libavcodec/aacenc.c

   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 libavcodec/aacenc.c
  24  * AAC encoder
  25  */
  26
  27 /***********************************
  28  *              TODOs:
  29  * psy model selection with some option
  30  * add sane pulse detection
  31  * add temporal noise shaping
  32  ***********************************/
  33
  34 #include "avcodec.h"
  35 #include "get_bits.h"
  36 #include "dsputil.h"
  37 #include "mpeg4audio.h"
  38
  39 #include "aacpsy.h"
  40 #include "aac.h"
  41 #include "aactab.h"
  42
  43 static const uint8_t swb_size_1024_96[] = {
  44     4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8,
  45     12, 12, 12, 12, 12, 16, 16, 24, 28, 36, 44,
  46     64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
  47 };
  48
  49 static const uint8_t swb_size_1024_64[] = {
  50     4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8,
  51     12, 12, 12, 16, 16, 16, 20, 24, 24, 28, 36,
  52     40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40
  53 };
  54
  55 static const uint8_t swb_size_1024_48[] = {
  56     4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8,
  57     12, 12, 12, 12, 16, 16, 20, 20, 24, 24, 28, 28,
  58     32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32,
  59     96
  60 };
  61
  62 static const uint8_t swb_size_1024_32[] = {
  63     4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8,
  64     12, 12, 12, 12, 16, 16, 20, 20, 24, 24, 28, 28,
  65     32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32
  66 };
  67
  68 static const uint8_t swb_size_1024_24[] = {
  69     4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
  70     12, 12, 12, 12, 16, 16, 16, 20, 20, 24, 24, 28, 28,
  71     32, 36, 36, 40, 44, 48, 52, 52, 64, 64, 64, 64, 64
  72 };
  73
  74 static const uint8_t swb_size_1024_16[] = {
  75     8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
  76     12, 12, 12, 12, 12, 12, 12, 12, 12, 16, 16, 16, 16, 20, 20, 20, 24, 24, 28, 28,
  77     32, 36, 40, 40, 44, 48, 52, 56, 60, 64, 64, 64
  78 };
  79
  80 static const uint8_t swb_size_1024_8[] = {
  81     12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12,
  82     16, 16, 16, 16, 16, 16, 16, 20, 20, 20, 20, 24, 24, 24, 28, 28,
  83     32, 36, 36, 40, 44, 48, 52, 56, 60, 64, 80
  84 };
  85
  86 static const uint8_t * const swb_size_1024[] = {
  87     swb_size_1024_96, swb_size_1024_96, swb_size_1024_64,
  88     swb_size_1024_48, swb_size_1024_48, swb_size_1024_32,
  89     swb_size_1024_24, swb_size_1024_24, swb_size_1024_16,
  90     swb_size_1024_16, swb_size_1024_16, swb_size_1024_8
  91 };
  92
  93 static const uint8_t swb_size_128_96[] = {
  94     4, 4, 4, 4, 4, 4, 8, 8, 8, 16, 28, 36
  95 };
  96
  97 static const uint8_t swb_size_128_48[] = {
  98     4, 4, 4, 4, 4, 8, 8, 8, 12, 12, 12, 16, 16, 16
  99 };
 100
 101 static const uint8_t swb_size_128_24[] = {
 102     4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 12, 12, 16, 16, 20
 103 };
 104
 105 static const uint8_t swb_size_128_16[] = {
 106     4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 12, 12, 16, 20, 20
 107 };
 108
 109 static const uint8_t swb_size_128_8[] = {
 110     4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 12, 16, 20, 20
 111 };
 112
 113 static const uint8_t * const swb_size_128[] = {
 114     /* the last entry on the following row is swb_size_128_64 but is a
 115        duplicate of swb_size_128_96 */
 116     swb_size_128_96, swb_size_128_96, swb_size_128_96,
 117     swb_size_128_48, swb_size_128_48, swb_size_128_48,
 118     swb_size_128_24, swb_size_128_24, swb_size_128_16,
 119     swb_size_128_16, swb_size_128_16, swb_size_128_8
 120 };
 121
 122 /** bits needed to code codebook run value for long windows */
 123 static const uint8_t run_value_bits_long[64] = {
 124      5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,
 125      5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5,  5, 10,
 126     10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10,
 127     10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 15
 128 };
 129
 130 /** bits needed to code codebook run value for short windows */
 131 static const uint8_t run_value_bits_short[16] = {
 132     3, 3, 3, 3, 3, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 9
 133 };
 134
 135 static const uint8_t* const run_value_bits[2] = {
