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         AV_WL16(extradata, flags1);
  49         AV_WL16(extradata+2, flags2);
  50     } else if (avctx->codec->id == CODEC_ID_WMAV2) {
  51         extradata= av_mallocz(10);
  52         avctx->extradata_size= 10;
  53         AV_WL32(extradata, flags1);
  54         AV_WL16(extradata+4, flags2);
  55     }else
  56         assert(0);
  57     avctx->extradata= extradata;
  58     s->use_exp_vlc = flags2 & 0x0001;
  59     s->use_bit_reservoir = flags2 & 0x0002;
  60     s->use_variable_block_len = flags2 & 0x0004;
  61
  62     ff_wma_init(avctx, flags2);
  63
  64     /* init MDCT */
  65     for(i = 0; i < s->nb_block_sizes; i++)
  66         ff_mdct_init(&s->mdct_ctx[i], s->frame_len_bits - i + 1, 0);
  67
  68     avctx->block_align=
  69     s->block_align= avctx->bit_rate*(int64_t)s->frame_len / (avctx->sample_rate*8);
  70 //av_log(NULL, AV_LOG_ERROR, "%d %d %d %d\n", s->block_align, avctx->bit_rate, s->frame_len, avctx->sample_rate);
  71     avctx->frame_size= s->frame_len;
  72
  73     return 0;
  74 }
  75
  76
  77 static void apply_window_and_mdct(AVCodecContext * avctx, signed short * audio, int len) {
  78     WMACodecContext *s = avctx->priv_data;
  79     int window_index= s->frame_len_bits - s->block_len_bits;
  80     int i, j, channel;
  81     const float * win = s->windows[window_index];
  82     int window_len = 1 << s->block_len_bits;
  83     float n = window_len/2;
  84
  85     for (channel = 0; channel < avctx->channels; channel++) {
  86         memcpy(s->output, s->frame_out[channel], sizeof(float)*window_len);
  87         j = channel;
  88         for (i = 0; i < len; i++, j += avctx->channels){
  89             s->output[i+window_len]  = audio[j] / n * win[window_len - i - 1];
  90             s->frame_out[channel][i] = audio[j] / n * win[i];
  91         }
  92         ff_mdct_calc(&s->mdct_ctx[window_index], s->coefs[channel], s->output, s->mdct_tmp);
  93     }
  94 }
  95
  96 //FIXME use for decoding too
  97 static void init_exp(WMACodecContext *s, int ch, const int *exp_param){
  98     int n;
  99     const uint16_t *ptr;
 100     float v, *q, max_scale, *q_end;
 101
 102     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 103     q = s->exponents[ch];
 104     q_end = q + s->block_len;
 105     max_scale = 0;
 106     while (q < q_end) {
 107         /* XXX: use a table */
 108         v = pow(10, *exp_param++ * (1.0 / 16.0));
 109         max_scale= FFMAX(max_scale, v);
 110         n = *ptr++;
 111         do {
 112             *q++ = v;
 113         } while (--n);
 114     }
 115     s->max_exponent[ch] = max_scale;
 116 }
 117
 118 static void encode_exp_vlc(WMACodecContext *s, int ch, const int *exp_param){
 119     int last_exp;
 120     const uint16_t *ptr;
 121     float *q, *q_end;
 122
 123     ptr = s->exponent_bands[s->frame_len_bits - s->block_len_bits];
 124     q = s->exponents[ch];
 125     q_end = q + s->block_len;
 126     if (s->version == 1) {
 127         last_exp= *exp_param++;
 128         assert(last_exp-10 >= 0 && last_exp-10 < 32);
 129         put_bits(&s->pb, 5, last_exp - 10);
 130         q+= *ptr++;
 131     }else
 132         last_exp = 36;
 133     while (q < q_end) {
 134         int exp = *exp_param++;
 135         int code = exp - last_exp + 60;
 136         assert(code >= 0 && code < 120);
 137         put_bits(&s->pb, ff_wma_scale_huffbits[code], ff_wma_scale_huffcodes[code]);
 138         /* XXX: use a table */
 139         q+= *ptr++;
 140         last_exp= exp;
 141     }
 142 }
 143
 144 static int encode_block(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], int total_gain){
 145     int v, bsize, ch, coef_nb_bits, parse_exponents;
 146     float mdct_norm;
 147     int nb_coefs[MAX_CHANNELS];
 148     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};
 149
 150     //FIXME remove duplication relative to decoder
 151     if (s->use_variable_block_len) {
 152         assert(0); //FIXME not implemented
 153     }else{
 154         /* fixed block len */
 155         s->next_block_len_bits = s->frame_len_bits;
 156         s->prev_block_len_bits = s->frame_len_bits;
