libavcodec/wma.c

   1 /*
   2  * WMA compatible codec
   3  * Copyright (c) 2002-2007 The FFmpeg Project
   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 #include "wmadata.h"
  25
  26 #undef NDEBUG
  27 #include <assert.h>
  28
  29 /* XXX: use same run/length optimization as mpeg decoders */
  30 //FIXME maybe split decode / encode or pass flag
  31 static void init_coef_vlc(VLC *vlc, uint16_t **prun_table,
  32                           uint16_t **plevel_table, uint16_t **pint_table,
  33                           const CoefVLCTable *vlc_table)
  34 {
  35     int n = vlc_table->n;
  36     const uint8_t  *table_bits   = vlc_table->huffbits;
  37     const uint32_t *table_codes  = vlc_table->huffcodes;
  38     const uint16_t *levels_table = vlc_table->levels;
  39     uint16_t *run_table, *level_table, *int_table;
  40     int i, l, j, k, level;
  41
  42     init_vlc(vlc, VLCBITS, n, table_bits, 1, 1, table_codes, 4, 4, 0);
  43
  44     run_table   = av_malloc(n * sizeof(uint16_t));
  45     level_table = av_malloc(n * sizeof(uint16_t));
  46     int_table   = av_malloc(n * sizeof(uint16_t));
  47     i = 2;
  48     level = 1;
  49     k = 0;
  50     while (i < n) {
  51         int_table[k] = i;
  52         l = levels_table[k++];
  53         for (j = 0; j < l; j++) {
  54             run_table[i]   = j;
  55             level_table[i] = level;
  56             i++;
  57         }
  58         level++;
  59     }
  60     *prun_table   = run_table;
  61     *plevel_table = level_table;
  62     *pint_table   = int_table;
  63 }
  64
  65 /**
  66  *@brief Get the samples per frame for this stream.
  67  *@param sample_rate output sample_rate
  68  *@param version wma version
  69  *@param decode_flags codec compression features
  70  *@return log2 of the number of output samples per frame
  71  */
  72 int av_cold ff_wma_get_frame_len_bits(int sample_rate, int version,
  73                                       unsigned int decode_flags)
  74 {
  75
  76     int frame_len_bits;
  77
  78     if (sample_rate <= 16000) {
  79         frame_len_bits = 9;
  80     } else if (sample_rate <= 22050 ||
  81              (sample_rate <= 32000 && version == 1)) {
  82         frame_len_bits = 10;
  83     } else if (sample_rate <= 48000) {
  84         frame_len_bits = 11;
  85     } else if (sample_rate <= 96000) {
  86         frame_len_bits = 12;
  87     } else {
  88         frame_len_bits = 13;
  89     }
  90
  91     if (version == 3) {
  92         int tmp = decode_flags & 0x6;
  93         if (tmp == 0x2) {
  94             ++frame_len_bits;
  95         } else if (tmp == 0x4) {
  96             --frame_len_bits;
  97         } else if (tmp == 0x6) {
  98             frame_len_bits -= 2;
  99         }
 100     }
 101
 102     return frame_len_bits;
 103 }
 104
 105 int ff_wma_init(AVCodecContext *avctx, int flags2)
 106 {
 107     WMACodecContext *s = avctx->priv_data;
 108     int i;
 109     float bps1, high_freq;
 110     volatile float bps;
 111     int sample_rate1;
 112     int coef_vlc_table;
 113
 114     if (   avctx->sample_rate <= 0 || avctx->sample_rate > 50000
 115         || avctx->channels    <= 0 || avctx->channels    > 8
 116         || avctx->bit_rate    <= 0)
 117         return -1;
 118
 119     s->sample_rate = avctx->sample_rate;
 120     s->nb_channels = avctx->channels;
 121     s->bit_rate    = avctx->bit_rate;
 122     s->block_align = avctx->block_align;
 123
 124     dsputil_init(&s->dsp, avctx);
 125
 126     if (avctx->codec->id == CODEC_ID_WMAV1) {
 127         s->version = 1;
 128     } else {
 129         s->version = 2;
 130     }
 131
 132     /* compute MDCT block size */
