takdec.c
Go to the documentation of this file.
1 /*
2  * TAK decoder
3  * Copyright (c) 2012 Paul B Mahol
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 
28 #include "libavutil/samplefmt.h"
29 #include "tak.h"
30 #include "avcodec.h"
31 #include "dsputil.h"
32 #include "internal.h"
33 #include "unary.h"
34 
35 #define MAX_SUBFRAMES 8 // max number of subframes per channel
36 #define MAX_PREDICTORS 256
37 
38 typedef struct MCDParam {
39  int8_t present; // decorrelation parameter availability for this channel
40  int8_t index; // index into array of decorrelation types
41  int8_t chan1;
42  int8_t chan2;
43 } MCDParam;
44 
45 typedef struct TAKDecContext {
46  AVCodecContext *avctx; // parent AVCodecContext
47  AVFrame frame; // AVFrame for decoded output
50  GetBitContext gb; // bitstream reader initialized to start at the current frame
51 
52  int uval;
53  int nb_samples; // number of samples in the current frame
55  unsigned int decode_buffer_size;
56  int32_t *decoded[TAK_MAX_CHANNELS]; // decoded samples for each channel
57 
59  int8_t sample_shift[TAK_MAX_CHANNELS]; // shift applied to every sample in the channel
61 
62  int8_t dmode; // channel decorrelation type in the current frame
63 
64  MCDParam mcdparams[TAK_MAX_CHANNELS]; // multichannel decorrelation parameters
65 
66  int16_t *residues;
67  unsigned int residues_buf_size;
69 
70 static const int8_t mc_dmodes[] = { 1, 3, 4, 6, };
71 
72 static const uint16_t predictor_sizes[] = {
73  4, 8, 12, 16, 24, 32, 48, 64, 80, 96, 128, 160, 192, 224, 256, 0,
74 };
75 
76 static const struct CParam {
77  int init;
78  int escape;
79  int scale;
80  int aescape;
81  int bias;
82 } xcodes[50] = {
83  { 0x01, 0x0000001, 0x0000001, 0x0000003, 0x0000008 },
84  { 0x02, 0x0000003, 0x0000001, 0x0000007, 0x0000006 },
85  { 0x03, 0x0000005, 0x0000002, 0x000000E, 0x000000D },
86  { 0x03, 0x0000003, 0x0000003, 0x000000D, 0x0000018 },
87  { 0x04, 0x000000B, 0x0000004, 0x000001C, 0x0000019 },
88  { 0x04, 0x0000006, 0x0000006, 0x000001A, 0x0000030 },
89  { 0x05, 0x0000016, 0x0000008, 0x0000038, 0x0000032 },
90  { 0x05, 0x000000C, 0x000000C, 0x0000034, 0x0000060 },
91  { 0x06, 0x000002C, 0x0000010, 0x0000070, 0x0000064 },
92  { 0x06, 0x0000018, 0x0000018, 0x0000068, 0x00000C0 },
93  { 0x07, 0x0000058, 0x0000020, 0x00000E0, 0x00000C8 },
94  { 0x07, 0x0000030, 0x0000030, 0x00000D0, 0x0000180 },
95  { 0x08, 0x00000B0, 0x0000040, 0x00001C0, 0x0000190 },
96  { 0x08, 0x0000060, 0x0000060, 0x00001A0, 0x0000300 },
97  { 0x09, 0x0000160, 0x0000080, 0x0000380, 0x0000320 },
98  { 0x09, 0x00000C0, 0x00000C0, 0x0000340, 0x0000600 },
99  { 0x0A, 0x00002C0, 0x0000100, 0x0000700, 0x0000640 },
100  { 0x0A, 0x0000180, 0x0000180, 0x0000680, 0x0000C00 },
101  { 0x0B, 0x0000580, 0x0000200, 0x0000E00, 0x0000C80 },
102  { 0x0B, 0x0000300, 0x0000300, 0x0000D00, 0x0001800 },
103  { 0x0C, 0x0000B00, 0x0000400, 0x0001C00, 0x0001900 },
104  { 0x0C, 0x0000600, 0x0000600, 0x0001A00, 0x0003000 },
105  { 0x0D, 0x0001600, 0x0000800, 0x0003800, 0x0003200 },
106  { 0x0D, 0x0000C00, 0x0000C00, 0x0003400, 0x0006000 },
107  { 0x0E, 0x0002C00, 0x0001000, 0x0007000, 0x0006400 },
108  { 0x0E, 0x0001800, 0x0001800, 0x0006800, 0x000C000 },
109  { 0x0F, 0x0005800, 0x0002000, 0x000E000, 0x000C800 },
110  { 0x0F, 0x0003000, 0x0003000, 0x000D000, 0x0018000 },
111  { 0x10, 0x000B000, 0x0004000, 0x001C000, 0x0019000 },
112  { 0x10, 0x0006000, 0x0006000, 0x001A000, 0x0030000 },
113  { 0x11, 0x0016000, 0x0008000, 0x0038000, 0x0032000 },
114  { 0x11, 0x000C000, 0x000C000, 0x0034000, 0x0060000 },
115  { 0x12, 0x002C000, 0x0010000, 0x0070000, 0x0064000 },
116  { 0x12, 0x0018000, 0x0018000, 0x0068000, 0x00C0000 },
117  { 0x13, 0x0058000, 0x0020000, 0x00E0000, 0x00C8000 },
118  { 0x13, 0x0030000, 0x0030000, 0x00D0000, 0x0180000 },
119  { 0x14, 0x00B0000, 0x0040000, 0x01C0000, 0x0190000 },
120  { 0x14, 0x0060000, 0x0060000, 0x01A0000, 0x0300000 },
121  { 0x15, 0x0160000, 0x0080000, 0x0380000, 0x0320000 },
122  { 0x15, 0x00C0000, 0x00C0000, 0x0340000, 0x0600000 },
123  { 0x16, 0x02C0000, 0x0100000, 0x0700000, 0x0640000 },
124  { 0x16, 0x0180000, 0x0180000, 0x0680000, 0x0C00000 },
125  { 0x17, 0x0580000, 0x0200000, 0x0E00000, 0x0C80000 },
