3 * Copyright (C) 2008 Konstantin Shishkov
5 * This file is part of FFmpeg.
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.
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.
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
27 /***********************************
29 * add sane pulse detection
30 * add temporal noise shaping
31 ***********************************/
36 #include "mpeg4audio.h"
44 static const uint8_t swb_size_1024_96[] = {
45 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8,
46 12, 12, 12, 12, 12, 16, 16, 24, 28, 36, 44,
47 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
50 static const uint8_t swb_size_1024_64[] = {
51 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8,
52 12, 12, 12, 16, 16, 16, 20, 24, 24, 28, 36,
53 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40
56 static const uint8_t swb_size_1024_48[] = {
57 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8,
58 12, 12, 12, 12, 16, 16, 20, 20, 24, 24, 28, 28,
59 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32,
63 static const uint8_t swb_size_1024_32[] = {
64 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8,
65 12, 12, 12, 12, 16, 16, 20, 20, 24, 24, 28, 28,
66 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
69 static const uint8_t swb_size_1024_24[] = {
70 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
71 12, 12, 12, 12, 16, 16, 16, 20, 20, 24, 24, 28, 28,
72 32, 36, 36, 40, 44, 48, 52, 52, 64, 64, 64, 64, 64
75 static const uint8_t swb_size_1024_16[] = {
76 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
77 12, 12, 12, 12, 12, 12, 12, 12, 12, 16, 16, 16, 16, 20, 20, 20, 24, 24, 28, 28,
78 32, 36, 40, 40, 44, 48, 52, 56, 60, 64, 64, 64
81 static const uint8_t swb_size_1024_8[] = {
82 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
83 16, 16, 16, 16, 16, 16, 16, 20, 20, 20, 20, 24, 24, 24, 28, 28,
84 32, 36, 36, 40, 44, 48, 52, 56, 60, 64, 80
87 static const uint8_t *swb_size_1024[] = {
88 swb_size_1024_96, swb_size_1024_96, swb_size_1024_64,
89 swb_size_1024_48, swb_size_1024_48, swb_size_1024_32,
90 swb_size_1024_24, swb_size_1024_24, swb_size_1024_16,
91 swb_size_1024_16, swb_size_1024_16, swb_size_1024_8
94 static const uint8_t swb_size_128_96[] = {
95 4, 4, 4, 4, 4, 4, 8, 8, 8, 16, 28, 36
98 static const uint8_t swb_size_128_48[] = {
99 4, 4, 4, 4, 4, 8, 8, 8, 12, 12, 12, 16, 16, 16
102 static const uint8_t swb_size_128_24[] = {
103 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 12, 12, 16, 16, 20
106 static const uint8_t swb_size_128_16[] = {
107 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 12, 12, 16, 20, 20
110 static const uint8_t swb_size_128_8[] = {
111 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 12, 16, 20, 20
114 static const uint8_t *swb_size_128[] = {
115 /* the last entry on the following row is swb_size_128_64 but is a
116 duplicate of swb_size_128_96 */
117 swb_size_128_96, swb_size_128_96, swb_size_128_96,
118 swb_size_128_48, swb_size_128_48, swb_size_128_48,
119 swb_size_128_24, swb_size_128_24, swb_size_128_16,
120 swb_size_128_16, swb_size_128_16, swb_size_128_8
123 /** default channel configurations */
124 static const uint8_t aac_chan_configs[6][5] = {
125 {1, TYPE_SCE}, // 1 channel - single channel element
126 {1, TYPE_CPE}, // 2 channels - channel pair
127 {2, TYPE_SCE, TYPE_CPE}, // 3 channels - center + stereo
128 {3, TYPE_SCE, TYPE_CPE, TYPE_SCE}, // 4 channels - front center + stereo + back center
129 {3, TYPE_SCE, TYPE_CPE, TYPE_CPE}, // 5 channels - front center + stereo + back stereo
130 {4, TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_LFE}, // 6 channels - front center + stereo + back stereo + LFE
134 * Make AAC audio config object.
