libavcodec/wmaenc.c

   1 /*
   2  * WMA compatible encoder
   3  * Copyright (c) 2007 Michael Niedermayer
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 #include "avcodec.h"
  23 #include "internal.h"
  24 #include "wma.h"
  25
  26 #undef NDEBUG
  27 #include <assert.h>
  28
  29
  30 static int encode_init(AVCodecContext * avctx){
  31     WMACodecContext *s = avctx->priv_data;
  32     int i, flags1, flags2, block_align;
  33     uint8_t *extradata;
  34
  35     s->avctx = avctx;
  36
  37     if(avctx->channels > MAX_CHANNELS) {
  38         av_log(avctx, AV_LOG_ERROR, "too many channels: got %i, need %i or fewer",
  39                avctx->channels, MAX_CHANNELS);
  40         return AVERROR(EINVAL);
  41     }
  42
  43     if (avctx->sample_rate > 48000) {
  44         av_log(avctx, AV_LOG_ERROR, "sample rate is too high: %d > 48kHz",
  45                avctx->sample_rate);
  46         return AVERROR(EINVAL);
  47     }
  48
  49     if(avctx->bit_rate < 24*1000) {
  50         av_log(avctx, AV_LOG_ERROR, "bitrate too low: got %i, need 24000 or higher\n",
  51                avctx->bit_rate);
  52         return AVERROR(EINVAL);
  53     }
  54
  55 #if FF_API_OLD_ENCODE_AUDIO
  56     if (!(avctx->coded_frame = avcodec_alloc_frame()))
  57         return AVERROR(ENOMEM);
  58 #endif
  59
  60     /* extract flag infos */
  61     flags1 = 0;
  62     flags2 = 1;
  63     if (avctx->codec->id == AV_CODEC_ID_WMAV1) {
  64         extradata= av_malloc(4);
  65         avctx->extradata_size= 4;
  66         AV_WL16(extradata, flags1);
  67         AV_WL16(extradata+2, flags2);
  68     } else if (avctx->codec->id == AV_CODEC_ID_WMAV2) {
  69         extradata= av_mallocz(10);
  70         avctx->extradata_size= 10;
  71         AV_WL32(extradata, flags1);
  72         AV_WL16(extradata+4, flags2);
  73     }else
  74         assert(0);
  75     avctx->extradata= extradata;
  76     s->use_exp_vlc = flags2 & 0x0001;
  77     s->use_bit_reservoir = flags2 & 0x0002;
  78     s->use_variable_block_len = flags2 & 0x0004;
  79     if (avctx->channels == 2)
  80         s->ms_stereo = 1;
  81
  82     ff_wma_init(avctx, flags2);
  83
  84     /* init MDCT */
  85     for(i = 0; i < s->nb_block_sizes; i++)
  86         ff_mdct_init(&s->mdct_ctx[i], s->frame_len_bits - i + 1, 0, 1.0);
  87
  88     block_align        = avctx->bit_rate * (int64_t)s->frame_len /
  89                          (avctx->sample_rate * 8);
  90     block_align        = FFMIN(block_align, MAX_CODED_SUPERFRAME_SIZE);
  91     avctx->block_align = block_align;
  92     avctx->bit_rate    = avctx->block_align * 8LL * avctx->sample_rate /
  93                          s->frame_len;
  94     avctx->frame_size = avctx->delay = s->frame_len;
  95
  96     return 0;
  97 }
  98
  99
 100 static void apply_window_and_mdct(AVCodecContext * avctx, const AVFrame *frame)
 101 {
 102     WMACodecContext *s = avctx->priv_data;
 103     float **audio      = (float **)frame->extended_data;
 104     int len            = frame->nb_samples;
 105     int window_index= s->frame_len_bits - s->block_len_bits;
 106     FFTContext *mdct = &s->mdct_ctx[window_index];
 107     int ch;
 108     const float * win = s->windows[window_index];
 109     int window_len = 1 << s->block_len_bits;
 110     float n = 2.0 * 32768.0 / window_len;
 111
 112     for (ch = 0; ch < avctx->channels; ch++) {
 113         memcpy(s->output, s->frame_out[ch], window_len * sizeof(*s->output));
 114         s->fdsp.vector_fmul_scalar(s->frame_out[ch], audio[ch], n, len);
 115         s->fdsp.vector_fmul_reverse(&s->output[window_len], s->frame_out[ch], win, len);
 116         s->fdsp.vector_fmul(s->frame_out[ch], s->frame_out[ch], win, len);
 117         mdct->mdct_calc(mdct, s->coefs[ch], s->output);
 118     }
 119 }
 120
 121 //FIXME use for decoding too
 122 static void init_exp(WMACodecContext *s, int ch, const int *exp_param){