126  { 0x17, 0x0300000, 0x0300000, 0x0D00000, 0x1800000 },
127  { 0x18, 0x0B00000, 0x0400000, 0x1C00000, 0x1900000 },
128  { 0x18, 0x0600000, 0x0600000, 0x1A00000, 0x3000000 },
129  { 0x19, 0x1600000, 0x0800000, 0x3800000, 0x3200000 },
130  { 0x19, 0x0C00000, 0x0C00000, 0x3400000, 0x6000000 },
131  { 0x1A, 0x2C00000, 0x1000000, 0x7000000, 0x6400000 },
132  { 0x1A, 0x1800000, 0x1800000, 0x6800000, 0xC000000 },
133 };
134 
136 {
137  ff_tak_init_crc();
138 }
139 
140 static int set_bps_params(AVCodecContext *avctx)
141 {
142  switch (avctx->bits_per_coded_sample) {
143  case 8:
144  avctx->sample_fmt = AV_SAMPLE_FMT_U8P;
145  break;
146  case 16:
148  break;
149  case 24:
151  break;
152  default:
153  av_log(avctx, AV_LOG_ERROR, "unsupported bits per sample: %d\n",
154  avctx->bits_per_coded_sample);
155  return AVERROR_INVALIDDATA;
156  }
158 
159  return 0;
160 }
161 
163 {
164  TAKDecContext *s = avctx->priv_data;
165  int shift = 3 - (avctx->sample_rate / 11025);
166  shift = FFMAX(0, shift);
167  s->uval = FFALIGN(avctx->sample_rate + 511 >> 9, 4) << shift;
168  s->subframe_scale = FFALIGN(avctx->sample_rate + 511 >> 9, 4) << 1;
169 }
170 
172 {
173  TAKDecContext *s = avctx->priv_data;
174 
175  ff_dsputil_init(&s->dsp, avctx);
176 
177  s->avctx = avctx;
179  avctx->coded_frame = &s->frame;
180 
181  set_sample_rate_params(avctx);
182 
183  return set_bps_params(avctx);
184 }
185 
186 static void decode_lpc(int32_t *coeffs, int mode, int length)
187 {
188  int i;
189 
190  if (length < 2)
191  return;
192 
193  if (mode == 1) {
194  int a1 = *coeffs++;
195  for (i = 0; i < length - 1 >> 1; i++) {
196  *coeffs += a1;
197  coeffs[1] += *coeffs;
198  a1 = coeffs[1];
199  coeffs += 2;
200  }
201  if (length - 1 & 1)
202  *coeffs += a1;
203  } else if (mode == 2) {
204  int a1 = coeffs[1];
205  int a2 = a1 + *coeffs;
206  coeffs[1] = a2;
207  if (length > 2) {
208  coeffs += 2;
209  for (i = 0; i < length - 2 >> 1; i++) {
210  int a3 = *coeffs + a1;
211  int a4 = a3 + a2;
212  *coeffs = a4;
213  a1 = coeffs[1] + a3;
214  a2 = a1 + a4;
215  coeffs[1] = a2;
216  coeffs += 2;
217  }
218  if (length & 1)
219  *coeffs += a1 + a2;
220  }
221  } else if (mode == 3) {
222  int a1 = coeffs[1];
223  int a2 = a1 + *coeffs;
224  coeffs[1] = a2;
225  if (length > 2) {
226  int a3 = coeffs[2];
227  int a4 = a3 + a1;
228  int a5 = a4 + a2;
229  coeffs += 3;
230  for (i = 0; i < length - 3; i++) {
231  a3 += *coeffs;
232  a4 += a3;
233  a5 += a4;
234  *coeffs = a5;
235  coeffs++;
236  }
237  }
238  }
239 }
240 
241 static int decode_segment(GetBitContext *gb, int mode, int32_t *decoded,
242  int len)
243 {
244  struct CParam code;
245  int i;
246 
247  if (!mode) {
248  memset(decoded, 0, len * sizeof(*decoded));
249  return 0;
250  }
251 
252  if (mode > FF_ARRAY_ELEMS(xcodes))
253  return AVERROR_INVALIDDATA;
254  code = xcodes[mode - 1];
255 
256  for (i = 0; i < len; i++) {
257  int x = get_bits_long(gb, code.init);
258  if (x >= code.escape && get_bits1(gb)) {
259  x |= 1 << code.init;
260  if (x >= code.aescape) {
261  int scale = get_unary(gb, 1, 9);
262  if (scale == 9) {
263  int scale_bits = get_bits(gb, 3);
264  if (scale_bits > 0) {
265  if (scale_bits == 7) {
266  scale_bits += get_bits(gb, 5);
267  if (scale_bits > 29)
268  return AVERROR_INVALIDDATA;
269  }
270  scale = get_bits_long(gb, scale_bits) + 1;
271  x += code.scale * scale;
272  }
273  x += code.bias;
274  } else
275  x += code.scale * scale - code.escape;
276  } else
277  x -= code.escape;
278  }
279  decoded[i] = (x >> 1) ^ -(x & 1);
280  }
281 
282  return 0;
283 }
284 
285 static int decode_residues(TAKDecContext *s, int32_t *decoded, int length)
286 {
287  GetBitContext *gb = &s->gb;
288  int i, mode, ret;
289 
290  if (length > s->nb_samples)
291  return AVERROR_INVALIDDATA;
292 
293  if (get_bits1(gb)) {
294  int wlength, rval;
295  int coding_mode[128];
296 
297  wlength = length / s->uval;
298 
299  rval = length - (wlength * s->uval);
300 
301  if (rval < s->uval / 2)
302  rval += s->uval;
303  else
304  wlength++;
305 
306  if (wlength <= 1 || wlength > 128)
307  return AVERROR_INVALIDDATA;
308 
309  coding_mode[0] = mode = get_bits(gb, 6);
310 
311  for (i = 1; i < wlength; i++) {
312  int c = get_unary(gb, 1, 6);
