libavcodec/nellymoserdec.c

   1 /*
   2  * NellyMoser audio decoder
   3  * Copyright (c) 2007 a840bda5870ba11f19698ff6eb9581dfb0f95fa5,
   4  *                    539459aeb7d425140b62a3ec7dbf6dc8e408a306, and
   5  *                    520e17cd55896441042b14df2566a6eb610ed444
   6  * Copyright (c) 2007 Loic Minier <lool at dooz.org>
   7  *                    Benjamin Larsson
   8  *
   9  * Permission is hereby granted, free of charge, to any person obtaining a
  10  * copy of this software and associated documentation files (the "Software"),
  11  * to deal in the Software without restriction, including without limitation
  12  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
  13  * and/or sell copies of the Software, and to permit persons to whom the
  14  * Software is furnished to do so, subject to the following conditions:
  15  *
  16  * The above copyright notice and this permission notice shall be included in
  17  * all copies or substantial portions of the Software.
  18  *
  19  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  20  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  21  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  22  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  23  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  24  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  25  * DEALINGS IN THE SOFTWARE.
  26  */
  27
  28 /**
  29  * @file nellymoserdec.c
  30  * The 3 alphanumeric copyright notices are md5summed they are from the original
  31  * implementors. The original code is available from http://code.google.com/p/nelly2pcm/
  32  */
  33
  34 #include "libavutil/random.h"
  35 #include "avcodec.h"
  36 #include "dsputil.h"
  37
  38 #define ALT_BITSTREAM_READER_LE
  39 #include "bitstream.h"
  40
  41 #define NELLY_BANDS       23
  42 #define NELLY_BLOCK_LEN   64
  43 #define NELLY_HEADER_BITS 116
  44 #define NELLY_DETAIL_BITS 198
  45 #define NELLY_BUF_LEN     128
  46 #define NELLY_FILL_LEN    124
  47 #define NELLY_BIT_CAP     6
  48 #define NELLY_BASE_OFF    4228
  49 #define NELLY_BASE_SHIFT  19
  50 #define NELLY_SAMPLES     (2 * NELLY_BUF_LEN)
  51
  52 static const float dequantization_table[127] = {
  53 0.0000000000,-0.8472560048, 0.7224709988, -1.5247479677, -0.4531480074, 0.3753609955, 1.4717899561,
  54 -1.9822579622, -1.1929379702, -0.5829370022, -0.0693780035, 0.3909569979,0.9069200158, 1.4862740040,
  55  2.2215409279, -2.3887870312, -1.8067539930, -1.4105420113, -1.0773609877, -0.7995010018,-0.5558109879,
  56 -0.3334020078, -0.1324490011, 0.0568020009, 0.2548770010, 0.4773550034, 0.7386850119, 1.0443060398,
  57 1.3954459429, 1.8098750114, 2.3918759823,-2.3893830776, -1.9884680510, -1.7514040470, -1.5643119812,
  58 -1.3922129869,-1.2164649963, -1.0469499826, -0.8905100226, -0.7645580173, -0.6454579830, -0.5259280205,
  59 -0.4059549868, -0.3029719889, -0.2096900046, -0.1239869967, -0.0479229987, 0.0257730000, 0.1001340002,
  60 0.1737180054, 0.2585540116, 0.3522900045, 0.4569880068, 0.5767750144, 0.7003160119, 0.8425520062,
  61 1.0093879700, 1.1821349859, 1.3534560204, 1.5320819616, 1.7332619429, 1.9722349644, 2.3978140354,
  62 -2.5756309032, -2.0573320389, -1.8984919786, -1.7727810144, -1.6662600040, -1.5742180347, -1.4993319511,
  63 -1.4316639900, -1.3652280569, -1.3000990152, -1.2280930281, -1.1588579416, -1.0921250582, -1.0135740042,
  64 -0.9202849865, -0.8287050128, -0.7374889851, -0.6447759867, -0.5590940118, -0.4857139885, -0.4110319912,
  65 -0.3459700048, -0.2851159871, -0.2341620028, -0.1870580018, -0.1442500055, -0.1107169986, -0.0739680007,
  66 -0.0365610011, -0.0073290002, 0.0203610007, 0.0479039997, 0.0751969963, 0.0980999991, 0.1220389977,
