libavcodec/wmaenc.c

   1 /*
   2  * WMA compatible encoder
   3  * Copyright (c) 2007 Michael Niedermayer
   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 "avcodec.h"
  23 #include "wma.h"
  24
  25 #undef NDEBUG
  26 #include <assert.h>
  27
  28
  29 static int encode_init(AVCodecContext * avctx){
  30     WMACodecContext *s = avctx->priv_data;
  31     int i, flags1, flags2;
  32     uint8_t *extradata;
  33
  34     s->avctx = avctx;
  35
  36     if(avctx->channels > MAX_CHANNELS)
  37         return -1;
  38
  39     if(avctx->bit_rate < 24*1000)
  40         return -1;
  41
  42     /* extract flag infos */
  43     flags1 = 0;
  44     flags2 = 1;
  45     if (avctx->codec->id == CODEC_ID_WMAV1) {
  46         extradata= av_malloc(4);
  47         avctx->extradata_size= 4;
  48         extradata[0] = flags1;
  49         extradata[1] = flags1>>8;
  50         extradata[2] = flags2;
  51         extradata[3] = flags2>>8;
  52     } else if (avctx->codec->id == CODEC_ID_WMAV2) {
  53         extradata= av_mallocz(10);
  54         avctx->extradata_size= 10;
  55         extradata[0] = flags1;
  56         extradata[1] = flags1>>8;
  57         extradata[2] = flags1>>16;
  58         extradata[3] = flags1>>24;
  59         extradata[4] = flags2;
  60         extradata[5] = flags2>>8;
  61     }else
  62         assert(0);
  63     avctx->extradata= extradata;
  64     s->use_exp_vlc = flags2 & 0x0001;
  65     s->use_bit_reservoir = flags2 & 0x0002;
  66     s->use_variable_block_len = flags2 & 0x0004;
  67
  68     ff_wma_init(avctx, flags2);
  69
  70     /* init MDCT */
  71     for(i = 0; i < s->nb_block_sizes; i++)
  72         ff_mdct_init(&s->mdct_ctx[i], s->frame_len_bits - i + 1, 0);
  73
  74     avctx->block_align=
  75     s->block_align= avctx->bit_rate*(int64_t)s->frame_len / (avctx->sample_rate*8);
  76 //av_log(NULL, AV_LOG_ERROR, "%d %d %d %d\n", s->block_align, avctx->bit_rate, s->frame_len, avctx->sample_rate);
  77     avctx->frame_size= s->frame_len;
  78
  79     return 0;
  80 }
  81
  82
  83 static void apply_window_and_mdct(AVCodecContext * avctx, signed short * audio, int len) {
  84     WMACodecContext *s = avctx->priv_data;
  85     int window_index= s->frame_len_bits - s->block_len_bits;
  86     int i, j, channel;
  87     const float * win = s->windows[window_index];
  88     int window_len = 1 << s->block_len_bits;
  89     float n = window_len/2;
  90
  91     for (channel = 0; channel < avctx->channels; channel++) {
  92         memcpy(s->output, s->frame_out[channel], sizeof(float)*window_len);
  93         j = channel;
  94         for (i = 0; i < len; i++, j += avctx->channels){
  95             s->output[i+window_len]  = audio[j] / n * win[window_len - i - 1];
  96             s->frame_out[channel][i] = audio[j] / n * win[i];
  97         }
  98         ff_mdct_calc(&s->mdct_ctx[window_index], s->coefs[channel], s->output, s->mdct_tmp);
  99     }
 100 }
 101
 102 //FIXME use for decoding too
 103 static void init_exp(WMACodecContext *s, int ch, int *exp_param){
 104     int n;
 105     const uint16_t *ptr;
 106     float v, *q, max_scale, *q_end;
 107
 108     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 109     q = s->exponents[ch];
 110     q_end = q + s->block_len;
 111     max_scale = 0;
 112     while (q < q_end) {
 113         /* XXX: use a table */
 114         v = pow(10, *exp_param++ * (1.0 / 16.0));
 115         max_scale= FFMAX(max_scale, v);
 116         n = *ptr++;
 117         do {
 118             *q++ = v;
 119         } while (--n);
 120     }
 121     s->max_exponent[ch] = max_scale;
 122 }
 123
 124 static void encode_exp_vlc(WMACodecContext *s, int ch, const int *exp_param){
 125     int last_exp;
 126     const uint16_t *ptr;
 127     float *q, *q_end;
 128
 129     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 130     q = s->exponents[ch];
 131     q_end = q + s->block_len;
 132     if (s->version == 1) {
 133         last_exp= *exp_param++;
 134         assert(last_exp-10 >= 0 && last_exp-10 < 32);
 135         put_bits(&s->pb, 5, last_exp - 10);
 136         q+= *ptr++;
 137     }else
 138         last_exp = 36;
 139     while (q < q_end) {
 140         int exp = *exp_param++;
 141         int code = exp - last_exp + 60;
 142         assert(code >= 0 && code < 120);
 143         put_bits(&s->pb, ff_wma_scale_huffbits[code], ff_wma_scale_huffcodes[code]);
 144         /* XXX: use a table */