135 * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
137 static void put_audio_specific_config(AVCodecContext *avctx)
140 AACEncContext *s = avctx->priv_data;
142 init_put_bits(&pb, avctx->extradata, avctx->extradata_size*8);
143 put_bits(&pb, 5, 2); //object type - AAC-LC
144 put_bits(&pb, 4, s->samplerate_index); //sample rate index
145 put_bits(&pb, 4, avctx->channels);
147 put_bits(&pb, 1, 0); //frame length - 1024 samples
148 put_bits(&pb, 1, 0); //does not depend on core coder
149 put_bits(&pb, 1, 0); //is not extension
153 static av_cold int aac_encode_init(AVCodecContext *avctx)
155 AACEncContext *s = avctx->priv_data;
157 const uint8_t *sizes[2];
160 avctx->frame_size = 1024;
162 for (i = 0; i < 16; i++)
163 if (avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
166 av_log(avctx, AV_LOG_ERROR, "Unsupported sample rate %d\n", avctx->sample_rate);
169 if (avctx->channels > 6) {
170 av_log(avctx, AV_LOG_ERROR, "Unsupported number of channels: %d\n", avctx->channels);
173 if (avctx->profile != FF_PROFILE_UNKNOWN && avctx->profile != FF_PROFILE_AAC_LOW) {
174 av_log(avctx, AV_LOG_ERROR, "Unsupported profile %d\n", avctx->profile);
177 if (1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * avctx->channels) {
178 av_log(avctx, AV_LOG_ERROR, "Too many bits per frame requested\n");
181 s->samplerate_index = i;
183 dsputil_init(&s->dsp, avctx);
184 ff_mdct_init(&s->mdct1024, 11, 0, 1.0);
185 ff_mdct_init(&s->mdct128, 8, 0, 1.0);
187 ff_kbd_window_init(ff_aac_kbd_long_1024, 4.0, 1024);
188 ff_kbd_window_init(ff_aac_kbd_short_128, 6.0, 128);
189 ff_init_ff_sine_windows(10);
190 ff_init_ff_sine_windows(7);
192 s->samples = av_malloc(2 * 1024 * avctx->channels * sizeof(s->samples[0]));
193 s->cpe = av_mallocz(sizeof(ChannelElement) * aac_chan_configs[avctx->channels-1][0]);
194 avctx->extradata = av_mallocz(2 + FF_INPUT_BUFFER_PADDING_SIZE);
195 avctx->extradata_size = 2;
196 put_audio_specific_config(avctx);
198 sizes[0] = swb_size_1024[i];
199 sizes[1] = swb_size_128[i];
200 lengths[0] = ff_aac_num_swb_1024[i];
201 lengths[1] = ff_aac_num_swb_128[i];
202 ff_psy_init(&s->psy, avctx, 2, sizes, lengths);
203 s->psypp = ff_psy_preprocess_init(avctx);
204 s->coder = &ff_aac_coders[2];
206 s->lambda = avctx->global_quality ? avctx->global_quality : 120;
210 if (avctx->channels > 5)
211 av_log(avctx, AV_LOG_ERROR, "This encoder does not yet enforce the restrictions on LFEs. "
212 "The output will most likely be an illegal bitstream.\n");
217 static void apply_window_and_mdct(AVCodecContext *avctx, AACEncContext *s,
218 SingleChannelElement *sce, short *audio, int channel)
221 const float * lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
222 const float * swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
223 const float * pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
225 if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
226 memcpy(s->output, sce->saved, sizeof(float)*1024);
227 if (sce->ics.window_sequence[0] == LONG_STOP_SEQUENCE) {
228 memset(s->output, 0, sizeof(s->output[0]) * 448);
229 for (i = 448; i < 576; i++)
230 s->output[i] = sce->saved[i] * pwindow[i - 448];
231 for (i = 576; i < 704; i++)
232 s->output[i] = sce->saved[i];
234 if (sce->ics.window_sequence[0] != LONG_START_SEQUENCE) {
235 for (i = 0, j = channel; i < 1024; i++, j += avctx->channels) {
236 s->output[i+1024] = audio[j] * lwindow[1024 - i - 1];
237 sce->saved[i] = audio[j] * lwindow[i];
240 for (i = 0, j = channel; i < 448; i++, j += avctx->channels)
241 s->output[i+1024] = audio[j];
242 for (; i < 576; i++, j += avctx->channels)
243 s->output[i+1024] = audio[j] * swindow[576 - i - 1];
244 memset(s->output+1024+576, 0, sizeof(s->output[0]) * 448);
245 for (i = 0, j = channel; i < 1024; i++, j += avctx->channels)
246 sce->saved[i] = audio[j];
248 ff_mdct_calc(&s->mdct1024, sce->coeffs, s->output);
250 for (k = 0; k < 1024; k += 128) {
251 for (i = 448 + k; i < 448 + k + 256; i++)
252 s->output[i - 448 - k] = (i < 1024)
254 : audio[channel + (i-1024)*avctx->channels];
255 s->dsp.vector_fmul (s->output, k ? swindow : pwindow, 128);
256 s->dsp.vector_fmul_reverse(s->output+128, s->output+128, swindow, 128);
257 ff_mdct_calc(&s->mdct128, sce->coeffs + k, s->output);
259 for (i = 0, j = channel; i < 1024; i++, j += avctx->channels)
260 sce->saved[i] = audio[j];
265 * Encode ics_info element.