313 
314  switch (c) {
315  case 6:
316  mode = get_bits(gb, 6);
317  break;
318  case 5:
319  case 4:
320  case 3: {
321  /* mode += sign ? (1 - c) : (c - 1) */
322  int sign = get_bits1(gb);
323  mode += (-sign ^ (c - 1)) + sign;
324  break;
325  }
326  case 2:
327  mode++;
328  break;
329  case 1:
330  mode--;
331  break;
332  }
333  coding_mode[i] = mode;
334  }
335 
336  i = 0;
337  while (i < wlength) {
338  int len = 0;
339 
340  mode = coding_mode[i];
341  do {
342  if (i >= wlength - 1)
343  len += rval;
344  else
345  len += s->uval;
346  i++;
347 
348  if (i == wlength)
349  break;
350  } while (coding_mode[i] == mode);
351 
352  if ((ret = decode_segment(gb, mode, decoded, len)) < 0)
353  return ret;
354  decoded += len;
355  }
356  } else {
357  mode = get_bits(gb, 6);
358  if ((ret = decode_segment(gb, mode, decoded, length)) < 0)
359  return ret;
360  }
361 
362  return 0;
363 }
364 
366 {
367  if (get_bits1(gb))
368  return get_bits(gb, 4) + 1;
369  else
370  return 0;
371 }
372 
373 static void decode_filter_coeffs(TAKDecContext *s, int filter_order, int size,
374  int filter_quant, int16_t *filter)
375 {
376  GetBitContext *gb = &s->gb;
377  int i, j, a, b;
378  int filter_tmp[MAX_PREDICTORS];
379  int16_t predictors[MAX_PREDICTORS];
380 
381  predictors[0] = get_sbits(gb, 10);
382  predictors[1] = get_sbits(gb, 10);
383  predictors[2] = get_sbits(gb, size) << (10 - size);
384  predictors[3] = get_sbits(gb, size) << (10 - size);
385  if (filter_order > 4) {
386  int av_uninit(code_size);
387  int code_size_base = size - get_bits1(gb);
388 
389  for (i = 4; i < filter_order; i++) {
390  if (!(i & 3))
391  code_size = code_size_base - get_bits(gb, 2);
392  predictors[i] = get_sbits(gb, code_size) << (10 - size);
393  }
394  }
395 
396  filter_tmp[0] = predictors[0] << 6;
397  for (i = 1; i < filter_order; i++) {
398  int *p1 = &filter_tmp[0];
399  int *p2 = &filter_tmp[i - 1];
400 
401  for (j = 0; j < (i + 1) / 2; j++) {
402  int tmp = *p1 + (predictors[i] * *p2 + 256 >> 9);
403  *p2 = *p2 + (predictors[i] * *p1 + 256 >> 9);
404  *p1 = tmp;
405  p1++;
406  p2--;
407  }
408 
409  filter_tmp[i] = predictors[i] << 6;
410  }
411 
412  a = 1 << (32 - (15 - filter_quant));
413  b = 1 << ((15 - filter_quant) - 1);
414  for (i = 0, j = filter_order - 1; i < filter_order / 2; i++, j--) {
415  filter[j] = a - ((filter_tmp[i] + b) >> (15 - filter_quant));
416  filter[i] = a - ((filter_tmp[j] + b) >> (15 - filter_quant));
417  }
418 }
419 
420 static int decode_subframe(TAKDecContext *s, int32_t *decoded,
421  int subframe_size, int prev_subframe_size)
422 {
424  GetBitContext *gb = &s->gb;
425  int i, ret;
426  int dshift, size, filter_quant, filter_order;
427 
428  memset(filter, 0, MAX_PREDICTORS * sizeof(*filter));
429 
430  if (!get_bits1(gb))
431  return decode_residues(s, decoded, subframe_size);
432 
433  filter_order = predictor_sizes[get_bits(gb, 4)];
434 
435  if (prev_subframe_size > 0 && get_bits1(gb)) {
436  if (filter_order > prev_subframe_size)
437  return AVERROR_INVALIDDATA;
438 
439  decoded -= filter_order;
440  subframe_size += filter_order;
441 
442  if (filter_order > subframe_size)
443  return AVERROR_INVALIDDATA;
444  } else {
445  int lpc_mode;
446 
447  if (filter_order > subframe_size)
448  return AVERROR_INVALIDDATA;
449 
450  lpc_mode = get_bits(gb, 2);
451  if (lpc_mode > 2)
452  return AVERROR_INVALIDDATA;
453 
454  if ((ret = decode_residues(s, decoded, filter_order)) < 0)
455  return ret;
456 
457  if (lpc_mode)
458  decode_lpc(decoded, lpc_mode, filter_order);
459  }
460 
461  dshift = get_bits_esc4(gb);
462  size = get_bits1(gb) + 6;
463 
464  filter_quant = 10;
465  if (get_bits1(gb)) {
466  filter_quant -= get_bits(gb, 3) + 1;
467  if (filter_quant < 3)
468  return AVERROR_INVALIDDATA;
469  }
470 
471  decode_filter_coeffs(s, filter_order, size, filter_quant, filter);
472 
473  if ((ret = decode_residues(s, &decoded[filter_order],
474  subframe_size - filter_order)) < 0)
475  return ret;
476 
478  FFALIGN(subframe_size + 16, 16) * sizeof(*s->residues));
479  if (!s->residues)
480  return AVERROR(ENOMEM);
481  memset(s->residues, 0, s->residues_buf_size);
482 
483  for (i = 0; i < filter_order; i++)
