1 /*
2 * ALAC (Apple Lossless Audio Codec) decoder
3 * Copyright (c) 2005 David Hammerton
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 /**
23 * @file
24 * ALAC (Apple Lossless Audio Codec) decoder
25 * @author 2005 David Hammerton
26 * @see http://crazney.net/programs/itunes/alac.html
27 *
28 * Note: This decoder expects a 36-byte QuickTime atom to be
29 * passed through the extradata[_size] fields. This atom is tacked onto
30 * the end of an 'alac' stsd atom and has the following format:
31 *
32 * 32bit atom size
33 * 32bit tag ("alac")
34 * 32bit tag version (0)
35 * 32bit samples per frame (used when not set explicitly in the frames)
36 * 8bit compatible version (0)
37 * 8bit sample size
38 * 8bit history mult (40)
39 * 8bit initial history (10)
40 * 8bit rice param limit (14)
41 * 8bit channels
42 * 16bit maxRun (255)
43 * 32bit max coded frame size (0 means unknown)
44 * 32bit average bitrate (0 means unknown)
45 * 32bit samplerate
46 */
47
48 #include <inttypes.h>
49
60
61 #define ALAC_EXTRADATA_SIZE 36
62
68
72
78
79 int extra_bits;
/**< number of extra bits beyond 16-bit */
80 int nb_samples;
/**< number of samples in the current frame */
81
85
87 {
89
90 if (x > 8) { /* RICE THRESHOLD */
91 /* use alternative encoding */
93 } else if (k != 1) {
95
96 /* multiply x by 2^k - 1, as part of their strange algorithm */
97 x = (x << k) - x;
98
99 if (extrabits > 1) {
100 x += extrabits - 1;
102 } else
104 }
105 return x;
106 }
107
109 int nb_samples,
int bps,
int rice_history_mult)
110 {
111 int i;
113 int sign_modifier = 0;
114
115 for (i = 0; i < nb_samples; i++) {
116 int k;
117 unsigned int x;
118
120 return -1;
121
122 /* calculate rice param and decode next value */
123 k =
av_log2((history >> 9) + 3);
126 x += sign_modifier;
127 sign_modifier = 0;
128 output_buffer[i] = (x >> 1) ^ -(x & 1);
129
130 /* update the history */
131 if (x > 0xffff)
132 history = 0xffff;
133 else
134 history += x * rice_history_mult -
135 ((history * rice_history_mult) >> 9);
136
137 /* special case: there may be compressed blocks of 0 */
138 if ((history < 128) && (i + 1 < nb_samples)) {
139 int block_size;
140
141 /* calculate rice param and decode block size */
142 k = 7 -
av_log2(history) + ((history + 16) >> 6);
145
146 if (block_size > 0) {
147 if (block_size >= nb_samples - i) {
149 "invalid zero block size of %d %d %d\n", block_size,
150 nb_samples, i);
151 block_size = nb_samples - i - 1;
152 }
153 memset(&output_buffer[i + 1], 0,
154 block_size * sizeof(*output_buffer));
155 i += block_size;
156 }
157 if (block_size <= 0xffff)
158 sign_modifier = 1;
159 history = 0;
160 }
161 }
162 return 0;
163 }
164
166 {
168 }
169
171 int nb_samples,
int bps, int16_t *lpc_coefs,
172 int lpc_order, int lpc_quant)
173 {
174 int i;
176
177 /* first sample always copies */
178 *buffer_out = *error_buffer;
179
180 if (nb_samples <= 1)
181 return;
182
183 if (!lpc_order) {
184 memcpy(&buffer_out[1], &error_buffer[1],