 133     s->frame_len_bits = ff_wma_get_frame_len_bits(s->sample_rate, s->version, 0);
 134
 135     s->frame_len = 1 << s->frame_len_bits;
 136     if (s->use_variable_block_len) {
 137         int nb_max, nb;
 138         nb = ((flags2 >> 3) & 3) + 1;
 139         if ((s->bit_rate / s->nb_channels) >= 32000)
 140             nb += 2;
 141         nb_max = s->frame_len_bits - BLOCK_MIN_BITS;
 142         if (nb > nb_max)
 143             nb = nb_max;
 144         s->nb_block_sizes = nb + 1;
 145     } else {
 146         s->nb_block_sizes = 1;
 147     }
 148
 149     /* init rate dependent parameters */
 150     s->use_noise_coding = 1;
 151     high_freq = s->sample_rate * 0.5;
 152
 153     /* if version 2, then the rates are normalized */
 154     sample_rate1 = s->sample_rate;
 155     if (s->version == 2) {
 156         if (sample_rate1 >= 44100) {
 157             sample_rate1 = 44100;
 158         } else if (sample_rate1 >= 22050) {
 159             sample_rate1 = 22050;
 160         } else if (sample_rate1 >= 16000) {
 161             sample_rate1 = 16000;
 162         } else if (sample_rate1 >= 11025) {
 163             sample_rate1 = 11025;
 164         } else if (sample_rate1 >= 8000) {
 165             sample_rate1 = 8000;
 166         }
 167     }
 168
 169     bps = (float)s->bit_rate / (float)(s->nb_channels * s->sample_rate);
 170     s->byte_offset_bits = av_log2((int)(bps * s->frame_len / 8.0 + 0.5)) + 2;
 171
 172     /* compute high frequency value and choose if noise coding should
 173        be activated */
 174     bps1 = bps;
 175     if (s->nb_channels == 2)
 176         bps1 = bps * 1.6;
 177     if (sample_rate1 == 44100) {
 178         if (bps1 >= 0.61) {
 179             s->use_noise_coding = 0;
 180         } else {
 181             high_freq = high_freq * 0.4;
 182         }
 183     } else if (sample_rate1 == 22050) {
 184         if (bps1 >= 1.16) {
 185             s->use_noise_coding = 0;
 186         } else if (bps1 >= 0.72) {
 187             high_freq = high_freq * 0.7;
 188         } else {
 189             high_freq = high_freq * 0.6;
 190         }
 191     } else if (sample_rate1 == 16000) {
 192         if (bps > 0.5) {
 193             high_freq = high_freq * 0.5;
 194         } else {
 195             high_freq = high_freq * 0.3;
 196         }
 197     } else if (sample_rate1 == 11025) {
 198         high_freq = high_freq * 0.7;
 199     } else if (sample_rate1 == 8000) {
 200         if (bps <= 0.625) {
 201             high_freq = high_freq * 0.5;
 202         } else if (bps > 0.75) {
 203             s->use_noise_coding = 0;
 204         } else {
 205             high_freq = high_freq * 0.65;
 206         }
 207     } else {
 208         if (bps >= 0.8) {
 209             high_freq = high_freq * 0.75;
 210         } else if (bps >= 0.6) {
 211             high_freq = high_freq * 0.6;
 212         } else {
 213             high_freq = high_freq * 0.5;
 214         }
 215     }
 216     dprintf(s->avctx, "flags2=0x%x\n", flags2);
 217     dprintf(s->avctx, "version=%d channels=%d sample_rate=%d bitrate=%d block_align=%d\n",
 218             s->version, s->nb_channels, s->sample_rate, s->bit_rate,
 219             s->block_align);
 220     dprintf(s->avctx, "bps=%f bps1=%f high_freq=%f bitoffset=%d\n",
 221             bps, bps1, high_freq, s->byte_offset_bits);
 222     dprintf(s->avctx, "use_noise_coding=%d use_exp_vlc=%d nb_block_sizes=%d\n",
 223             s->use_noise_coding, s->use_exp_vlc, s->nb_block_sizes);
 224
 225     /* compute the scale factor band sizes for each MDCT block size */
 226     {
 227         int a, b, pos, lpos, k, block_len, i, j, n;
 228         const uint8_t *table;
 229
 230         if (s->version == 1) {