 136     run_value_bits_long, run_value_bits_short
 137 };
 138
 139 /** default channel configurations */
 140 static const uint8_t aac_chan_configs[6][5] = {
 141  {1, TYPE_SCE},                               // 1 channel  - single channel element
 142  {1, TYPE_CPE},                               // 2 channels - channel pair
 143  {2, TYPE_SCE, TYPE_CPE},                     // 3 channels - center + stereo
 144  {3, TYPE_SCE, TYPE_CPE, TYPE_SCE},           // 4 channels - front center + stereo + back center
 145  {3, TYPE_SCE, TYPE_CPE, TYPE_CPE},           // 5 channels - front center + stereo + back stereo
 146  {4, TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_LFE}, // 6 channels - front center + stereo + back stereo + LFE
 147 };
 148
 149 /**
 150  * structure used in optimal codebook search
 151  */
 152 typedef struct BandCodingPath {
 153     int prev_idx; ///< pointer to the previous path point
 154     int codebook; ///< codebook for coding band run
 155     int bits;     ///< number of bit needed to code given number of bands
 156 } BandCodingPath;
 157
 158 /**
 159  * AAC encoder context
 160  */
 161 typedef struct {
 162     PutBitContext pb;
 163     MDCTContext mdct1024;                        ///< long (1024 samples) frame transform context
 164     MDCTContext mdct128;                         ///< short (128 samples) frame transform context
 165     DSPContext  dsp;
 166     DECLARE_ALIGNED_16(FFTSample, output[2048]); ///< temporary buffer for MDCT input coefficients
 167     int16_t* samples;                            ///< saved preprocessed input
 168
 169     int samplerate_index;                        ///< MPEG-4 samplerate index
 170
 171     ChannelElement *cpe;                         ///< channel elements
 172     AACPsyContext psy;                           ///< psychoacoustic model context
 173     int last_frame;
 174 } AACEncContext;
 175
 176 /**
 177  * Make AAC audio config object.
 178  * @see 1.6.2.1 "Syntax - AudioSpecificConfig"
 179  */
 180 static void put_audio_specific_config(AVCodecContext *avctx)
 181 {
 182     PutBitContext pb;
 183     AACEncContext *s = avctx->priv_data;
 184
 185     init_put_bits(&pb, avctx->extradata, avctx->extradata_size*8);
 186     put_bits(&pb, 5, 2); //object type - AAC-LC
 187     put_bits(&pb, 4, s->samplerate_index); //sample rate index
 188     put_bits(&pb, 4, avctx->channels);
 189     //GASpecificConfig
 190     put_bits(&pb, 1, 0); //frame length - 1024 samples
 191     put_bits(&pb, 1, 0); //does not depend on core coder
 192     put_bits(&pb, 1, 0); //is not extension
 193     flush_put_bits(&pb);
 194 }
 195
 196 static av_cold int aac_encode_init(AVCodecContext *avctx)
 197 {
 198     AACEncContext *s = avctx->priv_data;
 199     int i;
 200
 201     avctx->frame_size = 1024;
 202
 203     for(i = 0; i < 16; i++)
 204         if(avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
 205             break;
 206     if(i == 16){
 207         av_log(avctx, AV_LOG_ERROR, "Unsupported sample rate %d\n", avctx->sample_rate);
 208         return -1;
 209     }
 210     if(avctx->channels > 6){
 211         av_log(avctx, AV_LOG_ERROR, "Unsupported number of channels: %d\n", avctx->channels);
 212         return -1;
 213     }
 214     s->samplerate_index = i;
 215
 216     dsputil_init(&s->dsp, avctx);
 217     ff_mdct_init(&s->mdct1024, 11, 0, 1.0);
 218     ff_mdct_init(&s->mdct128,   8, 0, 1.0);
 219     // window init
 220     ff_kbd_window_init(ff_aac_kbd_long_1024, 4.0, 1024);
 221     ff_kbd_window_init(ff_aac_kbd_short_128, 6.0, 128);
 222     ff_sine_window_init(ff_sine_1024, 1024);
 223     ff_sine_window_init(ff_sine_128, 128);
 224
 225     s->samples = av_malloc(2 * 1024 * avctx->channels * sizeof(s->samples[0]));
 226     s->cpe = av_mallocz(sizeof(ChannelElement) * aac_chan_configs[avctx->channels-1][0]);
 227     if(ff_aac_psy_init(&s->psy, avctx, AAC_PSY_3GPP,
 228                        aac_chan_configs[avctx->channels-1][0], 0,
 229                        swb_size_1024[i], ff_aac_num_swb_1024[i], swb_size_128[i], ff_aac_num_swb_128[i]) < 0){
 230         av_log(avctx, AV_LOG_ERROR, "Cannot initialize selected model.\n");
 231         return -1;
 232     }
 233     avctx->extradata = av_malloc(2);
 234     avctx->extradata_size = 2;
 235     put_audio_specific_config(avctx);
 236     return 0;
 237 }
 238
 239 /**
 240  * Encode ics_info element.