266 * @see Table 4.6 (syntax of ics_info)
268 static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
272 put_bits(&s->pb, 1, 0); // ics_reserved bit
273 put_bits(&s->pb, 2, info->window_sequence[0]);
274 put_bits(&s->pb, 1, info->use_kb_window[0]);
275 if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
276 put_bits(&s->pb, 6, info->max_sfb);
277 put_bits(&s->pb, 1, 0); // no prediction
279 put_bits(&s->pb, 4, info->max_sfb);
280 for (w = 1; w < 8; w++)
281 put_bits(&s->pb, 1, !info->group_len[w]);
287 * @see 4.6.8.1 "Joint Coding - M/S Stereo"
289 static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
293 put_bits(pb, 2, cpe->ms_mode);
294 if (cpe->ms_mode == 1)
295 for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
296 for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
297 put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
301 * Produce integer coefficients from scalefactors provided by the model.
303 static void adjust_frame_information(AACEncContext *apc, ChannelElement *cpe, int chans)
306 int start, sum, maxsfb, cmaxsfb;
308 for (ch = 0; ch < chans; ch++) {
309 IndividualChannelStream *ics = &cpe->ch[ch].ics;
312 cpe->ch[ch].pulse.num_pulse = 0;
313 for (w = 0; w < ics->num_windows*16; w += 16) {
314 for (g = 0; g < ics->num_swb; g++) {
317 if (!ch && cpe->ms_mask[w + g]) {
318 for (i = 0; i < ics->swb_sizes[g]; i++) {
319 cpe->ch[0].coeffs[start+i] = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) / 2.0;
320 cpe->ch[1].coeffs[start+i] = cpe->ch[0].coeffs[start+i] - cpe->ch[1].coeffs[start+i];
323 start += ics->swb_sizes[g];
325 for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w+cmaxsfb-1]; cmaxsfb--)
327 maxsfb = FFMAX(maxsfb, cmaxsfb);
329 ics->max_sfb = maxsfb;
331 //adjust zero bands for window groups
332 for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
333 for (g = 0; g < ics->max_sfb; g++) {
335 for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
336 if (!cpe->ch[ch].zeroes[w2*16 + g]) {
341 cpe->ch[ch].zeroes[w*16 + g] = i;
346 if (chans > 1 && cpe->common_window) {
347 IndividualChannelStream *ics0 = &cpe->ch[0].ics;
348 IndividualChannelStream *ics1 = &cpe->ch[1].ics;
350 ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
351 ics1->max_sfb = ics0->max_sfb;
352 for (w = 0; w < ics0->num_windows*16; w += 16)
353 for (i = 0; i < ics0->max_sfb; i++)
354 if (cpe->ms_mask[w+i])
356 if (msc == 0 || ics0->max_sfb == 0)
359 cpe->ms_mode = msc < ics0->max_sfb ? 1 : 2;
364 * Encode scalefactor band coding type.
366 static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
370 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
371 s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
375 * Encode scalefactors.