484  s->residues[i] = *decoded++ >> dshift;
485 
486  for (i = 0; i < subframe_size - filter_order; i++) {
487  int v = 1 << (filter_quant - 1);
488 
489  v += s->dsp.scalarproduct_int16(&s->residues[i], filter,
490  FFALIGN(filter_order, 16));
491 
492  v = (av_clip(v >> filter_quant, -8192, 8191) << dshift) - *decoded;
493  *decoded++ = v;
494  s->residues[filter_order + i] = v >> dshift;
495  }
496 
497  emms_c();
498 
499  return 0;
500 }
501 
502 static int decode_channel(TAKDecContext *s, int chan)
503 {
504  AVCodecContext *avctx = s->avctx;
505  GetBitContext *gb = &s->gb;
506  int32_t *decoded = s->decoded[chan];
507  int left = s->nb_samples - 1;
508  int i, prev, ret, nb_subframes;
509  int subframe_len[MAX_SUBFRAMES];
510 
511  s->sample_shift[chan] = get_bits_esc4(gb);
512  if (s->sample_shift[chan] >= avctx->bits_per_coded_sample)
513  return AVERROR_INVALIDDATA;
514 
515  /* NOTE: TAK 2.2.0 appears to set the sample value to 0 if
516  * bits_per_coded_sample - sample_shift is 1, but this produces
517  * non-bit-exact output. Reading the 1 bit using get_sbits() instead
518  * of skipping it produces bit-exact output. This has been reported
519  * to the TAK author. */
520  *decoded++ = get_sbits(gb,
521  avctx->bits_per_coded_sample -
522  s->sample_shift[chan]);
523  s->lpc_mode[chan] = get_bits(gb, 2);
524  nb_subframes = get_bits(gb, 3) + 1;
525 
526  i = 0;
527  if (nb_subframes > 1) {
528  if (get_bits_left(gb) < (nb_subframes - 1) * 6)
529  return AVERROR_INVALIDDATA;
530 
531  prev = 0;
532  for (; i < nb_subframes - 1; i++) {
533  int subframe_end = get_bits(gb, 6) * s->subframe_scale;
534  if (subframe_end <= prev)
535  return AVERROR_INVALIDDATA;
536  subframe_len[i] = subframe_end - prev;
537  left -= subframe_len[i];
538  prev = subframe_end;
539  }
540 
541  if (left <= 0)
542  return AVERROR_INVALIDDATA;
543  }
544  subframe_len[i] = left;
545 
546  prev = 0;
547  for (i = 0; i < nb_subframes; i++) {
548  if ((ret = decode_subframe(s, decoded, subframe_len[i], prev)) < 0)
549  return ret;
550  decoded += subframe_len[i];
551  prev = subframe_len[i];
552  }
553 
554  return 0;
555 }
556 
557 static int decorrelate(TAKDecContext *s, int c1, int c2, int length)
558 {
559  GetBitContext *gb = &s->gb;
560  int32_t *p1 = s->decoded[c1] + 1;
561  int32_t *p2 = s->decoded[c2] + 1;
562  int i;
563  int dshift, dfactor;
564 
565  switch (s->dmode) {
566  case 1: /* left/side */
567  for (i = 0; i < length; i++) {
568  int32_t a = p1[i];
569  int32_t b = p2[i];
570  p2[i] = a + b;
571  }
572  break;
573  case 2: /* side/right */
574  for (i = 0; i < length; i++) {
575  int32_t a = p1[i];
576  int32_t b = p2[i];
577  p1[i] = b - a;
578  }
579  break;
580  case 3: /* side/mid */
581  for (i = 0; i < length; i++) {
582  int32_t a = p1[i];
583  int32_t b = p2[i];
584  a -= b >> 1;
585  p1[i] = a;
586  p2[i] = a + b;
587  }
588  break;
589  case 4: /* side/left with scale factor */
590  FFSWAP(int32_t*, p1, p2);
591  case 5: /* side/right with scale factor */
592  dshift = get_bits_esc4(gb);
593  dfactor = get_sbits(gb, 10);
594  for (i = 0; i < length; i++) {
595  int32_t a = p1[i];
596  int32_t b = p2[i];
597  b = dfactor * (b >> dshift) + 128 >> 8 << dshift;
598  p1[i] = b - a;
599  }
600  break;
601  case 6:
602  FFSWAP(int32_t*, p1, p2);
603  case 7: {
605  int length2, order_half, filter_order, dval1, dval2;
606  int av_uninit(code_size);
607 
608  memset(filter, 0, MAX_PREDICTORS * sizeof(*filter));
609 
610  if (length < 256)
611  return AVERROR_INVALIDDATA;
612 
613  dshift = get_bits_esc4(gb);
614  filter_order = 8 << get_bits1(gb);
615  dval1 = get_bits1(gb);
616  dval2 = get_bits1(gb);
617 
618  for (i = 0; i < filter_order; i++) {
619  if (!(i & 3))
620  code_size = 14 - get_bits(gb, 3);
621  filter[i] = get_sbits(gb, code_size);
622  }
623 
624  order_half = filter_order / 2;
625  length2 = length - (filter_order - 1);
626 
627  /* decorrelate beginning samples */
628  if (dval1) {
629  for (i = 0; i < order_half; i++) {
630  int32_t a = p1[i];
631  int32_t b = p2[i];
632  p1[i] = a + b;
633  }
634  }
635 
636  /* decorrelate ending samples */
637  if (dval2) {
638  for (i = length2 + order_half; i < length; i++) {
639  int32_t a = p1[i];
