720
ff_af_queue_add
int ff_af_queue_add(AudioFrameQueue *afq, const AVFrame *f)
Add a frame to the queue.
Definition: audio_frame_queue.c:44
AV_CH_LAYOUT_STEREO
#define AV_CH_LAYOUT_STEREO
Definition: channel_layout.h:86
ff_opus_psy_postencode_update
void ff_opus_psy_postencode_update(OpusPsyContext *s, CeltFrame *f, OpusRangeCoder *rc)
Definition: opusenc_psy.c:479
av_frame_alloc
AVFrame * av_frame_alloc(void)
Allocate an AVFrame and set its fields to default values.
Definition: frame.c:189
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
OPUS_MAX_LOOKAHEAD
#define OPUS_MAX_LOOKAHEAD
Definition: opusenc.h:32
OpusEncContext::afq
AudioFrameQueue afq
Definition: opusenc.c:38
AV_CODEC_CAP_EXPERIMENTAL
#define AV_CODEC_CAP_EXPERIMENTAL
Codec is experimental and is thus avoided in favor of non experimental encoders.
Definition: avcodec.h:1029
av_cold
#define av_cold
Definition: attributes.h:84
OPUS_BLOCK_SIZE
#define OPUS_BLOCK_SIZE(x)
Definition: opusenc.h:39
AVCodecContext::extradata_size
int extradata_size
Definition: avcodec.h:1667
OpusEncContext::rc
OpusRangeCoder * rc
Definition: opusenc.c:52
OpusEncContext::scratch
float scratch[2048]
Definition: opusenc.c:57
s
#define s(width, name)
Definition: cbs_vp9.c:257
frame_size
int frame_size
Definition: mxfenc.c:2215
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
ff_opus_psy_celt_frame_process
int ff_opus_psy_celt_frame_process(OpusPsyContext *s, CeltFrame *f, int index)
Definition: opusenc_psy.c:455
CELT_MAX_FINE_BITS
#define CELT_MAX_FINE_BITS
Definition: opus_celt.h:40
bits
uint8_t bits
Definition: vp3data.h:202
AudioFrameQueue
Definition: audio_frame_queue.h:32
av_frame_clone
AVFrame * av_frame_clone(const AVFrame *src)
Create a new frame that references the same data as src.
Definition: frame.c:540
MDCT15Context
Definition: mdct15.h:28
f
#define f(width, name)
Definition: cbs_vp9.c:255
av_packet_new_side_data
uint8_t * av_packet_new_side_data(AVPacket *pkt, enum AVPacketSideDataType type, int size)
Allocate new information of a packet.
Definition: avpacket.c:329
OpusPsyContext
Definition: opusenc_psy.h:54
OpusEncContext::mdct
MDCT15Context * mdct[CELT_BLOCK_NB]
Definition: opusenc.c:40
FFABS
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:72
AVCodecDefault
Definition: internal.h:231
OPUS_MODE_NB
@ OPUS_MODE_NB
Definition: opus.h:67
ff_celt_bitalloc
void ff_celt_bitalloc(CeltFrame *f, OpusRangeCoder *rc, int encode)
Definition: opus.c:555
OpusPacketInfo
Definition: opusenc.h:47
LIBAVUTIL_VERSION_INT
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:67
NULL
#define NULL
Definition: coverity.c:32
OPUS_BANDWITH_NB
@ OPUS_BANDWITH_NB
Definition: opus.h:77
opusenc.h
OpusEncContext::packet
OpusPacketInfo packet
Definition: opusenc.c:47
ff_celt_pvq_uninit
void av_cold ff_celt_pvq_uninit(CeltPVQ **pvq)
Definition: opus_pvq.c:914
period
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default minimum maximum flags name is the option keep it simple and lowercase description are in without period
Definition: writing_filters.txt:89
AVCodecContext::bit_rate
int64_t bit_rate
the average bitrate
Definition: avcodec.h:1615
av_default_item_name
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:191
ff_bufqueue_discard_all
static void ff_bufqueue_discard_all(struct FFBufQueue *queue)
Unref and remove all buffers from the queue.