377 static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s,
378 SingleChannelElement *sce)
380 int off = sce->sf_idx[0], diff;
383 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
384 for (i = 0; i < sce->ics.max_sfb; i++) {
385 if (!sce->zeroes[w*16 + i]) {
386 diff = sce->sf_idx[w*16 + i] - off + SCALE_DIFF_ZERO;
387 if (diff < 0 || diff > 120)
388 av_log(avctx, AV_LOG_ERROR, "Scalefactor difference is too big to be coded\n");
389 off = sce->sf_idx[w*16 + i];
390 put_bits(&s->pb, ff_aac_scalefactor_bits[diff], ff_aac_scalefactor_code[diff]);
399 static void encode_pulses(AACEncContext *s, Pulse *pulse)
403 put_bits(&s->pb, 1, !!pulse->num_pulse);
404 if (!pulse->num_pulse)
407 put_bits(&s->pb, 2, pulse->num_pulse - 1);
408 put_bits(&s->pb, 6, pulse->start);
409 for (i = 0; i < pulse->num_pulse; i++) {
410 put_bits(&s->pb, 5, pulse->pos[i]);
411 put_bits(&s->pb, 4, pulse->amp[i]);
416 * Encode spectral coefficients processed by psychoacoustic model.
418 static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
422 for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
424 for (i = 0; i < sce->ics.max_sfb; i++) {
425 if (sce->zeroes[w*16 + i]) {
426 start += sce->ics.swb_sizes[i];
429 for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++)
430 s->coder->quantize_and_encode_band(s, &s->pb, sce->coeffs + start + w2*128,
431 sce->ics.swb_sizes[i],
432 sce->sf_idx[w*16 + i],
433 sce->band_type[w*16 + i],
435 start += sce->ics.swb_sizes[i];
441 * Encode one channel of audio data.
443 static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s,
444 SingleChannelElement *sce,
447 put_bits(&s->pb, 8, sce->sf_idx[0]);
449 put_ics_info(s, &sce->ics);
450 encode_band_info(s, sce);
451 encode_scale_factors(avctx, s, sce);
452 encode_pulses(s, &sce->pulse);
453 put_bits(&s->pb, 1, 0); //tns
454 put_bits(&s->pb, 1, 0); //ssr
455 encode_spectral_coeffs(s, sce);
460 * Write some auxiliary information about the created AAC file.
462 static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s,
465 int i, namelen, padbits;
467 namelen = strlen(name) + 2;
468 put_bits(&s->pb, 3, TYPE_FIL);
469 put_bits(&s->pb, 4, FFMIN(namelen, 15));
471 put_bits(&s->pb, 8, namelen - 16);
472 put_bits(&s->pb, 4, 0); //extension type - filler
473 padbits = 8 - (put_bits_count(&s->pb) & 7);
474 align_put_bits(&s->pb);
475 for (i = 0; i < namelen - 2; i++)
476 put_bits(&s->pb, 8, name[i]);
477 put_bits(&s->pb, 12 - padbits, 0);
480 static int aac_encode_frame(AVCodecContext *avctx,
481 uint8_t *frame, int buf_size, void *data)
483 AACEncContext *s = avctx->priv_data;
484 int16_t *samples = s->samples, *samples2, *la;
486 int i, j, chans, tag, start_ch;
487 const uint8_t *chan_map = aac_chan_configs[avctx->channels-1];
488 int chan_el_counter[4];
489 FFPsyWindowInfo windows[avctx->channels];
495 memcpy(s->samples + 1024 * avctx->channels, data,
496 1024 * avctx->channels * sizeof(s->samples[0]));
499 samples2 = s->samples + 1024 * avctx->channels;
500 for (i = 0; i < chan_map[0]; i++) {
502 chans = tag == TYPE_CPE ? 2 : 1;
503 ff_psy_preprocess(s->psypp, (uint16_t*)data + start_ch,
504 samples2 + start_ch, start_ch, chans);
509 if (!avctx->frame_number) {
510 memcpy(s->samples, s->samples + 1024 * avctx->channels,
511 1024 * avctx->channels * sizeof(s->samples[0]));
516 for (i = 0; i < chan_map[0]; i++) {
517 FFPsyWindowInfo* wi = windows + start_ch;