640  int32_t b = p2[i];
641  p1[i] = a + b;
642  }
643  }
644 
646  FFALIGN(length + 16, 16) * sizeof(*s->residues));
647  if (!s->residues)
648  return AVERROR(ENOMEM);
649  memset(s->residues, 0, s->residues_buf_size);
650 
651  for (i = 0; i < length; i++)
652  s->residues[i] = p2[i] >> dshift;
653 
654  p1 += order_half;
655 
656  for (i = 0; i < length2; i++) {
657  int v = 1 << 9;
658 
659  v += s->dsp.scalarproduct_int16(&s->residues[i], filter,
660  FFALIGN(filter_order, 16));
661 
662  p1[i] = (av_clip(v >> 10, -8192, 8191) << dshift) - p1[i];
663  }
664 
665  emms_c();
666  break;
667  }
668  }
669 
670  return 0;
671 }
672 
673 static int tak_decode_frame(AVCodecContext *avctx, void *data,
674  int *got_frame_ptr, AVPacket *pkt)
675 {
676  TAKDecContext *s = avctx->priv_data;
677  GetBitContext *gb = &s->gb;
678  int chan, i, ret, hsize;
679 
680  if (pkt->size < TAK_MIN_FRAME_HEADER_BYTES)
681  return AVERROR_INVALIDDATA;
682 
683  init_get_bits(gb, pkt->data, pkt->size * 8);
684 
685  if ((ret = ff_tak_decode_frame_header(avctx, gb, &s->ti, 0)) < 0)
686  return ret;
687 
689  av_log_missing_feature(avctx, "frame metadata", 1);
690  return AVERROR_PATCHWELCOME;
691  }
692 
693  hsize = get_bits_count(gb) / 8;
694  if (avctx->err_recognition & AV_EF_CRCCHECK) {
695  if (ff_tak_check_crc(pkt->data, hsize)) {
696  av_log(avctx, AV_LOG_ERROR, "CRC error\n");
697  return AVERROR_INVALIDDATA;
698  }
699  }
700 
701  if (s->ti.codec != TAK_CODEC_MONO_STEREO &&
703  av_log(avctx, AV_LOG_ERROR, "unsupported codec: %d\n", s->ti.codec);
704  return AVERROR_PATCHWELCOME;
705  }
706  if (s->ti.data_type) {
707  av_log(avctx, AV_LOG_ERROR,
708  "unsupported data type: %d\n", s->ti.data_type);
709  return AVERROR_INVALIDDATA;
710  }
711  if (s->ti.codec == TAK_CODEC_MONO_STEREO && s->ti.channels > 2) {
712  av_log(avctx, AV_LOG_ERROR,
713  "invalid number of channels: %d\n", s->ti.channels);
714  return AVERROR_INVALIDDATA;
715  }
716  if (s->ti.channels > 6) {
717  av_log(avctx, AV_LOG_ERROR,
718  "unsupported number of channels: %d\n", s->ti.channels);
719  return AVERROR_INVALIDDATA;
720  }
721 
722  if (s->ti.frame_samples <= 0) {
723  av_log(avctx, AV_LOG_ERROR, "unsupported/invalid number of samples\n");
724  return AVERROR_INVALIDDATA;
725  }
726 
727  if (s->ti.bps != avctx->bits_per_coded_sample) {
728  avctx->bits_per_coded_sample = s->ti.bps;
729  if ((ret = set_bps_params(avctx)) < 0)
730  return ret;
731  }
732  if (s->ti.sample_rate != avctx->sample_rate) {
733  avctx->sample_rate = s->ti.sample_rate;
734  set_sample_rate_params(avctx);
735  }
736  if (s->ti.ch_layout)
737  avctx->channel_layout = s->ti.ch_layout;
738  avctx->channels = s->ti.channels;
739 
741  : s->ti.frame_samples;
742 
743  s->frame.nb_samples = s->nb_samples;
744  if ((ret = ff_get_buffer(avctx, &s->frame)) < 0)
745  return ret;
746 
747  if (avctx->bits_per_coded_sample <= 16) {
748  int buf_size = av_samples_get_buffer_size(NULL, avctx->channels,
749  s->nb_samples,
750  AV_SAMPLE_FMT_S32P, 0);
751  av_fast_malloc(&s->decode_buffer, &s->decode_buffer_size, buf_size);
752  if (!s->decode_buffer)
753  return AVERROR(ENOMEM);
755  s->decode_buffer, avctx->channels,
757  if (ret < 0)
758  return ret;
759  } else {
760  for (chan = 0; chan < avctx->channels; chan++)
761  s->decoded[chan] = (int32_t *)s->frame.extended_data[chan];
762  }
763 
764  if (s->nb_samples < 16) {
765  for (chan = 0; chan < avctx->channels; chan++) {
766  int32_t *decoded = s->decoded[chan];
767  for (i = 0; i < s->nb_samples; i++)
768  decoded[i] = get_sbits(gb, avctx->bits_per_coded_sample);
769  }
770  } else {
771  if (s->ti.codec == TAK_CODEC_MONO_STEREO) {
772  for (chan = 0; chan < avctx->channels; chan++)
773  if (ret = decode_channel(s, chan))
774  return ret;
775 
776  if (avctx->channels == 2) {
777  if (get_bits1(gb)) {
778  // some kind of subframe length, but it seems to be unused
779  skip_bits(gb, 6);
780  }
781 
782  s->dmode = get_bits(gb, 3);
783  if (ret = decorrelate(s, 0, 1, s->nb_samples - 1))
784  return ret;
785  }
786  } else if (s->ti.codec == TAK_CODEC_MULTICHANNEL) {
787  if (get_bits1(gb)) {
788  int ch_mask = 0;
789 
790  chan = get_bits(gb, 4) + 1;
