2 * TTA (The Lossless True Audio) decoder
3 * Copyright (c) 2006 Alex Beregszaszi
5 * This file is part of FFmpeg.
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.
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.
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
24 * TTA (The Lossless True Audio) decoder
25 * @see http://www.true-audio.com/
26 * @see http://tta.corecodec.org/
27 * @author Alex Beregszaszi
30 #define BITSTREAM_READER_LE
35 #include "libavutil/crc.h"
36 #include "libavutil/intreadwrite.h"
37 #include "libavutil/opt.h"
39 #define FORMAT_SIMPLE 1
40 #define FORMAT_ENCRYPTED 2
43 typedef struct TTAFilter {
44 int32_t shift, round, error;
45 int32_t qm[MAX_ORDER];
46 int32_t dx[MAX_ORDER];
47 int32_t dl[MAX_ORDER];
50 typedef struct TTARice {
51 uint32_t k0, k1, sum0, sum1;
54 typedef struct TTAChannel {
60 typedef struct TTAContext {
62 AVCodecContext *avctx;
64 const AVCRC *crc_table;
66 int format, channels, bps;
68 int frame_length, last_frame_length;
70 int32_t *decode_buffer;
77 static const uint32_t shift_1[] = {
78 0x00000001, 0x00000002, 0x00000004, 0x00000008,
79 0x00000010, 0x00000020, 0x00000040, 0x00000080,
80 0x00000100, 0x00000200, 0x00000400, 0x00000800,
81 0x00001000, 0x00002000, 0x00004000, 0x00008000,
82 0x00010000, 0x00020000, 0x00040000, 0x00080000,
83 0x00100000, 0x00200000, 0x00400000, 0x00800000,
84 0x01000000, 0x02000000, 0x04000000, 0x08000000,
85 0x10000000, 0x20000000, 0x40000000, 0x80000000,
86 0x80000000, 0x80000000, 0x80000000, 0x80000000,
87 0x80000000, 0x80000000, 0x80000000, 0x80000000
90 static const uint32_t * const shift_16 = shift_1 + 4;
92 static const int32_t ttafilter_configs[4] = {
99 static void ttafilter_init(TTAContext *s, TTAFilter *c, int32_t shift) {
100 memset(c, 0, sizeof(TTAFilter));
101 if (s->format == FORMAT_ENCRYPTED) {
103 for (i = 0; i < 8; i++)
104 c->qm[i] = sign_extend(s->crc_pass[i], 8);
107 c->round = shift_1[shift-1];
108 // c->round = 1 << (shift - 1);
111 static inline void ttafilter_process(TTAFilter *c, int32_t *in)
113 register int32_t *dl = c->dl, *qm = c->qm, *dx = c->dx, sum = c->round;
116 qm[0] -= dx[0]; qm[1] -= dx[1]; qm[2] -= dx[2]; qm[3] -= dx[3];
117 qm[4] -= dx[4]; qm[5] -= dx[5]; qm[6] -= dx[6]; qm[7] -= dx[7];
118 } else if (c->error > 0) {
119 qm[0] += dx[0]; qm[1] += dx[1]; qm[2] += dx[2]; qm[3] += dx[3];
120 qm[4] += dx[4]; qm[5] += dx[5]; qm[6] += dx[6]; qm[7] += dx[7];
123 sum += dl[0] * qm[0] + dl[1] * qm[1] + dl[2] * qm[2] + dl[3] * qm[3] +
124 dl[4] * qm[4] + dl[5] * qm[5] + dl[6] * qm[6] + dl[7] * qm[7];
126 dx[0] = dx[1]; dx[1] = dx[2]; dx[2] = dx[3]; dx[3] = dx[4];
127 dl[0] = dl[1]; dl[1] = dl[2]; dl[2] = dl[3]; dl[3] = dl[4];
129 dx[4] = ((dl[4] >> 30) | 1);
130 dx[5] = ((dl[5] >> 30) | 2) & ~1;
131 dx[6] = ((dl[6] >> 30) | 2) & ~1;
132 dx[7] = ((dl[7] >> 30) | 4) & ~3;
135 *in += (sum >> c->shift);
137 dl[4] = -dl[5]; dl[5] = -dl[6];
