libavcodec/atrac1.c

   1 /*
   2  * Atrac 1 compatible decoder
   3  * Copyright (c) 2009 Maxim Poliakovski
   4  * Copyright (c) 2009 Benjamin Larsson
   5  *
   6  * This file is part of FFmpeg.
   7  *
   8  * FFmpeg is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * FFmpeg is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with FFmpeg; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22
  23 /**
  24  * @file libavcodec/atrac1.c
  25  * Atrac 1 compatible decoder.
  26  * This decoder handles raw ATRAC1 data and probably SDDS data.
  27  */
  28
  29 /* Many thanks to Tim Craig for all the help! */
  30
  31 #include <math.h>
  32 #include <stddef.h>
  33 #include <stdio.h>
  34
  35 #include "avcodec.h"
  36 #include "get_bits.h"
  37 #include "dsputil.h"
  38
  39 #include "atrac.h"
  40 #include "atrac1data.h"
  41
  42 #define AT1_MAX_BFU      52                 ///< max number of block floating units in a sound unit
  43 #define AT1_SU_SIZE      212                ///< number of bytes in a sound unit
  44 #define AT1_SU_SAMPLES   512                ///< number of samples in a sound unit
  45 #define AT1_FRAME_SIZE   AT1_SU_SIZE * 2
  46 #define AT1_SU_MAX_BITS  AT1_SU_SIZE * 8
  47 #define AT1_MAX_CHANNELS 2
  48
  49 #define AT1_QMF_BANDS    3
  50 #define IDX_LOW_BAND     0
  51 #define IDX_MID_BAND     1
  52 #define IDX_HIGH_BAND    2
  53
  54 /**
  55  * Sound unit struct, one unit is used per channel
  56  */
  57 typedef struct {
  58     int                 log2_block_count[AT1_QMF_BANDS];    ///< log2 number of blocks in a band
  59     int                 num_bfus;                           ///< number of Block Floating Units
  60     float*              spectrum[2];
  61     DECLARE_ALIGNED_16(float, spec1[AT1_SU_SAMPLES]);       ///< mdct buffer
  62     DECLARE_ALIGNED_16(float, spec2[AT1_SU_SAMPLES]);       ///< mdct buffer
  63     DECLARE_ALIGNED_16(float, fst_qmf_delay[46]);           ///< delay line for the 1st stacked QMF filter
  64     DECLARE_ALIGNED_16(float, snd_qmf_delay[46]);           ///< delay line for the 2nd stacked QMF filter
  65     DECLARE_ALIGNED_16(float, last_qmf_delay[256+23]);      ///< delay line for the last stacked QMF filter
  66 } AT1SUCtx;
  67
  68 /**
  69  * The atrac1 context, holds all needed parameters for decoding
  70  */
  71 typedef struct {
  72     AT1SUCtx            SUs[AT1_MAX_CHANNELS];              ///< channel sound unit
  73     DECLARE_ALIGNED_16(float, spec[AT1_SU_SAMPLES]);        ///< the mdct spectrum buffer
  74
  75     DECLARE_ALIGNED_16(float,  low[256]);
  76     DECLARE_ALIGNED_16(float,  mid[256]);
  77     DECLARE_ALIGNED_16(float, high[512]);
  78     float*              bands[3];
  79     DECLARE_ALIGNED_16(float, out_samples[AT1_MAX_CHANNELS][AT1_SU_SAMPLES]);
  80     FFTContext          mdct_ctx[3];
  81     int                 channels;
  82     DSPContext          dsp;
  83 } AT1Ctx;
  84
  85 /** size of the transform in samples in the long mode for each QMF band */
  86 static const uint16_t samples_per_band[3] = {128, 128, 256};
  87 static const uint8_t   mdct_long_nbits[3] = {7, 7, 8};
  88
  89
  90 static void at1_imdct(AT1Ctx *q, float *spec, float *out, int nbits,
  91                       int rev_spec)
  92 {
  93     FFTContext* mdct_context = &q->mdct_ctx[nbits - 5 - (nbits > 6)];
  94     int transf_size = 1 << nbits;
  95
  96     if (rev_spec) {
  97         int i;
  98         for (i = 0; i < transf_size / 2; i++)
