libavcodec/amr.c

   1 /*
   2  * AMR Audio decoder stub
   3  * Copyright (c) 2003 the ffmpeg project
   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  /** @file
  23  * Adaptive Multi-Rate (AMR) Audio decoder stub.
  24  *
  25  * This code implements both an AMR-NarrowBand (AMR-NB) and an AMR-WideBand
  26  * (AMR-WB) audio encoder/decoder through external reference code from
  27  * http://www.3gpp.org/. The license of the code from 3gpp is unclear so you
  28  * have to download the code separately. Two versions exists: One fixed-point
  29  * and one with floats. For some reason the float-encoder is significant faster
  30  * at least on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip at MR102).
  31  * Both float and fixed point are supported for AMR-NB, but only float for
  32  * AMR-WB.
  33  *
  34  * \section AMR-NB
  35  *
  36  * \subsection Float
  37  * The float version (default) can be downloaded from:
  38  * http://www.3gpp.org/ftp/Specs/archive/26_series/26.104/26104-510.zip
  39  * Extract the source into \c "ffmpeg/libavcodec/amr_float".
  40  *
  41  * \subsection Fixed-point
  42  * The fixed-point (TS26.073) can be downloaded from:
  43  * http://www.3gpp.org/ftp/Specs/archive/26_series/26.073/26073-510.zip.
  44  * Extract the source into \c "ffmpeg/libavcodec/amr".
  45  * To use the fixed version run \c "./configure" with \c "--enable-amr_nb-fixed".
  46  *
  47  * \subsection Specification
  48  * The specification for AMR-NB can be found in TS 26.071
  49  * (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
  50  * info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm.
  51  *
  52  * \section AMR-WB
  53  * \subsection Float
  54  * The reference code can be downloaded from:
  55  * http://www.3gpp.org/ftp/Specs/archive/26_series/26.204/26204-510.zip
  56  * It should be extracted to \c "ffmpeg/libavcodec/amrwb_float". Enable it with
  57  * \c "--enable-amr_wb".
  58  *
  59  * \subsection Fixed-point
  60  * If someone wants to use the fixed point version it can be downloaded from:
  61  * http://www.3gpp.org/ftp/Specs/archive/26_series/26.173/26173-571.zip.
  62  *
  63  * \subsection Specification
  64  * The specification for AMR-WB can be downloaded from:
  65  * http://www.3gpp.org/ftp/Specs/archive/26_series/26.171/26171-500.zip.
  66  *
  67  */
  68
  69 #include "avcodec.h"
  70
  71 #ifdef CONFIG_AMR_NB_FIXED
  72
  73 #define MMS_IO
  74
  75 #include "amr/sp_dec.h"
  76 #include "amr/d_homing.h"
  77 #include "amr/typedef.h"
  78 #include "amr/sp_enc.h"
  79 #include "amr/sid_sync.h"
  80 #include "amr/e_homing.h"
  81
  82 #else
  83 #include "amr_float/interf_dec.h"
  84 #include "amr_float/interf_enc.h"
  85 #endif
  86
  87 /* Common code for fixed and float version*/
  88 typedef struct AMR_bitrates
  89 {
  90     int startrate;
  91     int stoprate;
  92     enum Mode mode;
  93 } AMR_bitrates;
  94
  95 /* Match desired bitrate with closest one*/
  96 static enum Mode getBitrateMode(int bitrate)
  97 {
  98     /* Adjusted so that all bitrates can be used from commandline where
  99        only a multiple of 1000 can be specified*/
 100     AMR_bitrates rates[]={ {0,4999,MR475}, //4
 101                            {5000,5899,MR515},//5
 102                            {5900,6699,MR59},//6
 103                            {6700,7000,MR67},//7
 104                            {7001,7949,MR74},//8
 105                            {7950,9999,MR795},//9
 106                            {10000,11999,MR102},//10
 107                            {12000,64000,MR122},//12
 108                          };
 109     int i;
 110
 111     for(i=0;i<8;i++)
 112     {
 113         if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
 114         {
