tta.c 13.3 KB
Newer Older
1 2 3 4
/*
 * TTA (The Lossless True Audio) decoder
 * Copyright (c) 2006 Alex Beregszaszi
 *
5 6 7
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
8 9
 * modify it under the terms of the GNU Lesser General Public
 * 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 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * 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 libavcodec/tta.c
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
 * TTA (The Lossless True Audio) decoder
 * (www.true-audio.com or tta.corecodec.org)
 * @author Alex Beregszaszi
 *
 */

#define ALT_BITSTREAM_READER_LE
//#define DEBUG
#include <limits.h>
#include "avcodec.h"
#include "bitstream.h"

#define FORMAT_INT 1
#define FORMAT_FLOAT 3

typedef struct TTAContext {
    AVCodecContext *avctx;
    GetBitContext gb;

    int flags, channels, bps, is_float, data_length;
    int frame_length, last_frame_length, total_frames;

46
    int32_t *decode_buffer;
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
} TTAContext;

#if 0
static inline int shift_1(int i)
{
    if (i < 32)
        return 1 << i;
    else
        return 0x80000000; // 16 << 31
}

static inline int shift_16(int i)
{
    if (i < 28)
        return 16 << i;
    else
        return 0x80000000; // 16 << 27
}
#else
66
static const uint32_t shift_1[] = {
67 68 69 70 71 72 73 74 75 76 77 78
    0x00000001, 0x00000002, 0x00000004, 0x00000008,
    0x00000010, 0x00000020, 0x00000040, 0x00000080,
    0x00000100, 0x00000200, 0x00000400, 0x00000800,
    0x00001000, 0x00002000, 0x00004000, 0x00008000,
    0x00010000, 0x00020000, 0x00040000, 0x00080000,
    0x00100000, 0x00200000, 0x00400000, 0x00800000,
    0x01000000, 0x02000000, 0x04000000, 0x08000000,
    0x10000000, 0x20000000, 0x40000000, 0x80000000,
    0x80000000, 0x80000000, 0x80000000, 0x80000000,
    0x80000000, 0x80000000, 0x80000000, 0x80000000
};

79
static const uint32_t * const shift_16 = shift_1 + 4;
80 81 82 83
#endif

#define MAX_ORDER 16
typedef struct TTAFilter {
84 85 86 87
    int32_t shift, round, error, mode;
    int32_t qm[MAX_ORDER];
    int32_t dx[MAX_ORDER];
    int32_t dl[MAX_ORDER];
88 89
} TTAFilter;

90
static const int32_t ttafilter_configs[4][2] = {
91 92 93 94 95 96
    {10, 1},
    {9, 1},
    {10, 1},
    {12, 0}
};

97
static void ttafilter_init(TTAFilter *c, int32_t shift, int32_t mode) {
98 99 100 101 102 103 104 105
    memset(c, 0, sizeof(TTAFilter));
    c->shift = shift;
   c->round = shift_1[shift-1];
//    c->round = 1 << (shift - 1);
    c->mode = mode;
}

// FIXME: copy paste from original
106
static inline void memshl(register int32_t *a, register int32_t *b) {
107 108 109 110 111 112 113 114 115 116 117 118
    *a++ = *b++;
    *a++ = *b++;
    *a++ = *b++;
    *a++ = *b++;
    *a++ = *b++;
    *a++ = *b++;
    *a++ = *b++;
    *a = *b;
}

// FIXME: copy paste from original
// mode=1 encoder, mode=0 decoder
119 120
static inline void ttafilter_process(TTAFilter *c, int32_t *in, int32_t mode) {
    register int32_t *dl = c->dl, *qm = c->qm, *dx = c->dx, sum = c->round;
121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178

    if (!c->error) {
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        sum += *dl++ * *qm, qm++;
        dx += 8;
    } else if(c->error < 0) {
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
        sum += *dl++ * (*qm -= *dx++), qm++;
    } else {
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
        sum += *dl++ * (*qm += *dx++), qm++;
    }

    *(dx-0) = ((*(dl-1) >> 30) | 1) << 2;
    *(dx-1) = ((*(dl-2) >> 30) | 1) << 1;
    *(dx-2) = ((*(dl-3) >> 30) | 1) << 1;
    *(dx-3) = ((*(dl-4) >> 30) | 1);

    // compress
    if (mode) {
        *dl = *in;
        *in -= (sum >> c->shift);
        c->error = *in;
    } else {
        c->error = *in;
        *in += (sum >> c->shift);
        *dl = *in;
    }

    if (c->mode) {
        *(dl-1) = *dl - *(dl-1);
        *(dl-2) = *(dl-1) - *(dl-2);
        *(dl-3) = *(dl-2) - *(dl-3);
    }

    memshl(c->dl, c->dl + 1);
    memshl(c->dx, c->dx + 1);
}

typedef struct TTARice {
179
    uint32_t k0, k1, sum0, sum1;
180 181
} TTARice;