 145         q+= *ptr++;
 146         last_exp= exp;
 147     }
 148 }
 149
 150 static int encode_block(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], int total_gain){
 151     int v, bsize, ch, coef_nb_bits, parse_exponents;
 152     float mdct_norm;
 153     int nb_coefs[MAX_CHANNELS];
 154     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};
 155
 156     //FIXME remove duplication relative to decoder
 157     if (s->use_variable_block_len) {
 158         assert(0); //FIXME not implemented
 159     }else{
 160         /* fixed block len */
 161         s->next_block_len_bits = s->frame_len_bits;
 162         s->prev_block_len_bits = s->frame_len_bits;
 163         s->block_len_bits = s->frame_len_bits;
 164     }
 165
 166     s->block_len = 1 << s->block_len_bits;
 167 //     assert((s->block_pos + s->block_len) <= s->frame_len);
 168     bsize = s->frame_len_bits - s->block_len_bits;
 169
 170     //FIXME factor
 171     v = s->coefs_end[bsize] - s->coefs_start;
 172     for(ch = 0; ch < s->nb_channels; ch++)
 173         nb_coefs[ch] = v;
 174     {
 175         int n4 = s->block_len / 2;
 176         mdct_norm = 1.0 / (float)n4;
 177         if (s->version == 1) {
 178             mdct_norm *= sqrt(n4);
 179         }
 180     }
 181
 182     if (s->nb_channels == 2) {
 183         put_bits(&s->pb, 1, s->ms_stereo= 1);
 184     }
 185
 186     for(ch = 0; ch < s->nb_channels; ch++) {
 187         if (s->channel_coded[ch]= 1) { //FIXME
 188             init_exp(s, ch, fixed_exp);
 189         }
 190     }
 191
 192     for(ch = 0; ch < s->nb_channels; ch++) {
 193         if (s->channel_coded[ch]) {
 194             int16_t *coefs1;
 195             float *coefs, *exponents, mult;
 196             int i, n;
 197
 198             coefs1 = s->coefs1[ch];
 199             exponents = s->exponents[ch];
 200             mult = pow(10, total_gain * 0.05) / s->max_exponent[ch];
 201             mult *= mdct_norm;
 202             coefs = src_coefs[ch];
 203             if (s->use_noise_coding && 0) {
 204                 assert(0); //FIXME not implemented
 205             } else {
 206                 coefs += s->coefs_start;
 207                 n = nb_coefs[ch];
 208                 for(i = 0;i < n; i++){
 209                     double t= *coefs++ / (exponents[i] * mult);
 210                     if(t<-32768 || t>32767)
 211                         return -1;
 212
 213                     coefs1[i] = lrint(t);
 214                 }
 215             }
 216         }
 217     }
 218
 219     v = 0;
 220     for(ch = 0; ch < s->nb_channels; ch++) {
 221         int a = s->channel_coded[ch];
 222         put_bits(&s->pb, 1, a);
 223         v |= a;
 224     }
 225
 226     if (!v)
 227         return 1;
 228
 229     for(v= total_gain-1; v>=127; v-= 127)
 230         put_bits(&s->pb, 7, 127);
 231     put_bits(&s->pb, 7, v);
 232
 233     coef_nb_bits= ff_wma_total_gain_to_bits(total_gain);
 234
 235     if (s->use_noise_coding) {
 236         for(ch = 0; ch < s->nb_channels; ch++) {
 237             if (s->channel_coded[ch]) {
 238                 int i, n;
 239                 n = s->exponent_high_sizes[bsize];
 240                 for(i=0;i<n;i++) {
 241                     put_bits(&s->pb, 1, s->high_band_coded[ch][i]= 0);
 242                     if (0)
 243                         nb_coefs[ch] -= s->exponent_high_bands[bsize][i];
 244                 }
 245             }
 246         }
 247     }
 248
 249     parse_exponents = 1;
 250     if (s->block_len_bits != s->frame_len_bits) {
 251         put_bits(&s->pb, 1, parse_exponents);
 252     }
 253
 254     if (parse_exponents) {
 255         for(ch = 0; ch < s->nb_channels; ch++) {
 256             if (s->channel_coded[ch]) {
 257                 if (s->use_exp_vlc) {
 258                     encode_exp_vlc(s, ch, fixed_exp);
 259                 } else {
 260                     assert(0); //FIXME not implemented
 261 //                    encode_exp_lsp(s, ch);
 262                 }
 263             }
 264         }
 265     } else {
 266         assert(0); //FIXME not implemented
 267     }
 268
 269     for(ch = 0; ch < s->nb_channels; ch++) {
 270         if (s->channel_coded[ch]) {
 271             int run, tindex;
 272             int16_t *ptr, *eptr;
 273             tindex = (ch == 1 && s->ms_stereo);
 274             ptr = &s->coefs1[ch][0];
 275             eptr = ptr + nb_coefs[ch];
 276
 277             run=0;