 115             return(rates[i].mode);
 116         }
 117     }
 118     /*Return highest possible*/
 119     return(MR122);
 120 }
 121
 122 static void amr_decode_fix_avctx(AVCodecContext * avctx)
 123 {
 124     const int is_amr_wb = 1 + (avctx->codec_id == CODEC_ID_AMR_WB);
 125
 126     if(avctx->sample_rate == 0)
 127     {
 128         avctx->sample_rate = 8000 * is_amr_wb;
 129     }
 130
 131     if(avctx->channels == 0)
 132     {
 133         avctx->channels = 1;
 134     }
 135
 136     avctx->frame_size = 160 * is_amr_wb;
 137 }
 138
 139 #ifdef CONFIG_AMR_NB_FIXED
 140 /* fixed point version*/
 141 /* frame size in serial bitstream file (frame type + serial stream + flags) */
 142 #define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)
 143
 144 typedef struct AMRContext {
 145     int frameCount;
 146     Speech_Decode_FrameState *speech_decoder_state;
 147     enum RXFrameType rx_type;
 148     enum Mode mode;
 149     Word16 reset_flag;
 150     Word16 reset_flag_old;
 151
 152     enum Mode enc_bitrate;
 153     Speech_Encode_FrameState *enstate;
 154     sid_syncState *sidstate;
 155     enum TXFrameType tx_frametype;
 156 } AMRContext;
 157
 158 static int amr_nb_decode_init(AVCodecContext * avctx)
 159 {
 160     AMRContext *s = avctx->priv_data;
 161
 162     s->frameCount=0;
 163     s->speech_decoder_state=NULL;
 164     s->rx_type = (enum RXFrameType)0;
 165     s->mode= (enum Mode)0;
 166     s->reset_flag=0;
 167     s->reset_flag_old=1;
 168
 169     if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
 170     {
 171         av_log(avctx, AV_LOG_ERROR, "Speech_Decode_Frame_init error\n");
 172         return -1;
 173     }
 174
 175     amr_decode_fix_avctx(avctx);
 176
 177     if(avctx->channels > 1)
 178     {
 179         av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
 180         return -1;
 181     }
 182
 183     return 0;
 184 }
 185
 186 static int amr_nb_encode_init(AVCodecContext * avctx)
 187 {
 188     AMRContext *s = avctx->priv_data;
 189
 190     s->frameCount=0;
 191     s->speech_decoder_state=NULL;
 192     s->rx_type = (enum RXFrameType)0;
 193     s->mode= (enum Mode)0;
 194     s->reset_flag=0;
 195     s->reset_flag_old=1;
 196
 197     if(avctx->sample_rate!=8000)
 198     {
 199         av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
 200         return -1;
 201     }
 202
 203     if(avctx->channels!=1)
 204     {
 205         av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
 206         return -1;
 207     }
 208
 209     avctx->frame_size=160;
 210     avctx->coded_frame= avcodec_alloc_frame();
 211
 212     if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
 213     {
 214         av_log(avctx, AV_LOG_ERROR, "Speech_Encode_Frame_init error\n");
 215         return -1;
 216     }
 217
 218     s->enc_bitrate=getBitrateMode(avctx->bit_rate);
 219
 220     return 0;
 221 }
 222
 223 static int amr_nb_encode_close(AVCodecContext * avctx)
 224 {
 225     AMRContext *s = avctx->priv_data;
 226
 227     Speech_Encode_Frame_exit(&s->enstate);
 228     sid_sync_exit (&s->sidstate);
 229     av_freep(&avctx->coded_frame);
 230     return 0;
 231 }
 232
 233 static int amr_nb_decode_close(AVCodecContext * avctx)
 234 {
 235     AMRContext *s = avctx->priv_data;
 236
 237     Speech_Decode_Frame_exit(&s->speech_decoder_state);
 238     return 0;
 239 }
 240
 241 static int amr_nb_decode_frame(AVCodecContext * avctx,
 242             void *data, int *data_size,
 243             uint8_t * buf, int buf_size)
 244 {
 245     AMRContext *s = avctx->priv_data;
 246     uint8_t*amrData=buf;
 247     int offset=0;
 248     UWord8 toc, q, ft;
 249     Word16 serial[SERIAL_FRAMESIZE];   /* coded bits */
 250     Word16 *synth;
 251     UWord8 *packed_bits;