 231             s->coefs_start = 3;
 232         } else {
 233             s->coefs_start = 0;
 234         }
 235         for (k = 0; k < s->nb_block_sizes; k++) {
 236             block_len = s->frame_len >> k;
 237
 238             if (s->version == 1) {
 239                 lpos = 0;
 240                 for (i = 0; i < 25; i++) {
 241                     a = wma_critical_freqs[i];
 242                     b = s->sample_rate;
 243                     pos = ((block_len * 2 * a) + (b >> 1)) / b;
 244                     if (pos > block_len)
 245                         pos = block_len;
 246                     s->exponent_bands[0][i] = pos - lpos;
 247                     if (pos >= block_len) {
 248                         i++;
 249                         break;
 250                     }
 251                     lpos = pos;
 252                 }
 253                 s->exponent_sizes[0] = i;
 254             } else {
 255                 /* hardcoded tables */
 256                 table = NULL;
 257                 a = s->frame_len_bits - BLOCK_MIN_BITS - k;
 258                 if (a < 3) {
 259                     if (s->sample_rate >= 44100) {
 260                         table = exponent_band_44100[a];
 261                     } else if (s->sample_rate >= 32000) {
 262                         table = exponent_band_32000[a];
 263                     } else if (s->sample_rate >= 22050) {
 264                         table = exponent_band_22050[a];
 265                     }
 266                 }
 267                 if (table) {
 268                     n = *table++;
 269                     for (i = 0; i < n; i++)
 270                         s->exponent_bands[k][i] = table[i];
 271                     s->exponent_sizes[k] = n;
 272                 } else {
 273                     j = 0;
 274                     lpos = 0;
 275                     for (i = 0; i < 25; i++) {
 276                         a = wma_critical_freqs[i];
 277                         b = s->sample_rate;
 278                         pos = ((block_len * 2 * a) + (b << 1)) / (4 * b);
 279                         pos <<= 2;
 280                         if (pos > block_len)
 281                             pos = block_len;
 282                         if (pos > lpos)
 283                             s->exponent_bands[k][j++] = pos - lpos;
 284                         if (pos >= block_len)
 285                             break;
 286                         lpos = pos;
 287                     }
 288                     s->exponent_sizes[k] = j;
 289                 }
 290             }
 291
 292             /* max number of coefs */
 293             s->coefs_end[k] = (s->frame_len - ((s->frame_len * 9) / 100)) >> k;
 294             /* high freq computation */
 295             s->high_band_start[k] = (int)((block_len * 2 * high_freq) /
 296                                           s->sample_rate + 0.5);
 297             n = s->exponent_sizes[k];
 298             j = 0;
 299             pos = 0;
 300             for (i = 0; i < n; i++) {
 301                 int start, end;
 302                 start = pos;
 303                 pos += s->exponent_bands[k][i];
 304                 end = pos;
 305                 if (start < s->high_band_start[k])
 306                     start = s->high_band_start[k];
 307                 if (end > s->coefs_end[k])
 308                     end = s->coefs_end[k];
 309                 if (end > start)
 310                     s->exponent_high_bands[k][j++] = end - start;
 311             }
 312             s->exponent_high_sizes[k] = j;
 313 #if 0
 314             tprintf(s->avctx, "%5d: coefs_end=%d high_band_start=%d nb_high_bands=%d: ",
 315                     s->frame_len >> k,