791  if (chan > avctx->channels)
792  return AVERROR_INVALIDDATA;
793 
794  for (i = 0; i < chan; i++) {
795  int nbit = get_bits(gb, 4);
796 
797  if (nbit >= avctx->channels)
798  return AVERROR_INVALIDDATA;
799 
800  if (ch_mask & 1 << nbit)
801  return AVERROR_INVALIDDATA;
802 
803  s->mcdparams[i].present = get_bits1(gb);
804  if (s->mcdparams[i].present) {
805  s->mcdparams[i].index = get_bits(gb, 2);
806  s->mcdparams[i].chan2 = get_bits(gb, 4);
807  if (s->mcdparams[i].index == 1) {
808  if ((nbit == s->mcdparams[i].chan2) ||
809  (ch_mask & 1 << s->mcdparams[i].chan2))
810  return AVERROR_INVALIDDATA;
811 
812  ch_mask |= 1 << s->mcdparams[i].chan2;
813  } else if (!(ch_mask & 1 << s->mcdparams[i].chan2)) {
814  return AVERROR_INVALIDDATA;
815  }
816  }
817  s->mcdparams[i].chan1 = nbit;
818 
819  ch_mask |= 1 << nbit;
820  }
821  } else {
822  chan = avctx->channels;
823  for (i = 0; i < chan; i++) {
824  s->mcdparams[i].present = 0;
825  s->mcdparams[i].chan1 = i;
826  }
827  }
828 
829  for (i = 0; i < chan; i++) {
830  if (s->mcdparams[i].present && s->mcdparams[i].index == 1)
831  if (ret = decode_channel(s, s->mcdparams[i].chan2))
832  return ret;
833 
834  if (ret = decode_channel(s, s->mcdparams[i].chan1))
835  return ret;
836 
837  if (s->mcdparams[i].present) {
838  s->dmode = mc_dmodes[s->mcdparams[i].index];
839  if (ret = decorrelate(s,
840  s->mcdparams[i].chan2,
841  s->mcdparams[i].chan1,
842  s->nb_samples - 1))
843  return ret;
844  }
845  }
846  }
847 
848  for (chan = 0; chan < avctx->channels; chan++) {
849  int32_t *decoded = s->decoded[chan];
850 
851  if (s->lpc_mode[chan])
852  decode_lpc(decoded, s->lpc_mode[chan], s->nb_samples);
853 
854  if (s->sample_shift[chan] > 0)
855  for (i = 0; i < s->nb_samples; i++)
856  decoded[i] <<= s->sample_shift[chan];
857  }
858  }
859 
860  align_get_bits(gb);
861  skip_bits(gb, 24);
862  if (get_bits_left(gb) < 0)
863  av_log(avctx, AV_LOG_DEBUG, "overread\n");
864  else if (get_bits_left(gb) > 0)
865  av_log(avctx, AV_LOG_DEBUG, "underread\n");
866 
867  if (avctx->err_recognition & AV_EF_CRCCHECK) {
868  if (ff_tak_check_crc(pkt->data + hsize,
869  get_bits_count(gb) / 8 - hsize)) {
870  av_log(avctx, AV_LOG_ERROR, "CRC error\n");
871  return AVERROR_INVALIDDATA;
872  }
873  }
874 
875  /* convert to output buffer */
876  switch (avctx->sample_fmt) {
877  case AV_SAMPLE_FMT_U8P:
878  for (chan = 0; chan < avctx->channels; chan++) {
879  uint8_t *samples = (uint8_t *)s->frame.extended_data[chan];
880  int32_t *decoded = s->decoded[chan];
881  for (i = 0; i < s->nb_samples; i++)
882  samples[i] = decoded[i] + 0x80;
883  }
884  break;
885  case AV_SAMPLE_FMT_S16P:
886  for (chan = 0; chan < avctx->channels; chan++) {
887  int16_t *samples = (int16_t *)s->frame.extended_data[chan];
888  int32_t *decoded = s->decoded[chan];
889  for (i = 0; i < s->nb_samples; i++)
890  samples[i] = decoded[i];
891  }
892  break;
893  case AV_SAMPLE_FMT_S32P:
894  for (chan = 0; chan < avctx->channels; chan++) {
895  int32_t *samples = (int32_t *)s->frame.extended_data[chan];
896  for (i = 0; i < s->nb_samples; i++)
897  samples[i] <<= 8;
898  }
899  break;
900  }
901 
902  *got_frame_ptr = 1;
903  *(AVFrame *)data = s->frame;
904 
905  return pkt->size;
906 }
907 
909 {
910  TAKDecContext *s = avctx->priv_data;
911 
912  av_freep(&s->decode_buffer);
913  av_freep(&s->residues);
914 
915  return 0;
916 }
917 
919  .name = "tak",
920  .type = AVMEDIA_TYPE_AUDIO,
921  .id = AV_CODEC_ID_TAK,
922  .priv_data_size = sizeof(TAKDecContext),
924  .init_static_data = tak_init_static_data,
927  .capabilities = CODEC_CAP_DR1,
928  .long_name = NULL_IF_CONFIG_SMALL("TAK (Tom's lossless Audio Kompressor)"),
929  .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_U8P,
933 };
#define TAK_MAX_CHANNELS
Definition: tak.h:65
static const int16_t coeffs[28]
av_cold void ff_dsputil_init(DSPContext *c, AVCodecContext *avctx)
Definition: dsputil.c:2656
static int set_bps_params(AVCodecContext *avctx)
Definition: takdec.c:140
static const uint16_t predictor_sizes[]
Definition: takdec.c:72
int ff_tak_decode_frame_header(AVCodecContext *avctx, GetBitContext *gb, TAKStreamInfo *ti, int log_level_offset)
Validate and decode a frame header.