 252     static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
 253     int i;
 254
 255     //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);
 256
 257     synth=data;
 258
 259 //    while(offset<buf_size)
 260     {
 261         toc=amrData[offset];
 262         /* read rest of the frame based on ToC byte */
 263         q  = (toc >> 2) & 0x01;
 264         ft = (toc >> 3) & 0x0F;
 265
 266         //printf("offset=%d, packet_size=%d amrData= 0x%X %X %X %X\n",offset,packed_size[ft],amrData[offset],amrData[offset+1],amrData[offset+2],amrData[offset+3]);
 267
 268         offset++;
 269
 270         packed_bits=amrData+offset;
 271
 272         offset+=packed_size[ft];
 273
 274         //Unsort and unpack bits
 275         s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);
 276
 277         //We have a new frame
 278         s->frameCount++;
 279
 280         if (s->rx_type == RX_NO_DATA)
 281         {
 282             s->mode = s->speech_decoder_state->prev_mode;
 283         }
 284         else {
 285             s->speech_decoder_state->prev_mode = s->mode;
 286         }
 287
 288         /* if homed: check if this frame is another homing frame */
 289         if (s->reset_flag_old == 1)
 290         {
 291             /* only check until end of first subframe */
 292             s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
 293         }
 294         /* produce encoder homing frame if homed & input=decoder homing frame */
 295         if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
 296         {
 297             for (i = 0; i < L_FRAME; i++)
 298             {
 299                 synth[i] = EHF_MASK;
 300             }
 301         }
 302         else
 303         {
 304             /* decode frame */
 305             Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
 306         }
 307
 308         //Each AMR-frame results in 160 16-bit samples
 309         *data_size+=160*2;
 310         synth+=160;
 311
 312         /* if not homed: check whether current frame is a homing frame */
 313         if (s->reset_flag_old == 0)
 314         {
 315             /* check whole frame */
 316             s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
 317         }
 318         /* reset decoder if current frame is a homing frame */
 319         if (s->reset_flag != 0)
 320         {
 321             Speech_Decode_Frame_reset(s->speech_decoder_state);
 322         }
 323         s->reset_flag_old = s->reset_flag;
 324
 325     }
 326     return offset;
 327 }
 328
 329
 330 static int amr_nb_encode_frame(AVCodecContext *avctx,
 331                             unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
 332 {
 333     short serial_data[250] = {0};
 334     AMRContext *s = avctx->priv_data;
 335     int written;
 336
 337     s->reset_flag = encoder_homing_frame_test(data);
 338
 339     Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode);
 340
 341     /* add frame type and mode */
 342     sid_sync (s->sidstate, s->mode, &s->tx_frametype);
 343
 344     written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
 345
 346     if (s->reset_flag != 0)
 347     {
 348         Speech_Encode_Frame_reset(s->enstate);
 349         sid_sync_reset(s->sidstate);
 350     }
 351     return written;
 352 }
 353
 354
 355 #elif defined(CONFIG_AMR_NB) /* Float point version*/
 356
 357 typedef struct AMRContext {
 358     int frameCount;
 359     void * decState;
 360     int *enstate;
 361     enum Mode enc_bitrate;
 362 } AMRContext;
 363
 364 static int amr_nb_decode_init(AVCodecContext * avctx)
 365 {
 366     AMRContext *s = avctx->priv_data;
 367
 368     s->frameCount=0;
 369     s->decState=Decoder_Interface_init();
 370     if(!s->decState)
 371     {