  67 0.1458999962, 0.1694349945, 0.1970459968, 0.2252430022, 0.2556869984, 0.2870100141, 0.3197099864,
  68 0.3525829911, 0.3889069855, 0.4334920049, 0.4769459963, 0.5204820037, 0.5644530058, 0.6122040153,
  69 0.6685929894, 0.7341650128, 0.8032159805, 0.8784040213, 0.9566209912, 1.0397069454, 1.1293770075,
  70 1.2211159468, 1.3080279827, 1.4024800062, 1.5056819916, 1.6227730513, 1.7724959850, 1.9430880547,
  71  2.2903931141
  72 };
  73
  74 static const uint8_t nelly_band_sizes_table[NELLY_BANDS] = {
  75 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 4, 4, 5, 6, 6, 7, 8, 9, 10, 12, 14, 15
  76 };
  77
  78 static const uint16_t nelly_init_table[64] = {
  79 3134, 5342, 6870, 7792, 8569, 9185, 9744, 10191, 10631, 11061, 11434, 11770,
  80 12116, 12513, 12925, 13300, 13674, 14027, 14352, 14716, 15117, 15477, 15824,
  81 16157, 16513, 16804, 17090, 17401, 17679, 17948, 18238, 18520, 18764, 19078,
  82 19381, 19640, 19921, 20205, 20500, 20813, 21162, 21465, 21794, 22137, 22453,
  83 22756, 23067, 23350, 23636, 23926, 24227, 24521, 24819, 25107, 25414, 25730,
  84 26120, 26497, 26895, 27344, 27877, 28463, 29426, 31355
  85 };
  86
  87 static const int16_t nelly_delta_table[32] = {
  88 -11725, -9420, -7910, -6801, -5948, -5233, -4599, -4039, -3507, -3030, -2596,
  89 -2170, -1774, -1383, -1016, -660, -329, -1, 337, 696, 1085, 1512, 1962, 2433,
  90 2968, 3569, 4314, 5279, 6622, 8154, 10076, 12975
  91 };
  92
  93 typedef struct NellyMoserDecodeContext {
  94     AVCodecContext* avctx;
  95     DECLARE_ALIGNED_16(float,float_buf[NELLY_SAMPLES]);
  96     float           state[128];
  97     AVRandomState   random_state;
  98     GetBitContext   gb;
  99     int             add_bias;
 100     float           scale_bias;
 101     DSPContext      dsp;
 102     MDCTContext     imdct_ctx;
 103     DECLARE_ALIGNED_16(float,imdct_tmp[NELLY_BUF_LEN]);
 104     DECLARE_ALIGNED_16(float,imdct_out[NELLY_BUF_LEN * 2]);
 105 } NellyMoserDecodeContext;
 106
 107 static DECLARE_ALIGNED_16(float,sine_window[128]);
 108
 109 static inline int signed_shift(int i, int shift) {
 110     if (shift > 0)
 111         return i << shift;
 112     return i >> -shift;
 113 }
 114
 115
 116 static void overlap_and_window(NellyMoserDecodeContext *s, float *state, float *audio, float *a_in)
 117 {
 118     int bot, top;
 119
 120     bot = 0;
 121     top = NELLY_BUF_LEN-1;
 122
 123     while (bot < NELLY_BUF_LEN) {
 124         audio[bot] = a_in [bot]*sine_window[bot]
 125                     +state[bot]*sine_window[top] + s->add_bias;
 126
 127         bot++;
 128         top--;
 129     }
 130     memcpy(state, a_in + NELLY_BUF_LEN, sizeof(float)*NELLY_BUF_LEN);
 131 }
 132
 133 static int sum_bits(short *buf, short shift, short off)
 134 {
 135     int i, ret = 0;
 136
 137     for (i = 0; i < NELLY_FILL_LEN; i++) {
 138         int b = buf[i]-off;
 139         b = ((b>>(shift-1))+1)>>1;
 140         ret += av_clip(b, 0, NELLY_BIT_CAP);
 141     }
 142
 143     return ret;
 144 }
 145
 146 static int headroom(int *la)
 147 {
 148     int l;
 149     if (*la == 0) {
 150         return 31;
 151     }
 152     l = 30 - av_log2(FFABS(*la));
 153     *la <<= l;
 154     return l;
 155 }
 156
 157
 158 static void get_sample_bits(const float *buf, int *bits)
 159 {
 160     int i, j;
 161     short sbuf[128];
 162     int bitsum = 0, last_bitsum, small_bitsum, big_bitsum;
 163     short shift, shift_saved;
 164     int max, sum, last_off, tmp;
 165     int big_off, small_off;
 166     int off;
 167
 168     max = 0;
 169     for (i = 0; i < NELLY_FILL_LEN; i++) {
 170         max = FFMAX(max, buf[i]);
 171     }
 172     shift = -16;
 173     shift += headroom(&max);
 174
 175     sum = 0;