  99             FFSWAP(float, spec[i], spec[transf_size - 1 - i]);
 100     }
 101     ff_imdct_half(mdct_context, out, spec);
 102 }
 103
 104
 105 static int at1_imdct_block(AT1SUCtx* su, AT1Ctx *q)
 106 {
 107     int          band_num, band_samples, log2_block_count, nbits, num_blocks, block_size;
 108     unsigned int start_pos, ref_pos = 0, pos = 0;
 109
 110     for (band_num = 0; band_num < AT1_QMF_BANDS; band_num++) {
 111         float *prev_buf;
 112         int j;
 113
 114         band_samples = samples_per_band[band_num];
 115         log2_block_count = su->log2_block_count[band_num];
 116
 117         /* number of mdct blocks in the current QMF band: 1 - for long mode */
 118         /* 4 for short mode(low/middle bands) and 8 for short mode(high band)*/
 119         num_blocks = 1 << log2_block_count;
 120
 121         if (num_blocks == 1) {
 122             /* mdct block size in samples: 128 (long mode, low & mid bands), */
 123             /* 256 (long mode, high band) and 32 (short mode, all bands) */
 124             block_size = band_samples >> log2_block_count;
 125
 126             /* calc transform size in bits according to the block_size_mode */
 127             nbits = mdct_long_nbits[band_num] - log2_block_count;
 128
 129             if (nbits != 5 && nbits != 7 && nbits != 8)
 130                 return -1;
 131         } else {
 132             block_size = 32;
 133             nbits = 5;
 134         }
 135
 136         start_pos = 0;
 137         prev_buf = &su->spectrum[1][ref_pos + band_samples - 16];
 138         for (j=0; j < num_blocks; j++) {
 139             at1_imdct(q, &q->spec[pos], &su->spectrum[0][ref_pos + start_pos], nbits, band_num);
 140
 141             /* overlap and window */
 142             q->dsp.vector_fmul_window(&q->bands[band_num][start_pos], prev_buf,
 143                                       &su->spectrum[0][ref_pos + start_pos], ff_sine_32, 0, 16);
 144
 145             prev_buf = &su->spectrum[0][ref_pos+start_pos + 16];
 146             start_pos += block_size;
 147             pos += block_size;
 148         }
 149
 150         if (num_blocks == 1)
 151             memcpy(q->bands[band_num] + 32, &su->spectrum[0][ref_pos + 16], 240 * sizeof(float));
 152
 153         ref_pos += band_samples;
 154     }
 155
 156     /* Swap buffers so the mdct overlap works */
 157     FFSWAP(float*, su->spectrum[0], su->spectrum[1]);
 158
 159     return 0;
 160 }
 161
 162 /**
 163  * Parse the block size mode byte
 164  */
 165
 166 static int at1_parse_bsm(GetBitContext* gb, int log2_block_cnt[AT1_QMF_BANDS])
 167 {
 168     int log2_block_count_tmp, i;
 169
 170     for (i = 0; i < 2; i++) {
 171         /* low and mid band */
 172         log2_block_count_tmp = get_bits(gb, 2);
 173         if (log2_block_count_tmp & 1)
 174             return -1;
 175         log2_block_cnt[i] = 2 - log2_block_count_tmp;
 176     }
 177
 178     /* high band */
 179     log2_block_count_tmp = get_bits(gb, 2);
 180     if (log2_block_count_tmp != 0 && log2_block_count_tmp != 3)
 181         return -1;
 182     log2_block_cnt[IDX_HIGH_BAND] = 3 - log2_block_count_tmp;
 183
 184     skip_bits(gb, 2);
 185     return 0;
 186 }
 187
 188
 189 static int at1_unpack_dequant(GetBitContext* gb, AT1SUCtx* su,
 190                               float spec[AT1_SU_SAMPLES])
 191 {
 192     int bits_used, band_num, bfu_num, i;
 193     uint8_t idwls[AT1_MAX_BFU];                 ///< the word length indexes for each BFU