 241  * @see Table 4.6 (syntax of ics_info)
 242  */
 243 static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
 244 {
 245     int i;
 246
 247     put_bits(&s->pb, 1, 0);                // ics_reserved bit
 248     put_bits(&s->pb, 2, info->window_sequence[0]);
 249     put_bits(&s->pb, 1, info->use_kb_window[0]);
 250     if(info->window_sequence[0] != EIGHT_SHORT_SEQUENCE){
 251         put_bits(&s->pb, 6, info->max_sfb);
 252         put_bits(&s->pb, 1, 0);            // no prediction
 253     }else{
 254         put_bits(&s->pb, 4, info->max_sfb);
 255         for(i = 1; i < info->num_windows; i++)
 256             put_bits(&s->pb, 1, info->group_len[i]);
 257     }
 258 }
 259
 260 /**
 261  * Calculate the number of bits needed to code all coefficient signs in current band.
 262  */
 263 static int calculate_band_sign_bits(AACEncContext *s, SingleChannelElement *sce,
 264                                     int group_len, int start, int size)
 265 {
 266     int bits = 0;
 267     int i, w;
 268     for(w = 0; w < group_len; w++){
 269         for(i = 0; i < size; i++){
 270             if(sce->icoefs[start + i])
 271                 bits++;
 272         }
 273         start += 128;
 274     }
 275     return bits;
 276 }
 277
 278 /**
 279  * Encode pulse data.
 280  */
 281 static void encode_pulses(AACEncContext *s, Pulse *pulse)
 282 {
 283     int i;
 284
 285     put_bits(&s->pb, 1, !!pulse->num_pulse);
 286     if(!pulse->num_pulse) return;
 287
 288     put_bits(&s->pb, 2, pulse->num_pulse - 1);
 289     put_bits(&s->pb, 6, pulse->start);
 290     for(i = 0; i < pulse->num_pulse; i++){
 291         put_bits(&s->pb, 5, pulse->pos[i]);
 292         put_bits(&s->pb, 4, pulse->amp[i]);
 293     }
 294 }
 295
 296 /**
 297  * Encode spectral coefficients processed by psychoacoustic model.
 298  */
 299 static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
 300 {
 301     int start, i, w, w2, wg;
 302
 303     w = 0;
 304     for(wg = 0; wg < sce->ics.num_window_groups; wg++){
 305         start = 0;
 306         for(i = 0; i < sce->ics.max_sfb; i++){
 307             if(sce->zeroes[w*16 + i]){
 308                 start += sce->ics.swb_sizes[i];
 309                 continue;
 310             }
 311             for(w2 = w; w2 < w + sce->ics.group_len[wg]; w2++){
 312                 encode_band_coeffs(s, sce, start + w2*128,
 313                                    sce->ics.swb_sizes[i],
 314                                    sce->band_type[w*16 + i]);
 315             }
 316             start += sce->ics.swb_sizes[i];
 317         }
 318         w += sce->ics.group_len[wg];
 319     }
 320 }
 321
 322 /**
 323  * Write some auxiliary information about the created AAC file.
 324  */
 325 static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s, const char *name)
 326 {
 327     int i, namelen, padbits;
 328
 329     namelen = strlen(name) + 2;
 330     put_bits(&s->pb, 3, TYPE_FIL);
 331     put_bits(&s->pb, 4, FFMIN(namelen, 15));
 332     if(namelen >= 15)
 333         put_bits(&s->pb, 8, namelen - 16);
 334     put_bits(&s->pb, 4, 0); //extension type - filler
 335     padbits = 8 - (put_bits_count(&s->pb) & 7);
 336     align_put_bits(&s->pb);
 337     for(i = 0; i < namelen - 2; i++)
 338         put_bits(&s->pb, 8, name[i]);
 339     put_bits(&s->pb, 12 - padbits, 0);
 340 }
 341
 342 static av_cold int aac_encode_end(AVCodecContext *avctx)
 343 {
 344     AACEncContext *s = avctx->priv_data;
 345
 346     ff_mdct_end(&s->mdct1024);
 347     ff_mdct_end(&s->mdct128);
 348     ff_aac_psy_end(&s->psy);
 349     av_freep(&s->samples);
 350     av_freep(&s->cpe);
 351     return 0;
 352 }
 353
 354 AVCodec aac_encoder = {
 355     "aac",
 356     CODEC_TYPE_AUDIO,
 357     CODEC_ID_AAC,
 358     sizeof(AACEncContext),
 359     aac_encode_init,
 360     aac_encode_frame,
 361     aac_encode_end,
 362     .capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
 363     .sample_fmts = (enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
 364     .long_name = NULL_IF_CONFIG_SMALL("Advanced Audio Coding"),
 365 };