 157         s->block_len_bits = s->frame_len_bits;
 158     }
 159
 160     s->block_len = 1 << s->block_len_bits;
 161 //     assert((s->block_pos + s->block_len) <= s->frame_len);
 162     bsize = s->frame_len_bits - s->block_len_bits;
 163
 164     //FIXME factor
 165     v = s->coefs_end[bsize] - s->coefs_start;
 166     for(ch = 0; ch < s->nb_channels; ch++)
 167         nb_coefs[ch] = v;
 168     {
 169         int n4 = s->block_len / 2;
 170         mdct_norm = 1.0 / (float)n4;
 171         if (s->version == 1) {
 172             mdct_norm *= sqrt(n4);
 173         }
 174     }
 175
 176     if (s->nb_channels == 2) {
 177         put_bits(&s->pb, 1, s->ms_stereo= 1);
 178     }
 179
 180     for(ch = 0; ch < s->nb_channels; ch++) {
 181         if ((s->channel_coded[ch]= 1)) { //FIXME only set channel_coded when needed, instead of always
 182             init_exp(s, ch, fixed_exp);
 183         }
 184     }
 185
 186     for(ch = 0; ch < s->nb_channels; ch++) {
 187         if (s->channel_coded[ch]) {
 188             int16_t *coefs1;
 189             float *coefs, *exponents, mult;
 190             int i, n;
 191
 192             coefs1 = s->coefs1[ch];
 193             exponents = s->exponents[ch];
 194             mult = pow(10, total_gain * 0.05) / s->max_exponent[ch];
 195             mult *= mdct_norm;
 196             coefs = src_coefs[ch];
 197             if (s->use_noise_coding && 0) {
 198                 assert(0); //FIXME not implemented
 199             } else {
 200                 coefs += s->coefs_start;
 201                 n = nb_coefs[ch];
 202                 for(i = 0;i < n; i++){
 203                     double t= *coefs++ / (exponents[i] * mult);
 204                     if(t<-32768 || t>32767)
 205                         return -1;
 206
 207                     coefs1[i] = lrint(t);
 208                 }
 209             }
 210         }
 211     }
 212
 213     v = 0;
 214     for(ch = 0; ch < s->nb_channels; ch++) {
 215         int a = s->channel_coded[ch];
 216         put_bits(&s->pb, 1, a);
 217         v |= a;
 218     }
 219
 220     if (!v)
 221         return 1;
 222
 223     for(v= total_gain-1; v>=127; v-= 127)
 224         put_bits(&s->pb, 7, 127);
 225     put_bits(&s->pb, 7, v);
 226
 227     coef_nb_bits= ff_wma_total_gain_to_bits(total_gain);
 228
 229     if (s->use_noise_coding) {
 230         for(ch = 0; ch < s->nb_channels; ch++) {
 231             if (s->channel_coded[ch]) {
 232                 int i, n;
 233                 n = s->exponent_high_sizes[bsize];
 234                 for(i=0;i<n;i++) {
 235                     put_bits(&s->pb, 1, s->high_band_coded[ch][i]= 0);
 236                     if (0)
 237                         nb_coefs[ch] -= s->exponent_high_bands[bsize][i];
 238                 }
 239             }
 240         }
 241     }
 242
 243     parse_exponents = 1;
 244     if (s->block_len_bits != s->frame_len_bits) {
 245         put_bits(&s->pb, 1, parse_exponents);
 246     }
 247
 248     if (parse_exponents) {
 249         for(ch = 0; ch < s->nb_channels; ch++) {
 250             if (s->channel_coded[ch]) {
 251                 if (s->use_exp_vlc) {
 252                     encode_exp_vlc(s, ch, fixed_exp);
 253                 } else {
 254                     assert(0); //FIXME not implemented
 255 //                    encode_exp_lsp(s, ch);
 256                 }
 257             }
 258         }
 259     } else {
 260         assert(0); //FIXME not implemented
 261     }
 262
 263     for(ch = 0; ch < s->nb_channels; ch++) {
 264         if (s->channel_coded[ch]) {
 265             int run, tindex;
 266             int16_t *ptr, *eptr;
 267             tindex = (ch == 1 && s->ms_stereo);
 268             ptr = &s->coefs1[ch][0];
 269             eptr = ptr + nb_coefs[ch];
 270
 271             run=0;
 272             for(;ptr < eptr; ptr++){
 273                 if(*ptr){
 274                     int level= *ptr;
 275                     int abs_level= FFABS(level);
 276                     int code= 0;
 277                     if(abs_level <= s->coef_vlcs[tindex]->max_level){
 278                         if(run < s->coef_vlcs[tindex]->levels[abs_level-1])