185 (nb_samples - 1) * sizeof(*buffer_out));
186 return;
187 }
188
189 if (lpc_order == 31) {
190 /* simple 1st-order prediction */
191 for (i = 1; i < nb_samples; i++) {
192 buffer_out[i] =
sign_extend(buffer_out[i - 1] + error_buffer[i],
193 bps);
194 }
195 return;
196 }
197
198 /* read warm-up samples */
199 for (i = 1; i <= lpc_order && i < nb_samples; i++)
200 buffer_out[i] =
sign_extend(buffer_out[i - 1] + error_buffer[i], bps);
201
202 /* NOTE: 4 and 8 are very common cases that could be optimized. */
203
204 for (; i < nb_samples; i++) {
205 int j;
207 int error_val = error_buffer[i];
208 int error_sign;
209 int d = *pred++;
210
211 /* LPC prediction */
212 for (j = 0; j < lpc_order; j++)
213 val += (pred[j] - d) * lpc_coefs[j];
214 val = (val + (1 << (lpc_quant - 1))) >> lpc_quant;
215 val += d + error_val;
217
218 /* adapt LPC coefficients */
220 if (error_sign) {
221 for (j = 0; j < lpc_order && error_val * error_sign > 0; j++) {
222 int sign;
223 val = d - pred[j];
225 lpc_coefs[j] -= sign;
226 val *= sign;
227 error_val -= (val >> lpc_quant) * (j + 1);
228 }
229 }
230 }
231 }
232
234 int decorr_shift, int decorr_left_weight)
235 {
236 int i;
237
238 for (i = 0; i < nb_samples; i++) {
240
241 a = buffer[0][i];
242 b = buffer[1][i];
243
244 a -= (b * decorr_left_weight) >> decorr_shift;
246
249 }
250 }
251
254 {
255 int i, ch;
256
257 for (ch = 0; ch < channels; ch++)
258 for (i = 0; i < nb_samples; i++)
259 buffer[ch][i] = (buffer[ch][i] << extra_bits) | extra_bits_buffer[ch][i];
260 }
261
263 int channels)
264 {
266 int has_size,
bps, is_compressed, decorr_shift, decorr_left_weight,
ret;
267 uint32_t output_samples;
268 int i, ch;
269
270 skip_bits(&alac->
gb, 4);
/* element instance tag */
272
273 /* the number of output samples is stored in the frame */
275
281 }
282
283 /* whether the frame is compressed */
285
286 if (has_size)
288 else
292 output_samples);
294 }
297 /* get output buffer */
301 }
else if (output_samples != alac->
nb_samples) {
305 }
308 for (ch = 0; ch < channels; ch++)
310 }
311
312 if (is_compressed) {
313 int16_t lpc_coefs[2][32];
314 int lpc_order[2];
315 int prediction_type[2];
316 int lpc_quant[2];
317 int rice_history_mult[2];
318
321 "Compression with rice limit 0");
323 }
324
327
328 for (ch = 0; ch < channels; ch++) {
331 rice_history_mult[ch] =
get_bits(&alac->
gb, 3);
333
336
337 /* read the predictor table */
338 for (i = lpc_order[ch] - 1; i >= 0; i--)
340 }
341
345 return -1;
346 for (ch = 0; ch < channels; ch++)
348 }
349 }
350 for (ch = 0; ch < channels; ch++) {
354 if(ret<0)
356
357 /* adaptive FIR filter */
358 if (prediction_type[ch] == 15) {
359 /* Prediction type 15 runs the adaptive FIR twice.
360 * The first pass uses the special-case coef_num = 31, while
361 * the second pass uses the coefs from the bitstream.
362 *
363 * However, this prediction type is not currently used by the
364 * reference encoder.