182
static void rice_init(TTARice *c, uint32_t k0, uint32_t k1)
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199
{
    c->k0 = k0;
    c->k1 = k1;
    c->sum0 = shift_16[k0];
    c->sum1 = shift_16[k1];
}

static int tta_get_unary(GetBitContext *gb)
{
    int ret = 0;

    // count ones
    while(get_bits1(gb))
        ret++;
    return ret;
}

200
static av_cold int tta_decode_init(AVCodecContext * avctx)
201 202 203 204 205 206 207 208 209 210 211
{
    TTAContext *s = avctx->priv_data;
    int i;

    s->avctx = avctx;

    // 30bytes includes a seektable with one frame
    if (avctx->extradata_size < 30)
        return -1;

    init_get_bits(&s->gb, avctx->extradata, avctx->extradata_size);
212
    if (show_bits_long(&s->gb, 32) == AV_RL32("TTA1"))
213 214 215
    {
        /* signature */
        skip_bits(&s->gb, 32);
216
//        if (get_bits_long(&s->gb, 32) != bswap_32(AV_RL32("TTA1"))) {
217 218 219 220
//            av_log(s->avctx, AV_LOG_ERROR, "Missing magic\n");
//            return -1;
//        }

221
        s->flags = get_bits(&s->gb, 16);
222 223 224 225 226 227
        if (s->flags != 1 && s->flags != 3)
        {
            av_log(s->avctx, AV_LOG_ERROR, "Invalid flags\n");
            return -1;
        }
        s->is_float = (s->flags == FORMAT_FLOAT);
228
        avctx->channels = s->channels = get_bits(&s->gb, 16);
229 230
        avctx->bits_per_coded_sample = get_bits(&s->gb, 16);
        s->bps = (avctx->bits_per_coded_sample + 7) / 8;
231
        avctx->sample_rate = get_bits_long(&s->gb, 32);
232 233 234 235
        if(avctx->sample_rate > 1000000){ //prevent FRAME_TIME * avctx->sample_rate from overflowing and sanity check
            av_log(avctx, AV_LOG_ERROR, "sample_rate too large\n");
            return -1;
        }
236
        s->data_length = get_bits_long(&s->gb, 32);
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
        skip_bits(&s->gb, 32); // CRC32 of header

        if (s->is_float)
        {
            avctx->sample_fmt = SAMPLE_FMT_FLT;
            av_log(s->avctx, AV_LOG_ERROR, "Unsupported sample format. Please contact the developers.\n");
            return -1;
        }
        else switch(s->bps) {
//            case 1: avctx->sample_fmt = SAMPLE_FMT_U8; break;
            case 2: avctx->sample_fmt = SAMPLE_FMT_S16; break;
//            case 3: avctx->sample_fmt = SAMPLE_FMT_S24; break;
            case 4: avctx->sample_fmt = SAMPLE_FMT_S32; break;
            default:
                av_log(s->avctx, AV_LOG_ERROR, "Invalid/unsupported sample format. Please contact the developers.\n");
                return -1;
        }

        // FIXME: horribly broken, but directly from reference source
#define FRAME_TIME 1.04489795918367346939
        s->frame_length = (int)(FRAME_TIME * avctx->sample_rate);

        s->last_frame_length = s->data_length % s->frame_length;
        s->total_frames = s->data_length / s->frame_length +
                        (s->last_frame_length ? 1 : 0);

        av_log(s->avctx, AV_LOG_DEBUG, "flags: %x chans: %d bps: %d rate: %d block: %d\n",
264
            s->flags, avctx->channels, avctx->bits_per_coded_sample, avctx->sample_rate,
265 266 267 268 269 270 271 272 273
            avctx->block_align);
        av_log(s->avctx, AV_LOG_DEBUG, "data_length: %d frame_length: %d last: %d total: %d\n",
            s->data_length, s->frame_length, s->last_frame_length, s->total_frames);

        // FIXME: seek table
        for (i = 0; i < s->total_frames; i++)
            skip_bits(&s->gb, 32);
        skip_bits(&s->gb, 32); // CRC32 of seektable

274 275 276 277 278
        if(s->frame_length >= UINT_MAX / (s->channels * sizeof(int32_t))){
            av_log(avctx, AV_LOG_ERROR, "frame_length too large\n");
            return -1;
        }

279
        s->decode_buffer = av_mallocz(sizeof(int32_t)*s->frame_length*s->channels);
280 281 282 283 284 285 286 287 288 289
    } else {
        av_log(avctx, AV_LOG_ERROR, "Wrong extradata present\n");
        return -1;
    }

    return 0;
}

static int tta_decode_frame(AVCodecContext *avctx,
        void *data, int *data_size,
Michael Niedermayer's avatar
Michael Niedermayer committed
290
        const uint8_t *buf, int buf_size)
291 292 293 294 295 296
{
    TTAContext *s = avctx->priv_data;
    int i;

    init_get_bits(&s->gb, buf, buf_size*8);
    {
297
        int32_t predictors[s->channels];
298 299 300
        TTAFilter filters[s->channels];
        TTARice rices[s->channels];
        int cur_chan = 0, framelen = s->frame_length;
301
        int32_t *p;
302 303 304 305 306 307 308 309 310 311 312 313 314 315