Definition: bufferqueue.h:111
celt_frame_mdct
static void celt_frame_mdct(OpusEncContext *s, CeltFrame *f)
Definition: opusenc.c:190
CELT_MAX_BANDS
#define CELT_MAX_BANDS
Definition: opus.h:45
celt_quant_fine
static void celt_quant_fine(CeltFrame *f, OpusRangeCoder *rc)
Definition: opusenc.c:382
celt_frame_setup_input
static void celt_frame_setup_input(OpusEncContext *s, CeltFrame *f)
Definition: opusenc.c:116
opus_gen_toc
static int opus_gen_toc(OpusEncContext *s, uint8_t *toc, int *size, int *fsize_needed)
Definition: opusenc.c:73
ff_celt_freq_range
const uint8_t ff_celt_freq_range[]
Definition: opustab.c:767
index
int index
Definition: gxfenc.c:89
float_dsp.h
for
for(j=16;j >0;--j)
Definition: h264pred_template.c:469
CELT_ENERGY_SILENCE
#define CELT_ENERGY_SILENCE
Definition: opus_celt.h:45
OpusEncContext::channels
int channels
Definition: opusenc.c:49
options
const OptionDef options[]
opustab.h
ff_opus_rc_enc_init
void ff_opus_rc_enc_init(OpusRangeCoder *rc)
Definition: opus_rc.c:402
AVPacket::size
int size
Definition: avcodec.h:1478
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:188
opus_pvq.h
opusenc_psy.h
FFMAX
#define FFMAX(a, b)
Definition: common.h:94
bps
unsigned bps
Definition: movenc.c:1497
AV_SAMPLE_FMT_NONE
@ AV_SAMPLE_FMT_NONE
Definition: samplefmt.h:59
sample
#define sample
Definition: flacdsp_template.c:44
size
int size
Definition: twinvq_data.h:11134
OpusEncContext::enc_id
uint8_t enc_id[64]
Definition: opusenc.c:44
OpusRangeCoder
Definition: opus_rc.h:40
OpusEncContext::psyctx
OpusPsyContext psyctx
Definition: opusenc.c:36
AV_CODEC_ID_OPUS
@ AV_CODEC_ID_OPUS
Definition: avcodec.h:624
AVFloatDSPContext
Definition: float_dsp.h:24
AVFrame::format
int format
format of the frame, -1 if unknown or unset Values correspond to enum AVPixelFormat for video frames,...
Definition: frame.h:368
FFMIN
#define FFMIN(a, b)
Definition: common.h:96
ff_celt_tf_select
const int8_t ff_celt_tf_select[4][2][2][2]
Definition: opustab.c:777
offset
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
Definition: writing_filters.txt:86
ff_bufqueue_add
static void ff_bufqueue_add(void *log, struct FFBufQueue *queue, AVFrame *buf)
Add a buffer to the queue.
Definition: bufferqueue.h:71
write_opuslacing
static int write_opuslacing(uint8_t *dst, int v)
Definition: opusenc.c:482
OpusEncContext
Definition: opusenc.c:33
input
and forward the test the status of outputs and forward it to the corresponding return FFERROR_NOT_READY If the filters stores internally one or a few frame for some input
Definition: filter_design.txt:172
opus_write_extradata
static void opus_write_extradata(AVCodecContext *avctx)
Definition: opusenc.c:60
ff_celt_beta_coef
const float ff_celt_beta_coef[]
Definition: opustab.c:799
AVCodecContext::channels
int channels
number of audio channels
Definition: avcodec.h:2226
AV_OPT_TYPE_FLOAT
@ AV_OPT_TYPE_FLOAT
Definition: opt.h:226
DECLARE_ALIGNED
#define DECLARE_ALIGNED(n, t, v)
Definition: mem.h:112
src1
#define src1
Definition: h264pred.c:139
OpusEncContext::last_quantized_energy
float last_quantized_energy[OPUS_MAX_CHANNELS][CELT_MAX_BANDS]
Definition: opusenc.c:55
bytestream_put_buffer
static av_always_inline void bytestream_put_buffer(uint8_t **b, const uint8_t *src, unsigned int size)
Definition: bytestream.h:368
AVFrame::nb_samples
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:361
lrintf
#define lrintf(x)
Definition: libm_mips.h:70
ff_bufqueue_peek
static AVFrame * ff_bufqueue_peek(struct FFBufQueue *queue, unsigned index)
Get a buffer from the queue without altering it.
Definition: bufferqueue.h:87
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
AVPacket::pts
int64_t pts
Presentation timestamp in AVStream->time_base units; the time at which the decompressed packet will b...