365 */
369 } else if (prediction_type[ch] > 0) {
371 prediction_type[ch]);
372 }
375 bps, lpc_coefs[ch], lpc_order[ch], lpc_quant[ch]);
376 }
377 } else {
378 /* not compressed, easy case */
381 return -1;
382 for (ch = 0; ch < channels; ch++) {
385 }
386 }
388 decorr_shift = 0;
389 decorr_left_weight = 0;
390 }
391
395 }
396
397 if (channels == 2 && decorr_left_weight) {
399 decorr_shift, decorr_left_weight);
400 }
401
405 }
406
409 case 16: {
410 for (ch = 0; ch < channels; ch++) {
411 int16_t *outbuffer = (int16_t *)frame->
extended_data[ch_index + ch];
414 }}
415 break;
416 case 24: {
417 for (ch = 0; ch < channels; ch++) {
420 }}
421 break;
422 }
423 }else{
425 case 16: {
426 int16_t *outbuffer = ((int16_t *)frame->
extended_data[0]) + ch_index;
428 for (ch = 0; ch < channels; ch++)
430 outbuffer += alac->
channels - channels;
431 }
432 }
433 break;
434 case 24: {
437 for (ch = 0; ch < channels; ch++)
439 outbuffer += alac->
channels - channels;
440 }
441 }
442 break;
443 case 32: {
446 for (ch = 0; ch < channels; ch++)
448 outbuffer += alac->
channels - channels;
449 }
450 }
451 break;
452 }
453 }
454
455 return 0;
456 }
457
459 int *got_frame_ptr,
AVPacket *avpkt)
460 {
464 int channels;
465 int ch,
ret, got_end;
466
469
470 got_end = 0;
472 ch = 0;
476 got_end = 1;
477 break;
478 }
482 }
483
484 channels = (element ==
TYPE_CPE) ? 2 : 1;
485 if (ch + channels > alac->
channels ||
489 }
490
493 channels);
496
497 ch += channels;
498 }
499 if (!got_end) {
502 }
503
507 }
508
510 *got_frame_ptr = 1;
511 else
513
515 }
516
518 {
520
521 int ch;
527 }
528
529 return 0;
530 }
531
533 {
534 int ch;
536
539 buf_size, buf_alloc_fail);
540
544 buf_size, buf_alloc_fail);
545 }
546
548 buf_size, buf_alloc_fail);
549 }
550 return 0;
551 buf_alloc_fail:
554 }
555
557 {
559
562
564
569 "max samples per frame invalid: %"PRIu32"\n",
572 }
577 alac->
rice_limit = bytestream2_get_byteu(&gb);
578 alac->
channels = bytestream2_get_byteu(&gb);
579 bytestream2_get_be16u(&gb); // maxRun
580 bytestream2_get_be32u(&gb); // max coded frame size
581 bytestream2_get_be32u(&gb); // average bitrate
582 bytestream2_get_be32u(&gb); // samplerate
583
584 return 0;
585 }
586
588 {
590 int req_packed;
593
594 /* initialize from the extradata */
598 }
601 return -1;
602 }
603
607 break;
608 case 24:
610 break;
613 }
615
619 } else {
622 else
624 }
629 }
631
635 }
636
637 return 0;
638 }
639
641 {
645 }
646
648 { "extra_bits_bug", "Force non-standard decoding process",
652 };
653
659 };
660
672 .priv_class = &alac_class
673 };
const char const char void * val
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
int nb_samples
number of samples in the current frame
This structure describes decoded (raw) audio or video data.
#define ALAC_EXTRADATA_SIZE
ptrdiff_t const GLvoid * data
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
#define AV_LOG_WARNING
Something somehow does not look correct.
static int decode_element(AVCodecContext *avctx, AVFrame *frame, int ch_index, int channels)
#define LIBAVUTIL_VERSION_INT
static av_cold int init(AVCodecContext *avctx)
#define AV_OPT_FLAG_AUDIO_PARAM
static av_always_inline void bytestream2_init(GetByteContext *g, const uint8_t *buf, int buf_size)
static const AVOption options[]
int bits_per_raw_sample
Bits per sample/pixel of internal libavcodec pixel/sample format.
int32_t * extra_bits_buffer[2]
static int get_sbits(GetBitContext *s, int n)
static int get_sbits_long(GetBitContext *s, int n)
Read 0-32 bits as a signed integer.