Definition: tak.c:121
static int16_t * samples
int size
This structure describes decoded (raw) audio or video data.
Definition: avcodec.h:989
#define c2
Definition: idct_sh4.c:28
int channels
Definition: tak.h:134
int av_samples_get_buffer_size(int *linesize, int nb_channels, int nb_samples, enum AVSampleFormat sample_fmt, int align)
Get the required buffer size for the given audio parameters.
Definition: samplefmt.c:108
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
Definition: get_bits.h:237
int aescape
Definition: takdec.c:80
AVFrame * coded_frame
the picture in the bitstream
Definition: avcodec.h:2725
static void align_get_bits(GetBitContext *s)
Definition: get_bits.h:412
MCDParam mcdparams[TAK_MAX_CHANNELS]
Definition: takdec.c:64
int size
Definition: avcodec.h:916
static int decode_subframe(TAKDecContext *s, int32_t *decoded, int subframe_size, int prev_subframe_size)
Definition: takdec.c:420
uint64_t ch_layout
Definition: tak.h:139
#define a1
Definition: regdef.h:47
int bits_per_raw_sample
Bits per sample/pixel of internal libavcodec pixel/sample format.
Definition: avcodec.h:2711
AVCodec.
Definition: avcodec.h:2960
static int get_sbits(GetBitContext *s, int n)
Definition: get_bits.h:223
int8_t lpc_mode[TAK_MAX_CHANNELS]
Definition: takdec.c:58
AVFrame frame
Definition: takdec.c:47
void av_freep(void *arg)
Free a memory block which has been allocated with av_malloc(z)() or av_realloc() and set the pointer ...
Definition: mem.c:151
struct MCDParam MCDParam
void av_fast_malloc(void *ptr, unsigned int *size, size_t min_size)
Allocate a buffer, reusing the given one if large enough.
Definition: utils.c:72
int av_samples_fill_arrays(uint8_t **audio_data, int *linesize, const uint8_t *buf, int nb_channels, int nb_samples, enum AVSampleFormat sample_fmt, int align)
Fill channel data pointers and linesize for samples with sample format sample_fmt.
Definition: samplefmt.c:140
int16_t * residues
Definition: takdec.c:66
unsigned int decode_buffer_size
Definition: takdec.c:55
static int decode(MimicContext *ctx, int quality, int num_coeffs, int is_iframe)
Definition: mimic.c:228
#define a3
Definition: regdef.h:49
int init
Definition: takdec.c:77
enum AVSampleFormat sample_fmt
audio sample format
Definition: avcodec.h:2112
uint8_t
int flags
Definition: tak.h:130
static av_cold int tak_decode_close(AVCodecContext *avctx)
Definition: takdec.c:908
#define b
Definition: input.c:52
int ff_tak_check_crc(const uint8_t *buf, unsigned int buf_size)
Definition: tak.c:70
#define emms_c()
Definition: internal.h:145
static void decode_filter_coeffs(TAKDecContext *s, int filter_order, int size, int filter_quant, int16_t *filter)
Definition: takdec.c:373
AVCodecContext * avctx
Definition: takdec.c:46
const char data[16]
Definition: mxf.c:66
uint8_t * data
Definition: avcodec.h:915
static int get_bits_count(const GetBitContext *s)
Definition: get_bits.h:192
int escape
Definition: takdec.c:78
signed 32 bits, planar
Definition: samplefmt.h:59
#define LOCAL_ALIGNED_16(t, v,...)
Definition: dsputil.h:602
int bits_per_coded_sample
bits per sample/pixel from the demuxer (needed for huffyuv).
Definition: avcodec.h:2704
int data_type
Definition: tak.h:132
static int decode_channel(TAKDecContext *s, int chan)
Definition: takdec.c:502
static int init(AVCodecParserContext *s)
Definition: h264_parser.c:335
#define MAX_SUBFRAMES
Definition: takdec.c:35
int8_t index
Definition: takdec.c:40
static int get_bits_left(GetBitContext *gb)
Definition: get_bits.h:547
int32_t(* scalarproduct_int16)(const int16_t *v1, const int16_t *v2, int len)
Calculate scalar product of two vectors.