 123     int n;
 124     const uint16_t *ptr;
 125     float v, *q, max_scale, *q_end;
 126
 127     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 128     q = s->exponents[ch];
 129     q_end = q + s->block_len;
 130     max_scale = 0;
 131     while (q < q_end) {
 132         /* XXX: use a table */
 133         v = pow(10, *exp_param++ * (1.0 / 16.0));
 134         max_scale= FFMAX(max_scale, v);
 135         n = *ptr++;
 136         do {
 137             *q++ = v;
 138         } while (--n);
 139     }
 140     s->max_exponent[ch] = max_scale;
 141 }
 142
 143 static void encode_exp_vlc(WMACodecContext *s, int ch, const int *exp_param){
 144     int last_exp;
 145     const uint16_t *ptr;
 146     float *q, *q_end;
 147
 148     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 149     q = s->exponents[ch];
 150     q_end = q + s->block_len;
 151     if (s->version == 1) {
 152         last_exp= *exp_param++;
 153         assert(last_exp-10 >= 0 && last_exp-10 < 32);
 154         put_bits(&s->pb, 5, last_exp - 10);
 155         q+= *ptr++;
 156     }else
 157         last_exp = 36;
 158     while (q < q_end) {
 159         int exp = *exp_param++;
 160         int code = exp - last_exp + 60;
 161         assert(code >= 0 && code < 120);
 162         put_bits(&s->pb, ff_aac_scalefactor_bits[code], ff_aac_scalefactor_code[code]);
 163         /* XXX: use a table */
 164         q+= *ptr++;
 165         last_exp= exp;
 166     }
 167 }
 168
 169 static int encode_block(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], int total_gain){
 170     int v, bsize, ch, coef_nb_bits, parse_exponents;
 171     float mdct_norm;
 172     int nb_coefs[MAX_CHANNELS];
 173     static const int fixed_exp[25]={20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20};
 174
 175     //FIXME remove duplication relative to decoder
 176     if (s->use_variable_block_len) {
 177         assert(0); //FIXME not implemented
 178     }else{
 179         /* fixed block len */
 180         s->next_block_len_bits = s->frame_len_bits;
 181         s->prev_block_len_bits = s->frame_len_bits;
 182         s->block_len_bits = s->frame_len_bits;
 183     }
 184
 185     s->block_len = 1 << s->block_len_bits;
 186 //     assert((s->block_pos + s->block_len) <= s->frame_len);
 187     bsize = s->frame_len_bits - s->block_len_bits;
 188
 189     //FIXME factor
 190     v = s->coefs_end[bsize] - s->coefs_start;
 191     for (ch = 0; ch < s->avctx->channels; ch++)
 192         nb_coefs[ch] = v;
 193     {
 194         int n4 = s->block_len / 2;
 195         mdct_norm = 1.0 / (float)n4;
 196         if (s->version == 1) {
 197             mdct_norm *= sqrt(n4);
 198         }
 199     }
 200
 201     if (s->avctx->channels == 2) {
 202         put_bits(&s->pb, 1, !!s->ms_stereo);
 203     }
 204
 205     for (ch = 0; ch < s->avctx->channels; ch++) {
 206         s->channel_coded[ch] = 1; //FIXME only set channel_coded when needed, instead of always
 207         if (s->channel_coded[ch]) {
 208             init_exp(s, ch, fixed_exp);
 209         }
 210     }
 211
 212     for (ch = 0; ch < s->avctx->channels; ch++) {
 213         if (s->channel_coded[ch]) {
 214             WMACoef *coefs1;
 215             float *coefs, *exponents, mult;
 216             int i, n;
 217
 218             coefs1 = s->coefs1[ch];
 219             exponents = s->exponents[ch];
 220             mult = pow(10, total_gain * 0.05) / s->max_exponent[ch];
 221             mult *= mdct_norm;
 222             coefs = src_coefs[ch];
 223             if (s->use_noise_coding && 0) {
 224                 assert(0); //FIXME not implemented
 225             } else {
 226                 coefs += s->coefs_start;
 227                 n = nb_coefs[ch];
 228                 for(i = 0;i < n; i++){
 229                     double t= *coefs++ / (exponents[i] * mult);
 230                     if(t<-32768 || t>32767)