519 chans = tag == TYPE_CPE ? 2 : 1;
521 samples2 = samples + start_ch;
522 la = samples2 + (448+64) * avctx->channels + start_ch;
525 for (j = 0; j < chans; j++) {
526 IndividualChannelStream *ics = &cpe->ch[j].ics;
528 wi[j] = ff_psy_suggest_window(&s->psy, samples2, la, start_ch + j, ics->window_sequence[0]);
529 ics->window_sequence[1] = ics->window_sequence[0];
530 ics->window_sequence[0] = wi[j].window_type[0];
531 ics->use_kb_window[1] = ics->use_kb_window[0];
532 ics->use_kb_window[0] = wi[j].window_shape;
533 ics->num_windows = wi[j].num_windows;
534 ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
535 ics->num_swb = s->psy.num_bands[ics->num_windows == 8];
536 for (k = 0; k < ics->num_windows; k++)
537 ics->group_len[k] = wi[j].grouping[k];
539 s->cur_channel = start_ch + j;
540 apply_window_and_mdct(avctx, s, &cpe->ch[j], samples2, j);
546 init_put_bits(&s->pb, frame, buf_size*8);
547 if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT))
548 put_bitstream_info(avctx, s, LIBAVCODEC_IDENT);
550 memset(chan_el_counter, 0, sizeof(chan_el_counter));
551 for (i = 0; i < chan_map[0]; i++) {
552 FFPsyWindowInfo* wi = windows + start_ch;
554 chans = tag == TYPE_CPE ? 2 : 1;
556 for (j = 0; j < chans; j++) {
557 s->cur_channel = start_ch + j;
558 ff_psy_set_band_info(&s->psy, s->cur_channel, cpe->ch[j].coeffs, &wi[j]);
559 s->coder->search_for_quantizers(avctx, s, &cpe->ch[j], s->lambda);
561 cpe->common_window = 0;
563 && wi[0].window_type[0] == wi[1].window_type[0]
564 && wi[0].window_shape == wi[1].window_shape) {
566 cpe->common_window = 1;
567 for (j = 0; j < wi[0].num_windows; j++) {
568 if (wi[0].grouping[j] != wi[1].grouping[j]) {
569 cpe->common_window = 0;
574 s->cur_channel = start_ch;
575 if (cpe->common_window && s->coder->search_for_ms)
576 s->coder->search_for_ms(s, cpe, s->lambda);
577 adjust_frame_information(s, cpe, chans);
578 put_bits(&s->pb, 3, tag);
579 put_bits(&s->pb, 4, chan_el_counter[tag]++);
581 put_bits(&s->pb, 1, cpe->common_window);
582 if (cpe->common_window) {
583 put_ics_info(s, &cpe->ch[0].ics);
584 encode_ms_info(&s->pb, cpe);
587 for (j = 0; j < chans; j++) {
588 s->cur_channel = start_ch + j;
589 encode_individual_channel(avctx, s, &cpe->ch[j], cpe->common_window);
594 frame_bits = put_bits_count(&s->pb);
595 if (frame_bits <= 6144 * avctx->channels - 3)
598 s->lambda *= avctx->bit_rate * 1024.0f / avctx->sample_rate / frame_bits;
602 put_bits(&s->pb, 3, TYPE_END);
603 flush_put_bits(&s->pb);
604 avctx->frame_bits = put_bits_count(&s->pb);
606 // rate control stuff
607 if (!(avctx->flags & CODEC_FLAG_QSCALE)) {
608 float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
610 s->lambda = FFMIN(s->lambda, 65536.f);
615 memcpy(s->samples, s->samples + 1024 * avctx->channels,
616 1024 * avctx->channels * sizeof(s->samples[0]));
617 return put_bits_count(&s->pb)>>3;
620 static av_cold int aac_encode_end(AVCodecContext *avctx)
622 AACEncContext *s = avctx->priv_data;
624 ff_mdct_end(&s->mdct1024);
625 ff_mdct_end(&s->mdct128);
627 ff_psy_preprocess_end(s->psypp);
628 av_freep(&s->samples);
633 AVCodec aac_encoder = {
637 sizeof(AACEncContext),
641 .capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY | CODEC_CAP_EXPERIMENTAL,
642 .sample_fmts = (const enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
643 .long_name = NULL_IF_CONFIG_SMALL("Advanced Audio Coding"),