        // FIXME: seeking
        s->total_frames--;
        if (!s->total_frames && s->last_frame_length)
            framelen = s->last_frame_length;

        // init per channel states
        for (i = 0; i < s->channels; i++) {
            predictors[i] = 0;
            ttafilter_init(&(filters[i]), ttafilter_configs[s->bps-1][0], ttafilter_configs[s->bps-1][1]);
            rice_init(&(rices[i]), 10, 10);
        }

        for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++) {
316
            int32_t *predictor = &(predictors[cur_chan]);
317 318
            TTAFilter *filter = &(filters[cur_chan]);
            TTARice *rice = &(rices[cur_chan]);
319 320
            uint32_t unary, depth, k;
            int32_t value;
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354

            unary = tta_get_unary(&s->gb);

            if (unary == 0) {
                depth = 0;
                k = rice->k0;
            } else {
                depth = 1;
                k = rice->k1;
                unary--;
            }

            if (k)
                value = (unary << k) + get_bits(&s->gb, k);
            else
                value = unary;

            // FIXME: copy paste from original
            switch (depth) {
            case 1:
                rice->sum1 += value - (rice->sum1 >> 4);
                if (rice->k1 > 0 && rice->sum1 < shift_16[rice->k1])
                    rice->k1--;
                else if(rice->sum1 > shift_16[rice->k1 + 1])
                    rice->k1++;
                value += shift_1[rice->k0];
            default:
                rice->sum0 += value - (rice->sum0 >> 4);
                if (rice->k0 > 0 && rice->sum0 < shift_16[rice->k0])
                    rice->k0--;
                else if(rice->sum0 > shift_16[rice->k0 + 1])
                    rice->k0++;
            }

355 356 357
            // extract coded value
#define UNFOLD(x) (((x)&1) ? (++(x)>>1) : (-(x)>>1))
            *p = UNFOLD(value);
358 359 360 361 362

            // run hybrid filter
            ttafilter_process(filter, p, 0);

            // fixed order prediction
363
#define PRED(x, k) (int32_t)((((uint64_t)x << k) - x) >> k)
364 365 366 367 368 369 370 371 372 373 374
            switch (s->bps) {
                case 1: *p += PRED(*predictor, 4); break;
                case 2:
                case 3: *p += PRED(*predictor, 5); break;
                case 4: *p += *predictor; break;
            }
            *predictor = *p;

#if 0
            // extract 32bit float from last two int samples
            if (s->is_float && ((p - data) & 1)) {
375 376 377
                uint32_t neg = *p & 0x80000000;
                uint32_t hi = *(p - 1);
                uint32_t lo = abs(*p) - 1;
378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396

                hi += (hi || lo) ? 0x3f80 : 0;
                // SWAP16: swap all the 16 bits
                *(p - 1) = (hi << 16) | SWAP16(lo) | neg;
            }
#endif

            /*if ((get_bits_count(&s->gb)+7)/8 > buf_size)
            {
                av_log(NULL, AV_LOG_INFO, "overread!!\n");
                break;
            }*/

            // flip channels
            if (cur_chan < (s->channels-1))
                cur_chan++;
            else {
                // decorrelate in case of stereo integer
                if (!s->is_float && (s->channels > 1)) {
397
                    int32_t *r = p - 1;
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427
                    for (*p += *r / 2; r > p - s->channels; r--)
                        *r = *(r + 1) - *r;
                }
                cur_chan = 0;
            }
        }

        skip_bits(&s->gb, 32); // frame crc

        // convert to output buffer
        switch(s->bps) {
            case 2: {
                uint16_t *samples = data;
                for (p = s->decode_buffer; p < s->decode_buffer + (framelen * s->channels); p++) {
//                    *samples++ = (unsigned char)*p;
//                    *samples++ = (unsigned char)(*p >> 8);
                    *samples++ = *p;
                }
                *data_size = (uint8_t *)samples - (uint8_t *)data;
                break;
            }
            default:
                av_log(s->avctx, AV_LOG_ERROR, "Error, only 16bit samples supported!\n");
        }
    }

//    return get_bits_count(&s->gb)+7)/8;
    return buf_size;
}

428
static av_cold int tta_decode_close(AVCodecContext *avctx) {
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445
    TTAContext *s = avctx->priv_data;

    if (s->decode_buffer)
        av_free(s->decode_buffer);

    return 0;
}

AVCodec tta_decoder = {
    "tta",
    CODEC_TYPE_AUDIO,
    CODEC_ID_TTA,
    sizeof(TTAContext),
    tta_decode_init,
    NULL,
    tta_decode_close,
    tta_decode_frame,
446
    .long_name = NULL_IF_CONFIG_SMALL("True Audio"),
447
};