138 dl[6] = *in - dl[7]; dl[7] = *in;
139 dl[5] += dl[6]; dl[4] += dl[5];
142 static void rice_init(TTARice *c, uint32_t k0, uint32_t k1)
146 c->sum0 = shift_16[k0];
147 c->sum1 = shift_16[k1];
150 static int tta_get_unary(GetBitContext *gb)
155 while (get_bits_left(gb) > 0 && get_bits1(gb))
160 static const int64_t tta_channel_layouts[7] = {
162 AV_CH_LAYOUT_STEREO|AV_CH_LOW_FREQUENCY,
165 AV_CH_LAYOUT_5POINT1_BACK,
166 AV_CH_LAYOUT_5POINT1_BACK|AV_CH_BACK_CENTER,
167 AV_CH_LAYOUT_7POINT1_WIDE
170 static int tta_check_crc(TTAContext *s, const uint8_t *buf, int buf_size)
174 CRC = AV_RL32(buf + buf_size);
175 crc = av_crc(s->crc_table, 0xFFFFFFFFU, buf, buf_size);
176 if (CRC != (crc ^ 0xFFFFFFFFU)) {
177 av_log(s->avctx, AV_LOG_ERROR, "CRC error\n");
178 return AVERROR_INVALIDDATA;
184 static uint64_t tta_check_crc64(uint8_t *pass)
186 uint64_t crc = UINT64_MAX, poly = 0x42F0E1EBA9EA3693U;
187 uint8_t *end = pass + strlen(pass);
191 crc ^= (uint64_t)*pass++ << 56;
192 for (i = 0; i < 8; i++)
193 crc = (crc << 1) ^ (poly & (((int64_t) crc) >> 63));
196 return crc ^ UINT64_MAX;
199 static av_cold int tta_decode_init(AVCodecContext * avctx)
201 TTAContext *s = avctx->priv_data;
206 // 30bytes includes a seektable with one frame
207 if (avctx->extradata_size < 30)
208 return AVERROR_INVALIDDATA;
210 init_get_bits(&s->gb, avctx->extradata, avctx->extradata_size * 8);
211 if (show_bits_long(&s->gb, 32) == AV_RL32("TTA1"))
213 if (avctx->err_recognition & AV_EF_CRCCHECK) {
214 s->crc_table = av_crc_get_table(AV_CRC_32_IEEE_LE);
215 tta_check_crc(s, avctx->extradata, 18);
219 skip_bits_long(&s->gb, 32);
221 s->format = get_bits(&s->gb, 16);
223 av_log(avctx, AV_LOG_ERROR, "Invalid format\n");
224 return AVERROR_INVALIDDATA;
226 if (s->format == FORMAT_ENCRYPTED) {
228 av_log(avctx, AV_LOG_ERROR, "Missing password for encrypted stream. Please use the -password option\n");
229 return AVERROR(EINVAL);
231 AV_WL64(s->crc_pass, tta_check_crc64(s->pass));
233 avctx->channels = s->channels = get_bits(&s->gb, 16);
234 if (s->channels > 1 && s->channels < 9)
235 avctx->channel_layout = tta_channel_layouts[s->channels-2];
236 avctx->bits_per_raw_sample = get_bits(&s->gb, 16);
237 s->bps = (avctx->bits_per_raw_sample + 7) / 8;
238 avctx->sample_rate = get_bits_long(&s->gb, 32);
239 s->data_length = get_bits_long(&s->gb, 32);
240 skip_bits_long(&s->gb, 32); // CRC32 of header
242 if (s->channels == 0) {
243 av_log(avctx, AV_LOG_ERROR, "Invalid number of channels\n");
244 return AVERROR_INVALIDDATA;
245 } else if (avctx->sample_rate == 0) {
246 av_log(avctx, AV_LOG_ERROR, "Invalid samplerate\n");
247 return AVERROR_INVALIDDATA;
251 case 1: avctx->sample_fmt = AV_SAMPLE_FMT_U8; break;
253 avctx->sample_fmt = AV_SAMPLE_FMT_S16;
256 avctx->sample_fmt = AV_SAMPLE_FMT_S32;
258 //case 4: avctx->sample_fmt = AV_SAMPLE_FMT_S32; break;
260 av_log(avctx, AV_LOG_ERROR, "Invalid/unsupported sample format.\n");