 194     uint8_t idsfs[AT1_MAX_BFU];                 ///< the scalefactor indexes for each BFU
 195
 196     /* parse the info byte (2nd byte) telling how much BFUs were coded */
 197     su->num_bfus = bfu_amount_tab1[get_bits(gb, 3)];
 198
 199     /* calc number of consumed bits:
 200         num_BFUs * (idwl(4bits) + idsf(6bits)) + log2_block_count(8bits) + info_byte(8bits)
 201         + info_byte_copy(8bits) + log2_block_count_copy(8bits) */
 202     bits_used = su->num_bfus * 10 + 32 +
 203                 bfu_amount_tab2[get_bits(gb, 2)] +
 204                 (bfu_amount_tab3[get_bits(gb, 3)] << 1);
 205
 206     /* get word length index (idwl) for each BFU */
 207     for (i = 0; i < su->num_bfus; i++)
 208         idwls[i] = get_bits(gb, 4);
 209
 210     /* get scalefactor index (idsf) for each BFU */
 211     for (i = 0; i < su->num_bfus; i++)
 212         idsfs[i] = get_bits(gb, 6);
 213
 214     /* zero idwl/idsf for empty BFUs */
 215     for (i = su->num_bfus; i < AT1_MAX_BFU; i++)
 216         idwls[i] = idsfs[i] = 0;
 217
 218     /* read in the spectral data and reconstruct MDCT spectrum of this channel */
 219     for (band_num = 0; band_num < AT1_QMF_BANDS; band_num++) {
 220         for (bfu_num = bfu_bands_t[band_num]; bfu_num < bfu_bands_t[band_num+1]; bfu_num++) {
 221             int pos;
 222
 223             int num_specs = specs_per_bfu[bfu_num];
 224             int word_len  = !!idwls[bfu_num] + idwls[bfu_num];
 225             float scale_factor = sf_table[idsfs[bfu_num]];
 226             bits_used += word_len * num_specs; /* add number of bits consumed by current BFU */
 227
 228             /* check for bitstream overflow */
 229             if (bits_used > AT1_SU_MAX_BITS)
 230                 return -1;
 231
 232             /* get the position of the 1st spec according to the block size mode */
 233             pos = su->log2_block_count[band_num] ? bfu_start_short[bfu_num] : bfu_start_long[bfu_num];
 234
 235             if (word_len) {
 236                 float   max_quant = 1.0 / (float)((1 << (word_len - 1)) - 1);
 237
 238                 for (i = 0; i < num_specs; i++) {
 239                     /* read in a quantized spec and convert it to
 240                      * signed int and then inverse quantization
 241                      */
 242                     spec[pos+i] = get_sbits(gb, word_len) * scale_factor * max_quant;
 243                 }
 244             } else { /* word_len = 0 -> empty BFU, zero all specs in the emty BFU */
 245                 memset(&spec[pos], 0, num_specs * sizeof(float));
 246             }
 247         }
 248     }
 249
 250     return 0;
 251 }
 252
 253
 254 void at1_subband_synthesis(AT1Ctx *q, AT1SUCtx* su, float *pOut)
 255 {
 256     float temp[256];
 257     float iqmf_temp[512 + 46];
 258
 259     /* combine low and middle bands */
 260     atrac_iqmf(q->bands[0], q->bands[1], 128, temp, su->fst_qmf_delay, iqmf_temp);
 261
 262     /* delay the signal of the high band by 23 samples */
 263     memcpy( su->last_qmf_delay,    &su->last_qmf_delay[256], sizeof(float) *  23);
 264     memcpy(&su->last_qmf_delay[23], q->bands[2],             sizeof(float) * 256);
 265
 266     /* combine (low + middle) and high bands */
 267     atrac_iqmf(temp, su->last_qmf_delay, 256, pOut, su->snd_qmf_delay, iqmf_temp);
 268 }
 269
 270
 271 static int atrac1_decode_frame(AVCodecContext *avctx, void *data,
 272                                int *data_size, AVPacket *avpkt)
 273 {
 274     const uint8_t *buf = avpkt->data;