 176     for (i = 0; i < NELLY_FILL_LEN; i++) {
 177         sbuf[i] = signed_shift(buf[i], shift);
 178         sbuf[i] = (3*sbuf[i])>>2;
 179         sum += sbuf[i];
 180     }
 181
 182     shift += 11;
 183     shift_saved = shift;
 184     sum -= NELLY_DETAIL_BITS << shift;
 185     shift += headroom(&sum);
 186     small_off = (NELLY_BASE_OFF * (sum>>16)) >> 15;
 187     shift = shift_saved - (NELLY_BASE_SHIFT+shift-31);
 188
 189     small_off = signed_shift(small_off, shift);
 190
 191     bitsum = sum_bits(sbuf, shift_saved, small_off);
 192
 193     if (bitsum != NELLY_DETAIL_BITS) {
 194         shift = 0;
 195         off = bitsum - NELLY_DETAIL_BITS;
 196
 197         for(shift=0; FFABS(off) <= 16383; shift++)
 198             off *= 2;
 199
 200         off = (off * NELLY_BASE_OFF) >> 15;
 201         shift = shift_saved-(NELLY_BASE_SHIFT+shift-15);
 202
 203         off = signed_shift(off, shift);
 204
 205         for (j = 1; j < 20; j++) {
 206             last_off = small_off;
 207             small_off += off;
 208             last_bitsum = bitsum;
 209
 210             bitsum = sum_bits(sbuf, shift_saved, small_off);
 211
 212             if ((bitsum-NELLY_DETAIL_BITS) * (last_bitsum-NELLY_DETAIL_BITS) <= 0)
 213                 break;
 214         }
 215
 216         if (bitsum > NELLY_DETAIL_BITS) {
 217             big_off = small_off;
 218             small_off = last_off;
 219             big_bitsum=bitsum;
 220             small_bitsum=last_bitsum;
 221         } else {
 222             big_off = last_off;
 223             big_bitsum=last_bitsum;
 224             small_bitsum=bitsum;
 225         }
 226
 227         while (bitsum != NELLY_DETAIL_BITS && j <= 19) {
 228             off = (big_off+small_off)>>1;
 229             bitsum = sum_bits(sbuf, shift_saved, off);
 230             if (bitsum > NELLY_DETAIL_BITS) {
 231                 big_off=off;
 232                 big_bitsum=bitsum;
 233             } else {
 234                 small_off = off;
 235                 small_bitsum=bitsum;
 236             }
 237             j++;
 238         }
 239
 240         if (abs(big_bitsum-NELLY_DETAIL_BITS) >=
 241             abs(small_bitsum-NELLY_DETAIL_BITS)) {
 242             bitsum = small_bitsum;
 243         } else {
 244             small_off = big_off;
 245             bitsum = big_bitsum;
 246         }
 247     }
 248
 249     for (i = 0; i < NELLY_FILL_LEN; i++) {
 250         tmp = sbuf[i]-small_off;
 251         tmp = ((tmp>>(shift_saved-1))+1)>>1;
 252         bits[i] = av_clip(tmp, 0, NELLY_BIT_CAP);
 253     }
 254
 255     if (bitsum > NELLY_DETAIL_BITS) {
 256         tmp = i = 0;
 257         while (tmp < NELLY_DETAIL_BITS) {
 258             tmp += bits[i];
 259             i++;
 260         }
 261
 262         bits[i-1] -= tmp - NELLY_DETAIL_BITS;
 263         for(; i < NELLY_FILL_LEN; i++)
 264             bits[i] = 0;
 265     }
 266 }
 267
 268 static void nelly_decode_block(NellyMoserDecodeContext *s,
 269                                const unsigned char block[NELLY_BLOCK_LEN],
 270                                float audio[NELLY_SAMPLES])
 271 {
 272     int i,j;
 273     float buf[NELLY_FILL_LEN], pows[NELLY_FILL_LEN];
 274     float *aptr, *bptr, *pptr, val, pval;
 275     int bits[NELLY_BUF_LEN];
 276     unsigned char v;
 277
 278     init_get_bits(&s->gb, block, NELLY_BLOCK_LEN * 8);
 279
 280     bptr = buf;
 281     pptr = pows;
 282     val = nelly_init_table[get_bits(&s->gb, 6)];
 283     for (i=0 ; i<NELLY_BANDS ; i++) {
 284         if (i > 0)
 285             val += nelly_delta_table[get_bits(&s->gb, 5)];
 286         pval = -pow(2, val/2048) * s->scale_bias;
 287         for (j = 0; j < nelly_band_sizes_table[i]; j++) {