Definition: avcodec.h:1470
ff_opus_rc_enc_laplace
void ff_opus_rc_enc_laplace(OpusRangeCoder *rc, int *value, uint32_t symbol, int decay)
Definition: opus_rc.c:314
av_get_bytes_per_sample
int av_get_bytes_per_sample(enum AVSampleFormat sample_fmt)
Return number of bytes per sample.
Definition: samplefmt.c:106
AVCodecContext::extradata
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
Definition: avcodec.h:1666
FF_CODEC_CAP_INIT_CLEANUP
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: internal.h:48
FFBufQueue
Structure holding the queue.
Definition: bufferqueue.h:49
AVFrame::extended_data
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:342
ff_opus_rc_put_raw
void ff_opus_rc_put_raw(OpusRangeCoder *rc, uint32_t val, uint32_t count)
CELT: write 0 - 31 bits to the rawbits buffer.
Definition: opus_rc.c:161
celt_encode_frame
static void celt_encode_frame(OpusEncContext *s, OpusRangeCoder *rc, CeltFrame *f, int index)
Definition: opusenc.c:417
AVSampleFormat
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
OPUS_MAX_CHANNELS
#define OPUS_MAX_CHANNELS
Definition: opusenc.h:34
uint8_t
uint8_t
Definition: audio_convert.c:194
ff_opus_psy_signal_eof
void ff_opus_psy_signal_eof(OpusPsyContext *s)
Definition: opusenc_psy.c:588
ff_mdct15_uninit
av_cold void ff_mdct15_uninit(MDCT15Context **ps)
Definition: mdct15.c:43
AVCodec::name
const char * name
Name of the codec implementation.
Definition: avcodec.h:3488
AV_PKT_DATA_SKIP_SAMPLES
@ AV_PKT_DATA_SKIP_SAMPLES
Recommmends skipping the specified number of samples.
Definition: avcodec.h:1300
len
int len
Definition: vorbis_enc_data.h:452
CELT_POSTFILTER_MINPERIOD
#define CELT_POSTFILTER_MINPERIOD
Definition: opus_celt.h:44
ff_opus_default_coupled_streams
const uint8_t ff_opus_default_coupled_streams[]
Definition: opustab.c:25
opusenc_class
static const AVClass opusenc_class
Definition: opusenc.c:707
ret
ret
Definition: filter_design.txt:187
OpusEncContext::avctx
AVCodecContext * avctx
Definition: opusenc.c:37
AVClass::class_name
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
frame
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
Definition: filter_design.txt:264
AV_INPUT_BUFFER_PADDING_SIZE
#define AV_INPUT_BUFFER_PADDING_SIZE
Definition: avcodec.h:790
OpusEncContext::enc_id_bits
int enc_id_bits
Definition: opusenc.c:45
left
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2] ... the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so ...,+,-,+,-,+,+,-,+,-,+,... hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32 - hcoeff[1] - hcoeff[2] - ... a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2} an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||......... intra?||||:Block01 :yes no||||:Block02 :....... ..........||||:Block03 ::y DC ::ref index:||||:Block04 ::cb DC ::motion x :||||......... :cr DC ::motion y :||||....... ..........|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------ ------------ ------------|||Y subbands||Cb subbands||Cr subbands||||--- ---||--- ---||--- ---|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------ ------------ ------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction ------------|\ Dequantization ------------------- \||Reference frames|\ IDWT|------- -------|Motion \|||Frame 0||Frame 1||Compensation . OBMC v -------|------- -------|--------------. \------> Frame n output Frame Frame<----------------------------------/|...|------------------- Range Coder:============Binary Range Coder:------------------- The implemented range coder is an adapted version based upon "Range encoding: an algorithm for removing redundancy from a digitised message." by G. N. N. Martin. The symbols encoded by the Snow range coder are bits(0|1). The associated probabilities are not fix but change depending on the symbol mix seen so far. bit seen|new state ---------+----------------------------------------------- 0|256 - state_transition_table[256 - old_state];1|state_transition_table[old_state];state_transition_table={ 0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:------------------------- FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1. the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled left
Definition: snow.txt:386
CELT_EMPH_COEFF
#define CELT_EMPH_COEFF
Definition: opusdsp.h:24
AVCodecContext
main external API structure.
