You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

527 lines
17KB

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