 278             for(;ptr < eptr; ptr++){
 279                 if(*ptr){
 280                     int level= *ptr;
 281                     int abs_level= FFABS(level);
 282                     int code= 0;
 283                     if(abs_level <= s->coef_vlcs[tindex]->max_level){
 284                         if(run < s->coef_vlcs[tindex]->levels[abs_level-1])
 285                             code= run + s->int_table[tindex][abs_level-1];
 286                     }
 287
 288                     assert(code < s->coef_vlcs[tindex]->n);
 289                     put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[code], s->coef_vlcs[tindex]->huffcodes[code]);
 290
 291                     if(code == 0){
 292                         if(1<<coef_nb_bits <= abs_level)
 293                             return -1;
 294
 295                         put_bits(&s->pb, coef_nb_bits, abs_level);
 296                         put_bits(&s->pb, s->frame_len_bits, run);
 297                     }
 298                     put_bits(&s->pb, 1, level < 0); //FIXME the sign is fliped somewhere
 299                     run=0;
 300                 }else{
 301                     run++;
 302                 }
 303             }
 304             if(run)
 305                 put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[1], s->coef_vlcs[tindex]->huffcodes[1]);
 306         }
 307         if (s->version == 1 && s->nb_channels >= 2) {
 308             align_put_bits(&s->pb);
 309         }
 310     }
 311     return 0;
 312 }
 313
 314 static int encode_frame(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], uint8_t *buf, int buf_size, int total_gain){
 315     init_put_bits(&s->pb, buf, buf_size);
 316
 317     if (s->use_bit_reservoir) {
 318         assert(0);//FIXME not implemented
 319     }else{
 320         if(encode_block(s, src_coefs, total_gain) < 0)
 321             return INT_MAX;
 322     }
 323
 324     align_put_bits(&s->pb);
 325
 326     return put_bits_count(&s->pb)/8 - s->block_align;
 327 }
 328
 329 static int encode_superframe(AVCodecContext *avctx,
 330                             unsigned char *buf, int buf_size, void *data){
 331     WMACodecContext *s = avctx->priv_data;
 332     short *samples = data;
 333     int i, total_gain, best;
 334
 335     s->block_len_bits= s->frame_len_bits; //required by non variable block len
 336     s->block_len = 1 << s->block_len_bits;
 337
 338     apply_window_and_mdct(avctx, samples, avctx->frame_size);
 339
 340     if (s->ms_stereo) {
 341         float a, b;
 342         int i;
 343
 344         for(i = 0; i < s->block_len; i++) {
 345             a = s->coefs[0][i]*0.5;
 346             b = s->coefs[1][i]*0.5;
 347             s->coefs[0][i] = a + b;
 348             s->coefs[1][i] = a - b;
 349         }
 350     }
 351
 352 #if 1
 353     total_gain= 128;
 354     for(i=64; i; i>>=1){
 355         int error= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
 356         if(error<0)
 357             total_gain-= i;
 358     }
 359 #else
 360     total_gain= 90;
 361     best= encode_frame(s, s->coefs, buf, buf_size, total_gain);
 362     for(i=32; i; i>>=1){
 363         int scoreL= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
 364         int scoreR= encode_frame(s, s->coefs, buf, buf_size, total_gain+i);
 365         av_log(NULL, AV_LOG_ERROR, "%d %d %d (%d)\n", scoreL, best, scoreR, total_gain);
 366         if(scoreL < FFMIN(best, scoreR)){
 367             best = scoreL;
 368             total_gain -= i;
 369         }else if(scoreR < best){
 370             best = scoreR;
 371             total_gain += i;
 372         }
 373     }
 374 #endif
 375
 376     encode_frame(s, s->coefs, buf, buf_size, total_gain);
 377     assert((put_bits_count(&s->pb) & 7) == 0);
 378     i= s->block_align - (put_bits_count(&s->pb)+7)/8;
 379     assert(i>=0);
 380     while(i--)
 381         put_bits(&s->pb, 8, 'N');
 382
 383     flush_put_bits(&s->pb);
 384     return pbBufPtr(&s->pb) - s->pb.buf;
 385 }
 386
 387 AVCodec wmav1_encoder =
 388 {
 389     "wmav1",
 390     CODEC_TYPE_AUDIO,
 391     CODEC_ID_WMAV1,
 392     sizeof(WMACodecContext),
 393     encode_init,
 394     encode_superframe,
 395     ff_wma_end,
 396 };
 397
 398 AVCodec wmav2_encoder =
 399 {
 400     "wmav2",
 401     CODEC_TYPE_AUDIO,
 402     CODEC_ID_WMAV2,
 403     sizeof(WMACodecContext),
 404     encode_init,
 405     encode_superframe,
 406     ff_wma_end,
 407 };