Definition: avcodec.h:1565
ff_celt_model_energy_small
const uint16_t ff_celt_model_energy_small[]
Definition: opustab.c:761
celt_quant_coarse
static void celt_quant_coarse(CeltFrame *f, OpusRangeCoder *rc, float last_energy[][CELT_MAX_BANDS])
Definition: opusenc.c:362
ff_celt_quant_bands
void ff_celt_quant_bands(CeltFrame *f, OpusRangeCoder *rc)
Definition: opus.c:446
ff_opus_rc_enc_log
void ff_opus_rc_enc_log(OpusRangeCoder *rc, int val, uint32_t bits)
Definition: opus_rc.c:131
temp
else temp
Definition: vf_mcdeint.c:256
AV_CODEC_CAP_DELAY
#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:1006
ff_celt_alpha_coef
const float ff_celt_alpha_coef[]
Definition: opustab.c:795
celt_enc_quant_pfilter
static void celt_enc_quant_pfilter(OpusRangeCoder *rc, CeltFrame *f)
Definition: opusenc.c:282
celt_quant_final
static void celt_quant_final(OpusEncContext *s, OpusRangeCoder *rc, CeltFrame *f)
Definition: opusenc.c:399
AV_CODEC_FLAG_BITEXACT
#define AV_CODEC_FLAG_BITEXACT
Use only bitexact stuff (except (I)DCT).
Definition: avcodec.h:908
AVCodecContext::frame_number
int frame_number
Frame counter, set by libavcodec.
Definition: avcodec.h:2256
ff_opus_psy_init
av_cold int ff_opus_psy_init(OpusPsyContext *s, AVCodecContext *avctx, struct FFBufQueue *bufqueue, OpusEncOptions *options)
Definition: opusenc_psy.c:516
quant
const uint8_t * quant
Definition: vorbis_enc_data.h:458
diff
static av_always_inline int diff(const uint32_t a, const uint32_t b)
Definition: vf_palettegen.c:136
OpusEncOptions
Definition: opusenc.h:43
FFALIGN
#define FFALIGN(x, a)
Definition: macros.h:48
alpha
static const int16_t alpha[]
Definition: ilbcdata.h:55
AVCodecContext::priv_data
void * priv_data
Definition: avcodec.h:1592
AVPacket
This structure stores compressed data.
Definition: avcodec.h:1454
OpusEncContext::pvq
CeltPVQ * pvq
Definition: opusenc.c:41
av_freep
#define av_freep(p)
Definition: tableprint_vlc.h:35
avpriv_float_dsp_alloc
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:135
CELT_BLOCK_NB
@ CELT_BLOCK_NB
Definition: opus_celt.h:62
bytestream.h
block
The exact code depends on how similar the blocks are and how related they are to the block
Definition: filter_design.txt:207
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:28
celt_apply_preemph_filter
static void celt_apply_preemph_filter(OpusEncContext *s, CeltFrame *f)
Definition: opusenc.c:156
ff_celt_mean_energy
const float ff_celt_mean_energy[]
Definition: opustab.c:787
OpusEncContext::bufqueue
struct FFBufQueue bufqueue
Definition: opusenc.c:42
CELT_OVERLAP
#define CELT_OVERLAP
Definition: opus.h:42
OPUS_RC_CHECKPOINT_BITS
#define OPUS_RC_CHECKPOINT_BITS(rc)
Definition: opus_rc.h:120
AV_CODEC_CAP_SMALL_LAST_FRAME
#define AV_CODEC_CAP_SMALL_LAST_FRAME
Codec can be fed a final frame with a smaller size.
Definition: avcodec.h:1011
OPUS_RC_CHECKPOINT_ROLLBACK
#define OPUS_RC_CHECKPOINT_ROLLBACK(rc)
Definition: opus_rc.h:123
ff_alloc_packet2
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
celt_enc_tf
static void celt_enc_tf(CeltFrame *f, OpusRangeCoder *rc)
Definition: opusenc.c:255
opus_encode_init
static av_cold int opus_encode_init(AVCodecContext *avctx)
Definition: opusenc.c:621
ff_opus_rc_enc_end
void ff_opus_rc_enc_end(OpusRangeCoder *rc, uint8_t *dst, int size)
Definition: opus_rc.c:360
CeltFrame
Definition: opus_celt.h:92
OpusEncContext::dsp
AVFloatDSPContext * dsp
Definition: opusenc.c:39