 231                         return -1;
 232
 233                     coefs1[i] = lrint(t);
 234                 }
 235             }
 236         }
 237     }
 238
 239     v = 0;
 240     for (ch = 0; ch < s->avctx->channels; ch++) {
 241         int a = s->channel_coded[ch];
 242         put_bits(&s->pb, 1, a);
 243         v |= a;
 244     }
 245
 246     if (!v)
 247         return 1;
 248
 249     for(v= total_gain-1; v>=127; v-= 127)
 250         put_bits(&s->pb, 7, 127);
 251     put_bits(&s->pb, 7, v);
 252
 253     coef_nb_bits= ff_wma_total_gain_to_bits(total_gain);
 254
 255     if (s->use_noise_coding) {
 256         for (ch = 0; ch < s->avctx->channels; ch++) {
 257             if (s->channel_coded[ch]) {
 258                 int i, n;
 259                 n = s->exponent_high_sizes[bsize];
 260                 for(i=0;i<n;i++) {
 261                     put_bits(&s->pb, 1, s->high_band_coded[ch][i]= 0);
 262                     if (0)
 263                         nb_coefs[ch] -= s->exponent_high_bands[bsize][i];
 264                 }
 265             }
 266         }
 267     }
 268
 269     parse_exponents = 1;
 270     if (s->block_len_bits != s->frame_len_bits) {
 271         put_bits(&s->pb, 1, parse_exponents);
 272     }
 273
 274     if (parse_exponents) {
 275         for (ch = 0; ch < s->avctx->channels; ch++) {
 276             if (s->channel_coded[ch]) {
 277                 if (s->use_exp_vlc) {
 278                     encode_exp_vlc(s, ch, fixed_exp);
 279                 } else {
 280                     assert(0); //FIXME not implemented
 281 //                    encode_exp_lsp(s, ch);
 282                 }
 283             }
 284         }
 285     } else {
 286         assert(0); //FIXME not implemented
 287     }
 288
 289     for (ch = 0; ch < s->avctx->channels; ch++) {
 290         if (s->channel_coded[ch]) {
 291             int run, tindex;
 292             WMACoef *ptr, *eptr;
 293             tindex = (ch == 1 && s->ms_stereo);
 294             ptr = &s->coefs1[ch][0];
 295             eptr = ptr + nb_coefs[ch];
 296
 297             run=0;
 298             for(;ptr < eptr; ptr++){
 299                 if(*ptr){
 300                     int level= *ptr;
 301                     int abs_level= FFABS(level);
 302                     int code= 0;
 303                     if(abs_level <= s->coef_vlcs[tindex]->max_level){
 304                         if(run < s->coef_vlcs[tindex]->levels[abs_level-1])
 305                             code= run + s->int_table[tindex][abs_level-1];
 306                     }
 307
 308                     assert(code < s->coef_vlcs[tindex]->n);
 309                     put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[code], s->coef_vlcs[tindex]->huffcodes[code]);
 310
 311                     if(code == 0){
 312                         if(1<<coef_nb_bits <= abs_level)
 313                             return -1;
 314
 315                         put_bits(&s->pb, coef_nb_bits, abs_level);
 316                         put_bits(&s->pb, s->frame_len_bits, run);
 317                     }
 318                     put_bits(&s->pb, 1, level < 0); //FIXME the sign is fliped somewhere
 319                     run=0;
 320                 }else{
 321                     run++;
 322                 }
 323             }
 324             if(run)
 325                 put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[1], s->coef_vlcs[tindex]->huffcodes[1]);
 326         }
 327         if (s->version == 1 && s->avctx->channels >= 2) {
 328             avpriv_align_put_bits(&s->pb);
 329         }
 330     }
 331     return 0;
 332 }
 333
 334 static int encode_frame(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], uint8_t *buf, int buf_size, int total_gain){
 335     init_put_bits(&s->pb, buf, buf_size);
 336
 337     if (s->use_bit_reservoir) {
 338         assert(0);//FIXME not implemented