 288             *bptr++ = val;
 289             *pptr++ = pval;
 290         }
 291
 292     }
 293
 294     get_sample_bits(buf, bits);
 295
 296     for (i = 0; i < 2; i++) {
 297         aptr = audio + i * NELLY_BUF_LEN;
 298
 299         init_get_bits(&s->gb, block, NELLY_BLOCK_LEN * 8);
 300         skip_bits(&s->gb, NELLY_HEADER_BITS + i*NELLY_DETAIL_BITS);
 301
 302         for (j = 0; j < NELLY_FILL_LEN; j++) {
 303             if (bits[j] <= 0) {
 304                 aptr[j] = M_SQRT1_2*pows[j];
 305                 if (av_random(&s->random_state) & 1)
 306                     aptr[j] *= -1.0;
 307             } else {
 308                 v = get_bits(&s->gb, bits[j]);
 309                 aptr[j] = dequantization_table[(1<<bits[j])-1+v]*pows[j];
 310             }
 311         }
 312         memset(&aptr[NELLY_FILL_LEN], 0,
 313                (NELLY_BUF_LEN - NELLY_FILL_LEN) * sizeof(float));
 314
 315         s->imdct_ctx.fft.imdct_calc(&s->imdct_ctx, s->imdct_out,
 316                                     aptr, s->imdct_tmp);
 317         /* XXX: overlapping and windowing should be part of a more
 318            generic imdct function */
 319         overlap_and_window(s, s->state, aptr, s->imdct_out);
 320     }
 321 }
 322
 323 static av_cold int decode_init(AVCodecContext * avctx) {
 324     NellyMoserDecodeContext *s = avctx->priv_data;
 325     int i;
 326
 327     s->avctx = avctx;
 328     av_init_random(0, &s->random_state);
 329     ff_mdct_init(&s->imdct_ctx, 8, 1);
 330
 331     dsputil_init(&s->dsp, avctx);
 332
 333     if(s->dsp.float_to_int16 == ff_float_to_int16_c) {
 334         s->add_bias = 385;
 335         s->scale_bias = 1.0/(8*32768);
 336     } else {
 337         s->add_bias = 0;
 338         s->scale_bias = 1.0/(1*8);
 339     }
 340
 341     /* Generate overlap window */
 342     if (!sine_window[0])
 343         for (i=0 ; i<128; i++) {
 344             sine_window[i] = sin((i + 0.5) / 256.0 * M_PI);
 345         }
 346
 347     return 0;
 348 }
 349
 350 static int decode_tag(AVCodecContext * avctx,
 351                       void *data, int *data_size,
 352                       const uint8_t * buf, int buf_size) {
 353     NellyMoserDecodeContext *s = avctx->priv_data;
 354     int blocks, i;
 355     int16_t* samples;
 356     *data_size = 0;
 357     samples = (int16_t*)data;
 358
 359     if (buf_size < avctx->block_align)
 360         return buf_size;
 361
 362     switch (buf_size) {
 363         case 64:    // 8000Hz
 364             blocks = 1; break;
 365         case 128:   // 11025Hz
 366             blocks = 2; break;
 367         case 256:   // 22050Hz
 368             blocks = 4; break;
 369         case 512:   // 44100Hz
 370             blocks = 8; break;
 371         default:
 372             av_log(avctx, AV_LOG_ERROR, "Tag size %d unknown, report sample!\n", buf_size);
 373             return buf_size;
 374     }
 375
 376     for (i=0 ; i<blocks ; i++) {
 377         nelly_decode_block(s, &buf[i*NELLY_BLOCK_LEN], s->float_buf);
 378         s->dsp.float_to_int16(&samples[i*NELLY_SAMPLES], s->float_buf, NELLY_SAMPLES);
 379         *data_size += NELLY_SAMPLES*sizeof(int16_t);
 380     }
 381
 382     return buf_size;
 383 }
 384
 385 static av_cold int decode_end(AVCodecContext * avctx) {
 386     NellyMoserDecodeContext *s = avctx->priv_data;
 387
 388     ff_mdct_end(&s->imdct_ctx);
 389     return 0;
 390 }
 391
 392 AVCodec nellymoser_decoder = {
 393     "nellymoser",
 394     CODEC_TYPE_AUDIO,
 395     CODEC_ID_NELLYMOSER,
 396     sizeof(NellyMoserDecodeContext),
 397     decode_init,
 398     NULL,
 399     decode_end,
 400     decode_tag,
 401     .long_name = "Nellymoser Asao",
 402 };
 403