 275     int buf_size       = avpkt->size;
 276     AT1Ctx *q          = avctx->priv_data;
 277     int ch, ret, i;
 278     GetBitContext gb;
 279     float* samples = data;
 280
 281
 282     if (buf_size < 212 * q->channels) {
 283         av_log(q,AV_LOG_ERROR,"Not enought data to decode!\n");
 284         return -1;
 285     }
 286
 287     for (ch = 0; ch < q->channels; ch++) {
 288         AT1SUCtx* su = &q->SUs[ch];
 289
 290         init_get_bits(&gb, &buf[212 * ch], 212 * 8);
 291
 292         /* parse block_size_mode, 1st byte */
 293         ret = at1_parse_bsm(&gb, su->log2_block_count);
 294         if (ret < 0)
 295             return ret;
 296
 297         ret = at1_unpack_dequant(&gb, su, q->spec);
 298         if (ret < 0)
 299             return ret;
 300
 301         ret = at1_imdct_block(su, q);
 302         if (ret < 0)
 303             return ret;
 304         at1_subband_synthesis(q, su, q->out_samples[ch]);
 305     }
 306
 307     /* round, convert to 16bit and interleave */
 308     if (q->channels == 1) {
 309         /* mono */
 310         q->dsp.vector_clipf(samples, q->out_samples[0], -32700.0 / (1 << 15),
 311                             32700.0 / (1 << 15), AT1_SU_SAMPLES);
 312     } else {
 313         /* stereo */
 314         for (i = 0; i < AT1_SU_SAMPLES; i++) {
 315             samples[i * 2]     = av_clipf(q->out_samples[0][i],
 316                                           -32700.0 / (1 << 15),
 317                                            32700.0 / (1 << 15));
 318             samples[i * 2 + 1] = av_clipf(q->out_samples[1][i],
 319                                           -32700.0 / (1 << 15),
 320                                            32700.0 / (1 << 15));
 321         }
 322     }
 323
 324     *data_size = q->channels * AT1_SU_SAMPLES * sizeof(*samples);
 325     return avctx->block_align;
 326 }
 327
 328
 329 static av_cold int atrac1_decode_init(AVCodecContext *avctx)
 330 {
 331     AT1Ctx *q = avctx->priv_data;
 332
 333     avctx->sample_fmt = SAMPLE_FMT_FLT;
 334
 335     q->channels = avctx->channels;
 336
 337     /* Init the mdct transforms */
 338     ff_mdct_init(&q->mdct_ctx[0], 6, 1, -1.0/ (1 << 15));
 339     ff_mdct_init(&q->mdct_ctx[1], 8, 1, -1.0/ (1 << 15));
 340     ff_mdct_init(&q->mdct_ctx[2], 9, 1, -1.0/ (1 << 15));
 341
 342     ff_sine_window_init(ff_sine_32, 32);
 343
 344     atrac_generate_tables();
 345
 346     dsputil_init(&q->dsp, avctx);
 347
 348     q->bands[0] = q->low;
 349     q->bands[1] = q->mid;
 350     q->bands[2] = q->high;
 351
 352     /* Prepare the mdct overlap buffers */
 353     q->SUs[0].spectrum[0] = q->SUs[0].spec1;
 354     q->SUs[0].spectrum[1] = q->SUs[0].spec2;
 355     q->SUs[1].spectrum[0] = q->SUs[1].spec1;
 356     q->SUs[1].spectrum[1] = q->SUs[1].spec2;
 357
 358     return 0;
 359 }
 360
 361
 362 static av_cold int atrac1_decode_end(AVCodecContext * avctx) {
 363     AT1Ctx *q = avctx->priv_data;
 364
 365     ff_mdct_end(&q->mdct_ctx[0]);
 366     ff_mdct_end(&q->mdct_ctx[1]);
 367     ff_mdct_end(&q->mdct_ctx[2]);
 368     return 0;
 369 }
 370
 371
 372 AVCodec atrac1_decoder = {
 373     .name = "atrac1",
 374     .type = CODEC_TYPE_AUDIO,
 375     .id = CODEC_ID_ATRAC1,
 376     .priv_data_size = sizeof(AT1Ctx),
 377     .init = atrac1_decode_init,
 378     .close = atrac1_decode_end,
 379     .decode = atrac1_decode_frame,
 380     .long_name = NULL_IF_CONFIG_SMALL("Atrac 1 (Adaptive TRansform Acoustic Coding)"),
 381 };