 339     }else{
 340         if(encode_block(s, src_coefs, total_gain) < 0)
 341             return INT_MAX;
 342     }
 343
 344     avpriv_align_put_bits(&s->pb);
 345
 346     return put_bits_count(&s->pb) / 8 - s->avctx->block_align;
 347 }
 348
 349 static int encode_superframe(AVCodecContext *avctx, AVPacket *avpkt,
 350                              const AVFrame *frame, int *got_packet_ptr)
 351 {
 352     WMACodecContext *s = avctx->priv_data;
 353     int i, total_gain, ret;
 354
 355     s->block_len_bits= s->frame_len_bits; //required by non variable block len
 356     s->block_len = 1 << s->block_len_bits;
 357
 358     apply_window_and_mdct(avctx, frame);
 359
 360     if (s->ms_stereo) {
 361         float a, b;
 362         int i;
 363
 364         for(i = 0; i < s->block_len; i++) {
 365             a = s->coefs[0][i]*0.5;
 366             b = s->coefs[1][i]*0.5;
 367             s->coefs[0][i] = a + b;
 368             s->coefs[1][i] = a - b;
 369         }
 370     }
 371
 372     if ((ret = ff_alloc_packet(avpkt, 2 * MAX_CODED_SUPERFRAME_SIZE))) {
 373         av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
 374         return ret;
 375     }
 376
 377 #if 1
 378     total_gain= 128;
 379     for(i=64; i; i>>=1){
 380         int error = encode_frame(s, s->coefs, avpkt->data, avpkt->size,
 381                                  total_gain - i);
 382         if(error<0)
 383             total_gain-= i;
 384     }
 385 #else
 386     total_gain= 90;
 387     best = encode_frame(s, s->coefs, avpkt->data, avpkt->size, total_gain);
 388     for(i=32; i; i>>=1){
 389         int scoreL = encode_frame(s, s->coefs, avpkt->data, avpkt->size, total_gain - i);
 390         int scoreR = encode_frame(s, s->coefs, avpkt->data, avpkt->size, total_gain + i);
 391         av_log(NULL, AV_LOG_ERROR, "%d %d %d (%d)\n", scoreL, best, scoreR, total_gain);
 392         if(scoreL < FFMIN(best, scoreR)){
 393             best = scoreL;
 394             total_gain -= i;
 395         }else if(scoreR < best){
 396             best = scoreR;
 397             total_gain += i;
 398         }
 399     }
 400 #endif
 401
 402     if ((i = encode_frame(s, s->coefs, avpkt->data, avpkt->size, total_gain)) >= 0) {
 403         av_log(avctx, AV_LOG_ERROR, "required frame size too large. please "
 404                "use a higher bit rate.\n");
 405         return AVERROR(EINVAL);
 406     }
 407     assert((put_bits_count(&s->pb) & 7) == 0);
 408     while (i++)
 409         put_bits(&s->pb, 8, 'N');
 410
 411     flush_put_bits(&s->pb);
 412
 413     if (frame->pts != AV_NOPTS_VALUE)
 414         avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
 415
 416     avpkt->size = avctx->block_align;
 417     *got_packet_ptr = 1;
 418     return 0;
 419 }
 420
 421 AVCodec ff_wmav1_encoder = {
 422     .name           = "wmav1",
 423     .type           = AVMEDIA_TYPE_AUDIO,
 424     .id             = AV_CODEC_ID_WMAV1,
 425     .priv_data_size = sizeof(WMACodecContext),
 426     .init           = encode_init,
 427     .encode2        = encode_superframe,
 428     .close          = ff_wma_end,
 429     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
 430                                                      AV_SAMPLE_FMT_NONE },
 431     .long_name      = NULL_IF_CONFIG_SMALL("Windows Media Audio 1"),
 432 };
 433
 434 AVCodec ff_wmav2_encoder = {
 435     .name           = "wmav2",
 436     .type           = AVMEDIA_TYPE_AUDIO,
 437     .id             = AV_CODEC_ID_WMAV2,
 438     .priv_data_size = sizeof(WMACodecContext),
 439     .init           = encode_init,
 440     .encode2        = encode_superframe,
 441     .close          = ff_wma_end,
 442     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
 443                                                      AV_SAMPLE_FMT_NONE },
 444     .long_name      = NULL_IF_CONFIG_SMALL("Windows Media Audio 2"),
 445 };