Definition: dsputil.h:450
int uval
Definition: takdec.c:52
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:88
av_cold void ff_tak_init_crc(void)
Definition: tak.c:62
void av_log(void *avcl, int level, const char *fmt,...)
Definition: log.c:146
const char * name
Name of the codec implementation.
Definition: avcodec.h:2967
int8_t present
Definition: takdec.c:39
int last_frame_samples
Definition: tak.h:138
static int decode_residues(TAKDecContext *s, int32_t *decoded, int length)
Definition: takdec.c:285
uint64_t channel_layout
Audio channel layout.
Definition: avcodec.h:2165
int bps
Definition: tak.h:135
static void filter(MpegAudioContext *s, int ch, const short *samples, int incr)
Definition: mpegaudioenc.c:318
static av_cold int tak_decode_init(AVCodecContext *avctx)
Definition: takdec.c:171
int8_t chan2
Definition: takdec.c:42
int err_recognition
Error recognition; may misdetect some more or less valid parts as errors.
Definition: avcodec.h:2602
static const struct CParam xcodes[50]
int scale
Definition: takdec.c:79
unsigned int residues_buf_size
Definition: takdec.c:67
int ff_get_buffer(AVCodecContext *avctx, AVFrame *frame)
Get a buffer for a frame.
Definition: utils.c:464
int32_t
#define a2
Definition: regdef.h:48
static int get_bits_esc4(GetBitContext *gb)
Definition: takdec.c:365
uint8_t * decode_buffer
Definition: takdec.c:54
static int decorrelate(TAKDecContext *s, int c1, int c2, int length)
Definition: takdec.c:557
static void decode_lpc(int32_t *coeffs, int mode, int length)
Definition: takdec.c:186
AVCodec ff_tak_decoder
Definition: takdec.c:918
NULL
Definition: eval.c:52
Definition: takdec.c:76
#define a5
Definition: regdef.h:51
#define TAK_FRAME_FLAG_HAS_METADATA
Definition: tak.h:63
external API header
#define TAK_MIN_FRAME_HEADER_BYTES
Definition: tak.h:98
AV_SAMPLE_FMT_NONE
Definition: avconv_filter.c:63
int sample_rate
samples per second
Definition: avcodec.h:2104
main external API structure.
Definition: avcodec.h:1339
static void close(AVCodecParserContext *s)
Definition: h264_parser.c:326
enum TAKCodecType codec
Definition: tak.h:131
struct TAKDecContext TAKDecContext
int32_t * decoded[TAK_MAX_CHANNELS]
Definition: takdec.c:56
static unsigned int get_bits1(GetBitContext *s)
Definition: get_bits.h:268
void avcodec_get_frame_defaults(AVFrame *frame)
Set the fields of the given AVFrame to default values.
Definition: utils.c:602
static void skip_bits(GetBitContext *s, int n)
Definition: get_bits.h:260
int frame_samples
Definition: tak.h:137
void av_log_missing_feature(void *avc, const char *feature, int want_sample)
Log a generic warning message about a missing feature.
Definition: utils.c:2005
TAK (Tom's lossless Audio Kompressor) decoder/demuxer common functions.
static int init_get_bits(GetBitContext *s, const uint8_t *buffer, int bit_size)
Initialize GetBitContext.
Definition: get_bits.h:372
GetBitContext gb
Definition: takdec.c:50
int8_t dmode
Definition: takdec.c:62
static unsigned int get_bits_long(GetBitContext *s, int n)
Read 0-32 bits.
Definition: get_bits.h:301
int sample_rate
Definition: tak.h:133
static const uint16_t scale[4]
int bias
Definition: takdec.c:81
int nb_samples
Definition: takdec.c:53
TAKStreamInfo ti
Definition: takdec.c:49
common internal api header.
int8_t chan1
Definition: takdec.c:41
static int get_unary(GetBitContext *gb, int stop, int len)
Get unary code of limited length.
Definition: unary.h:33
static int decode_segment(GetBitContext *gb, int mode, int32_t *decoded, int len)
Definition: takdec.c:241
AVSampleFormat
Audio Sample Formats.
Definition: samplefmt.h:49
static const int8_t mc_dmodes[]
Definition: takdec.c:70
int subframe_scale
Definition: takdec.c:60
DSP utils.
#define a4
Definition: regdef.h:50
void * priv_data
Definition: avcodec.h:1382
static int tak_decode_frame(AVCodecContext *avctx, void *data, int *got_frame_ptr, AVPacket *pkt)
Definition: takdec.c:673
unsigned 8 bits, planar
Definition: samplefmt.h:57
int len
int channels
number of audio channels
Definition: avcodec.h:2105
#define MAX_PREDICTORS
Definition: takdec.c:36
signed 16 bits, planar
Definition: samplefmt.h:58
static void set_sample_rate_params(AVCodecContext *avctx)
Definition: takdec.c:162
int8_t sample_shift[TAK_MAX_CHANNELS]
Definition: takdec.c:59
static av_cold void tak_init_static_data(AVCodec *codec)
Definition: takdec.c:135
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: avcodec.h:1028
This structure stores compressed data.
Definition: avcodec.h:898
DSPContext dsp
Definition: takdec.c:48
int nb_samples
number of audio samples (per channel) described by this frame
Definition: avcodec.h:1042
for(j=16;j >0;--j)
#define c1
Definition: idct_sh4.c:27
DSPContext.
Definition: dsputil.h:194