261 return AVERROR_INVALIDDATA;
265 if (avctx->sample_rate > 0x7FFFFFu) {
266 av_log(avctx, AV_LOG_ERROR, "sample_rate too large\n");
267 return AVERROR(EINVAL);
269 s->frame_length = 256 * avctx->sample_rate / 245;
271 s->last_frame_length = s->data_length % s->frame_length;
272 total_frames = s->data_length / s->frame_length +
273 (s->last_frame_length ? 1 : 0);
275 av_log(avctx, AV_LOG_DEBUG, "format: %d chans: %d bps: %d rate: %d block: %d\n",
276 s->format, avctx->channels, avctx->bits_per_coded_sample, avctx->sample_rate,
278 av_log(avctx, AV_LOG_DEBUG, "data_length: %d frame_length: %d last: %d total: %d\n",
279 s->data_length, s->frame_length, s->last_frame_length, total_frames);
282 if (avctx->extradata_size <= 26 || total_frames > INT_MAX / 4 ||
283 avctx->extradata_size - 26 < total_frames * 4)
284 av_log(avctx, AV_LOG_WARNING, "Seek table missing or too small\n");
285 else if (avctx->err_recognition & AV_EF_CRCCHECK) {
286 if (tta_check_crc(s, avctx->extradata + 22, total_frames * 4))
287 return AVERROR_INVALIDDATA;
289 skip_bits_long(&s->gb, 32 * total_frames);
290 skip_bits_long(&s->gb, 32); // CRC32 of seektable
292 if(s->frame_length >= UINT_MAX / (s->channels * sizeof(int32_t))){
293 av_log(avctx, AV_LOG_ERROR, "frame_length too large\n");
294 return AVERROR_INVALIDDATA;
298 s->decode_buffer = av_mallocz(sizeof(int32_t)*s->frame_length*s->channels);
299 if (!s->decode_buffer)
300 return AVERROR(ENOMEM);
302 s->decode_buffer = NULL;
303 s->ch_ctx = av_malloc(avctx->channels * sizeof(*s->ch_ctx));
305 av_freep(&s->decode_buffer);
306 return AVERROR(ENOMEM);
309 av_log(avctx, AV_LOG_ERROR, "Wrong extradata present\n");
310 return AVERROR_INVALIDDATA;
316 static int tta_decode_frame(AVCodecContext *avctx, void *data,
317 int *got_frame_ptr, AVPacket *avpkt)
319 AVFrame *frame = data;
320 const uint8_t *buf = avpkt->data;
321 int buf_size = avpkt->size;
322 TTAContext *s = avctx->priv_data;
324 int cur_chan = 0, framelen = s->frame_length;
327 if (avctx->err_recognition & AV_EF_CRCCHECK) {
328 if (buf_size < 4 || tta_check_crc(s, buf, buf_size - 4))
329 return AVERROR_INVALIDDATA;
332 init_get_bits(&s->gb, buf, buf_size*8);
334 /* get output buffer */
335 frame->nb_samples = framelen;
336 if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
339 // decode directly to output buffer for 24-bit sample format
341 s->decode_buffer = (int32_t *)frame->data[0];
343 // init per channel states
344 for (i = 0; i < s->channels; i++) {
345 s->ch_ctx[i].predictor = 0;
346 ttafilter_init(s, &s->ch_ctx[i].filter, ttafilter_configs[s->bps-1]);
347 rice_init(&s->ch_ctx[i].rice, 10, 10);
351 for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++) {
352 int32_t *predictor = &s->ch_ctx[cur_chan].predictor;
353 TTAFilter *filter = &s->ch_ctx[cur_chan].filter;
354 TTARice *rice = &s->ch_ctx[cur_chan].rice;
355 uint32_t unary, depth, k;