 279                             code= run + s->int_table[tindex][abs_level-1];
 280                     }
 281
 282                     assert(code < s->coef_vlcs[tindex]->n);
 283                     put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[code], s->coef_vlcs[tindex]->huffcodes[code]);
 284
 285                     if(code == 0){
 286                         if(1<<coef_nb_bits <= abs_level)
 287                             return -1;
 288
 289                         put_bits(&s->pb, coef_nb_bits, abs_level);
 290                         put_bits(&s->pb, s->frame_len_bits, run);
 291                     }
 292                     put_bits(&s->pb, 1, level < 0); //FIXME the sign is fliped somewhere
 293                     run=0;
 294                 }else{
 295                     run++;
 296                 }
 297             }
 298             if(run)
 299                 put_bits(&s->pb, s->coef_vlcs[tindex]->huffbits[1], s->coef_vlcs[tindex]->huffcodes[1]);
 300         }
 301         if (s->version == 1 && s->nb_channels >= 2) {
 302             align_put_bits(&s->pb);
 303         }
 304     }
 305     return 0;
 306 }
 307
 308 static int encode_frame(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE], uint8_t *buf, int buf_size, int total_gain){
 309     init_put_bits(&s->pb, buf, buf_size);
 310
 311     if (s->use_bit_reservoir) {
 312         assert(0);//FIXME not implemented
 313     }else{
 314         if(encode_block(s, src_coefs, total_gain) < 0)
 315             return INT_MAX;
 316     }
 317
 318     align_put_bits(&s->pb);
 319
 320     return put_bits_count(&s->pb)/8 - s->block_align;
 321 }
 322
 323 static int encode_superframe(AVCodecContext *avctx,
 324                             unsigned char *buf, int buf_size, void *data){
 325     WMACodecContext *s = avctx->priv_data;
 326     short *samples = data;
 327     int i, total_gain;
 328
 329     s->block_len_bits= s->frame_len_bits; //required by non variable block len
 330     s->block_len = 1 << s->block_len_bits;
 331
 332     apply_window_and_mdct(avctx, samples, avctx->frame_size);
 333
 334     if (s->ms_stereo) {
 335         float a, b;
 336         int i;
 337
 338         for(i = 0; i < s->block_len; i++) {
 339             a = s->coefs[0][i]*0.5;
 340             b = s->coefs[1][i]*0.5;
 341             s->coefs[0][i] = a + b;
 342             s->coefs[1][i] = a - b;
 343         }
 344     }
 345
 346 #if 1
 347     total_gain= 128;
 348     for(i=64; i; i>>=1){
 349         int error= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
 350         if(error<0)
 351             total_gain-= i;
 352     }
 353 #else
 354     total_gain= 90;
 355     best= encode_frame(s, s->coefs, buf, buf_size, total_gain);
 356     for(i=32; i; i>>=1){
 357         int scoreL= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
 358         int scoreR= encode_frame(s, s->coefs, buf, buf_size, total_gain+i);
 359         av_log(NULL, AV_LOG_ERROR, "%d %d %d (%d)\n", scoreL, best, scoreR, total_gain);
 360         if(scoreL < FFMIN(best, scoreR)){
 361             best = scoreL;
 362             total_gain -= i;
 363         }else if(scoreR < best){
 364             best = scoreR;
 365             total_gain += i;
 366         }
 367     }
 368 #endif
 369
 370     encode_frame(s, s->coefs, buf, buf_size, total_gain);
 371     assert((put_bits_count(&s->pb) & 7) == 0);
 372     i= s->block_align - (put_bits_count(&s->pb)+7)/8;
 373     assert(i>=0);
 374     while(i--)
 375         put_bits(&s->pb, 8, 'N');
 376
 377     flush_put_bits(&s->pb);
 378     return pbBufPtr(&s->pb) - s->pb.buf;
 379 }
 380
 381 AVCodec wmav1_encoder =
 382 {
 383     "wmav1",
 384     CODEC_TYPE_AUDIO,
 385     CODEC_ID_WMAV1,
 386     sizeof(WMACodecContext),
 387     encode_init,
 388     encode_superframe,
 389     ff_wma_end,
 390     .long_name = NULL_IF_CONFIG_SMALL("Windows Media Audio 1"),
 391 };
 392
 393 AVCodec wmav2_encoder =
 394 {
 395     "wmav2",
 396     CODEC_TYPE_AUDIO,
 397     CODEC_ID_WMAV2,
 398     sizeof(WMACodecContext),
 399     encode_init,
 400     encode_superframe,
 401     ff_wma_end,
 402     .long_name = NULL_IF_CONFIG_SMALL("Windows Media Audio 2"),
 403 };