 316                     s->coefs_end[k],
 317                     s->high_band_start[k],
 318                     s->exponent_high_sizes[k]);
 319             for (j = 0; j < s->exponent_high_sizes[k]; j++)
 320                 tprintf(s->avctx, " %d", s->exponent_high_bands[k][j]);
 321             tprintf(s->avctx, "\n");
 322 #endif
 323         }
 324     }
 325
 326 #ifdef TRACE
 327     {
 328         int i, j;
 329         for (i = 0; i < s->nb_block_sizes; i++) {
 330             tprintf(s->avctx, "%5d: n=%2d:",
 331                     s->frame_len >> i,
 332                     s->exponent_sizes[i]);
 333             for (j = 0; j < s->exponent_sizes[i]; j++)
 334                 tprintf(s->avctx, " %d", s->exponent_bands[i][j]);
 335             tprintf(s->avctx, "\n");
 336         }
 337     }
 338 #endif
 339
 340     /* init MDCT windows : simple sinus window */
 341     for (i = 0; i < s->nb_block_sizes; i++) {
 342         int n;
 343         n = 1 << (s->frame_len_bits - i);
 344         ff_sine_window_init(ff_sine_windows[s->frame_len_bits - i - 7], n);
 345         s->windows[i] = ff_sine_windows[s->frame_len_bits - i - 7];
 346     }
 347
 348     s->reset_block_lengths = 1;
 349
 350     if (s->use_noise_coding) {
 351
 352         /* init the noise generator */
 353         if (s->use_exp_vlc) {
 354             s->noise_mult = 0.02;
 355         } else {
 356             s->noise_mult = 0.04;
 357         }
 358
 359 #ifdef TRACE
 360         for (i = 0; i < NOISE_TAB_SIZE; i++)
 361             s->noise_table[i] = 1.0 * s->noise_mult;
 362 #else
 363         {
 364             unsigned int seed;
 365             float norm;
 366             seed = 1;
 367             norm = (1.0 / (float)(1LL << 31)) * sqrt(3) * s->noise_mult;
 368             for (i = 0; i < NOISE_TAB_SIZE; i++) {
 369                 seed = seed * 314159 + 1;
 370                 s->noise_table[i] = (float)((int)seed) * norm;
 371             }
 372         }
 373 #endif
 374     }
 375
 376     /* choose the VLC tables for the coefficients */
 377     coef_vlc_table = 2;
 378     if (s->sample_rate >= 32000) {
 379         if (bps1 < 0.72) {
 380             coef_vlc_table = 0;
 381         } else if (bps1 < 1.16) {
 382             coef_vlc_table = 1;
 383         }
 384     }
 385     s->coef_vlcs[0]= &coef_vlcs[coef_vlc_table * 2    ];
 386     s->coef_vlcs[1]= &coef_vlcs[coef_vlc_table * 2 + 1];
 387     init_coef_vlc(&s->coef_vlc[0], &s->run_table[0], &s->level_table[0], &s->int_table[0],
 388                   s->coef_vlcs[0]);
 389     init_coef_vlc(&s->coef_vlc[1], &s->run_table[1], &s->level_table[1], &s->int_table[1],
 390                   s->coef_vlcs[1]);
 391
 392     return 0;
 393 }
 394
 395 int ff_wma_total_gain_to_bits(int total_gain)
 396 {
 397          if (total_gain < 15) return 13;
 398     else if (total_gain < 32) return 12;
 399     else if (total_gain < 40) return 11;
 400     else if (total_gain < 45) return 10;
 401     else                      return  9;
 402 }
 403
 404 int ff_wma_end(AVCodecContext *avctx)
 405 {
 406     WMACodecContext *s = avctx->priv_data;
 407     int i;
 408
 409     for (i = 0; i < s->nb_block_sizes; i++)
 410         ff_mdct_end(&s->mdct_ctx[i]);
 411
 412     if (s->use_exp_vlc) {
 413         free_vlc(&s->exp_vlc);
 414     }
 415     if (s->use_noise_coding) {
 416         free_vlc(&s->hgain_vlc);
 417     }
 418     for (i = 0; i < 2; i++) {
 419         free_vlc(&s->coef_vlc[i]);
 420         av_free(s->run_table[i]);
 421         av_free(s->level_table[i]);
 422         av_free(s->int_table[i]);
 423     }
 424
 425     return 0;
 426 }
 427
 428 /**
 429  * Decode an uncompressed coefficient.