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
int32_t * predict_error_buffer[2]
static int get_unary_0_9(GetBitContext *gb)
#define LIBAVCODEC_VERSION_MAJOR
void void avpriv_request_sample(void *avc, const char *msg,...) av_printf_format(2
Log a generic warning message about a missing feature.
enum AVSampleFormat sample_fmt
audio sample format
uint8_t rice_initial_history
static av_cold int alac_decode_close(AVCodecContext *avctx)
Multithreading support functions.
uint8_t * extradata
some codecs need / can use extradata like Huffman tables.
#define CODEC_CAP_DR1
Codec uses get_buffer() for allocating buffers and supports custom allocators.
static int get_bits_count(const GetBitContext *s)
static av_always_inline void bytestream2_skipu(GetByteContext *g, unsigned int size)
static const AVClass alac_class
bitstream reader API header.
int32_t * output_samples_buffer[2]
static int alac_decode_frame(AVCodecContext *avctx, void *data, int *got_frame_ptr, AVPacket *avpkt)
int extra_bits
number of extra bits beyond 16-bit
int av_sample_fmt_is_planar(enum AVSampleFormat sample_fmt)
Check if the sample format is planar.
static int get_bits_left(GetBitContext *gb)
static void append_extra_bits(int32_t *buffer[2], int32_t *extra_bits_buffer[2], int extra_bits, int channels, int nb_samples)
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
static void lpc_prediction(int32_t *error_buffer, int32_t *buffer_out, int nb_samples, int bps, int16_t *lpc_coefs, int lpc_order, int lpc_quant)
enum AVSampleFormat request_sample_fmt
desired sample format
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
static int sign_only(int v)
const char * name
Name of the codec implementation.
Libavcodec external API header.
uint64_t channel_layout
Audio channel layout.
#define ALAC_MAX_CHANNELS
uint32_t max_samples_per_frame
#define ONLY_IF_THREADS_ENABLED(x)
Define a function with only the non-default version specified.
static void decorrelate_stereo(int32_t *buffer[2], int nb_samples, int decorr_shift, int decorr_left_weight)
audio channel layout utility functions
static unsigned int show_bits(GetBitContext *s, int n)
Show 1-25 bits.
static int alac_set_info(ALACContext *alac)
uint8_t rice_history_mult
const uint64_t ff_alac_channel_layouts[ALAC_MAX_CHANNELS+1]
static const float pred[4]
#define AVERROR_PATCHWELCOME
Not yet implemented in FFmpeg, patches welcome.
static int init_get_bits8(GetBitContext *s, const uint8_t *buffer, int byte_size)
Initialize GetBitContext.
int ff_thread_get_buffer(AVCodecContext *avctx, ThreadFrame *f, int flags)
Wrapper around get_buffer() for frame-multithreaded codecs.
main external API structure.
static unsigned int get_bits1(GetBitContext *s)
Describe the class of an AVClass context structure.
static void skip_bits(GetBitContext *s, int n)
static unsigned int decode_scalar(GetBitContext *gb, int k, int bps)
#define AV_OPT_FLAG_DECODING_PARAM
a generic parameter which can be set by the user for demuxing or decoding
static unsigned int get_bits_long(GetBitContext *s, int n)
Read 0-32 bits.
const uint8_t ff_alac_channel_layout_offsets[ALAC_MAX_CHANNELS][ALAC_MAX_CHANNELS]
static av_const int sign_extend(int val, unsigned bits)
static int decode(AVCodecContext *avctx, void *data, int *got_sub, AVPacket *avpkt)
common internal api header.
static av_cold int alac_decode_init(AVCodecContext *avctx)
#define CODEC_CAP_FRAME_THREADS
Codec supports frame-level multithreading.
#define FF_ALLOC_OR_GOTO(ctx, p, size, label)
static int allocate_buffers(ALACContext *alac)
int channels
number of audio channels
static int rice_decompress(ALACContext *alac, int32_t *output_buffer, int nb_samples, int bps, int rice_history_mult)
static int init_thread_copy(AVCodecContext *avctx)
uint8_t ** extended_data
pointers to the data planes/channels.
This structure stores compressed data.
int nb_samples
number of audio samples (per channel) described by this frame