358 unary = tta_get_unary(&s->gb);
369 if (get_bits_left(&s->gb) < k) {
370 ret = AVERROR_INVALIDDATA;
375 if (k > MIN_CACHE_BITS) {
376 ret = AVERROR_INVALIDDATA;
379 value = (unary << k) + get_bits(&s->gb, k);
383 // FIXME: copy paste from original
386 rice->sum1 += value - (rice->sum1 >> 4);
387 if (rice->k1 > 0 && rice->sum1 < shift_16[rice->k1])
389 else if(rice->sum1 > shift_16[rice->k1 + 1])
391 value += shift_1[rice->k0];
393 rice->sum0 += value - (rice->sum0 >> 4);
394 if (rice->k0 > 0 && rice->sum0 < shift_16[rice->k0])
396 else if(rice->sum0 > shift_16[rice->k0 + 1])
400 // extract coded value
401 *p = 1 + ((value >> 1) ^ ((value & 1) - 1));
404 ttafilter_process(filter, p);
406 // fixed order prediction
407 #define PRED(x, k) (int32_t)((((uint64_t)x << k) - x) >> k)
409 case 1: *p += PRED(*predictor, 4); break;
411 case 3: *p += PRED(*predictor, 5); break;
412 case 4: *p += *predictor; break;
417 if (cur_chan < (s->channels-1))
420 // decorrelate in case of multiple channels
421 if (s->channels > 1) {
423 for (*p += *r / 2; r > p - s->channels; r--)
428 // check for last frame
429 if (i == s->last_frame_length && get_bits_left(&s->gb) / 8 == 4) {
430 frame->nb_samples = framelen = s->last_frame_length;
436 align_get_bits(&s->gb);
437 if (get_bits_left(&s->gb) < 32) {
438 ret = AVERROR_INVALIDDATA;
441 skip_bits_long(&s->gb, 32); // frame crc
443 // convert to output buffer
446 uint8_t *samples = (uint8_t *)frame->data[0];
447 for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++)
448 *samples++ = *p + 0x80;
452 int16_t *samples = (int16_t *)frame->data[0];
453 for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++)
458 // shift samples for 24-bit sample format
459 int32_t *samples = (int32_t *)frame->data[0];
460 for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++)
462 // reset decode buffer
463 s->decode_buffer = NULL;
472 // reset decode buffer
474 s->decode_buffer = NULL;
478 static av_cold int tta_decode_close(AVCodecContext *avctx) {
479 TTAContext *s = avctx->priv_data;
482 av_free(s->decode_buffer);
483 s->decode_buffer = NULL;
484 av_freep(&s->ch_ctx);
489 #define OFFSET(x) offsetof(TTAContext, x)
490 #define DEC (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
491 static const AVOption options[] = {
492 { "password", "Set decoding password", OFFSET(pass), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, DEC },
496 static const AVClass tta_decoder_class = {
497 .class_name = "TTA Decoder",
498 .item_name = av_default_item_name,
500 .version = LIBAVUTIL_VERSION_INT,
503 AVCodec ff_tta_decoder = {
505 .type = AVMEDIA_TYPE_AUDIO,
506 .id = AV_CODEC_ID_TTA,
507 .priv_data_size = sizeof(TTAContext),
508 .init = tta_decode_init,
509 .close = tta_decode_close,
510 .decode = tta_decode_frame,
511 .capabilities = CODEC_CAP_DR1,
512 .long_name = NULL_IF_CONFIG_SMALL("TTA (True Audio)"),
513 .priv_class = &tta_decoder_class,