 430  * @param s codec context
 431  * @return the decoded coefficient
 432  */
 433 unsigned int ff_wma_get_large_val(GetBitContext* gb)
 434 {
 435     /** consumes up to 34 bits */
 436     int n_bits = 8;
 437     /** decode length */
 438     if (get_bits1(gb)) {
 439         n_bits += 8;
 440         if (get_bits1(gb)) {
 441             n_bits += 8;
 442             if (get_bits1(gb)) {
 443                 n_bits += 7;
 444             }
 445         }
 446     }
 447     return get_bits_long(gb, n_bits);
 448 }
 449
 450 /**
 451  * Decode run level compressed coefficients.
 452  * @param avctx codec context
 453  * @param gb bitstream reader context
 454  * @param vlc vlc table for get_vlc2
 455  * @param level_table level codes
 456  * @param run_table run codes
 457  * @param version 0 for wma1,2 1 for wmapro
 458  * @param ptr output buffer
 459  * @param offset offset in the output buffer
 460  * @param num_coefs number of input coefficents
 461  * @param block_len input buffer length (2^n)
 462  * @param frame_len_bits number of bits for escaped run codes
 463  * @param coef_nb_bits number of bits for escaped level codes
 464  * @return 0 on success, -1 otherwise
 465  */
 466 int ff_wma_run_level_decode(AVCodecContext* avctx, GetBitContext* gb,
 467                             VLC *vlc,
 468                             const uint16_t *level_table, const uint16_t *run_table,
 469                             int version, WMACoef *ptr, int offset,
 470                             int num_coefs, int block_len, int frame_len_bits,
 471                             int coef_nb_bits)
 472 {
 473     int code, level, sign;
 474     const unsigned int coef_mask = block_len - 1;
 475     for (; offset < num_coefs; offset++) {
 476         code = get_vlc2(gb, vlc->table, VLCBITS, VLCMAX);
 477         if (code > 1) {
 478             /** normal code */
 479             offset += run_table[code];
 480             level = level_table[code];
 481         } else if (code == 1) {
 482             /** EOB */
 483             break;
 484         } else {
 485             /** escape */
 486             if (!version) {
 487                 level = get_bits(gb, coef_nb_bits);
 488                 /** NOTE: this is rather suboptimal. reading
 489                     block_len_bits would be better */
 490                 offset += get_bits(gb, frame_len_bits);
 491             } else {
 492                 level = ff_wma_get_large_val(gb);
 493                 /** escape decode */
 494                 if (get_bits1(gb)) {
 495                     if (get_bits1(gb)) {
 496                         if (get_bits1(gb)) {
 497                             av_log(avctx,AV_LOG_ERROR,
 498                                 "broken escape sequence\n");
 499                             return -1;
 500                         } else
 501                             offset += get_bits(gb, frame_len_bits) + 4;
 502                     } else
 503                         offset += get_bits(gb, 2) + 1;
 504                 }
 505             }
 506         }
 507         sign = get_bits1(gb) - 1;
 508         ptr[offset & coef_mask] = (level^sign) - sign;
 509     }
 510     /** NOTE: EOB can be omitted */
 511     if (offset > num_coefs) {
 512         av_log(avctx, AV_LOG_ERROR, "overflow in spectral RLE, ignoring\n");
 513         return -1;
 514     }
 515
 516     return 0;
 517 }
 518