 372         av_log(avctx, AV_LOG_ERROR, "Decoder_Interface_init error\r\n");
 373         return -1;
 374     }
 375
 376     amr_decode_fix_avctx(avctx);
 377
 378     if(avctx->channels > 1)
 379     {
 380         av_log(avctx, AV_LOG_ERROR, "amr_nb: multichannel decoding not supported\n");
 381         return -1;
 382     }
 383
 384     return 0;
 385 }
 386
 387 static int amr_nb_encode_init(AVCodecContext * avctx)
 388 {
 389     AMRContext *s = avctx->priv_data;
 390
 391     s->frameCount=0;
 392
 393     if(avctx->sample_rate!=8000)
 394     {
 395         av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
 396         return -1;
 397     }
 398
 399     if(avctx->channels!=1)
 400     {
 401         av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
 402         return -1;
 403     }
 404
 405     avctx->frame_size=160;
 406     avctx->coded_frame= avcodec_alloc_frame();
 407
 408     s->enstate=Encoder_Interface_init(0);
 409     if(!s->enstate)
 410     {
 411         av_log(avctx, AV_LOG_ERROR, "Encoder_Interface_init error\n");
 412         return -1;
 413     }
 414
 415     s->enc_bitrate=getBitrateMode(avctx->bit_rate);
 416
 417     return 0;
 418 }
 419
 420 static int amr_nb_decode_close(AVCodecContext * avctx)
 421 {
 422     AMRContext *s = avctx->priv_data;
 423
 424     Decoder_Interface_exit(s->decState);
 425     return 0;
 426 }
 427
 428 static int amr_nb_encode_close(AVCodecContext * avctx)
 429 {
 430     AMRContext *s = avctx->priv_data;
 431
 432     Encoder_Interface_exit(s->enstate);
 433     av_freep(&avctx->coded_frame);
 434     return 0;
 435 }
 436
 437 static int amr_nb_decode_frame(AVCodecContext * avctx,
 438             void *data, int *data_size,
 439             uint8_t * buf, int buf_size)
 440 {
 441     AMRContext *s = avctx->priv_data;
 442     uint8_t*amrData=buf;
 443     static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
 444     enum Mode dec_mode;
 445     int packet_size;
 446
 447     /* av_log(NULL,AV_LOG_DEBUG,"amr_decode_frame buf=%p buf_size=%d frameCount=%d!!\n",buf,buf_size,s->frameCount); */
 448
 449     dec_mode = (buf[0] >> 3) & 0x000F;
 450     packet_size = block_size[dec_mode]+1;
 451
 452     if(packet_size > buf_size) {
 453         av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size);
 454         return -1;
 455     }
 456
 457     s->frameCount++;
 458     /* av_log(NULL,AV_LOG_DEBUG,"packet_size=%d amrData= 0x%X %X %X %X\n",packet_size,amrData[0],amrData[1],amrData[2],amrData[3]); */
 459     /* call decoder */
 460     Decoder_Interface_Decode(s->decState, amrData, data, 0);
 461     *data_size=160*2;
 462
 463     return packet_size;
 464 }
 465
 466 static int amr_nb_encode_frame(AVCodecContext *avctx,
 467                             unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
 468 {
 469     AMRContext *s = avctx->priv_data;
 470     int written;
 471
 472     s->enc_bitrate=getBitrateMode(avctx->bit_rate);
 473
 474     written = Encoder_Interface_Encode(s->enstate,
 475         s->enc_bitrate,
 476         data,
 477         frame,
 478         0);
 479     /* av_log(NULL,AV_LOG_DEBUG,"amr_nb_encode_frame encoded %u bytes, bitrate %u, first byte was %#02x\n",written, s->enc_bitrate, frame[0] ); */
 480
 481     return written;
 482 }
 483
 484 #endif
 485
 486 #if defined(CONFIG_AMR_NB) || defined(CONFIG_AMR_NB_FIXED)
 487
 488 AVCodec amr_nb_decoder =
 489 {
 490     "amr_nb",
 491     CODEC_TYPE_AUDIO,
 492     CODEC_ID_AMR_NB,
 493     sizeof(AMRContext),
 494     amr_nb_decode_init,
 495     NULL,
 496     amr_nb_decode_close,
 497     amr_nb_decode_frame,
 498 };
 499
 500 AVCodec amr_nb_encoder =
 501 {
 502     "amr_nb",
 503     CODEC_TYPE_AUDIO,
 504     CODEC_ID_AMR_NB,
 505     sizeof(AMRContext),
 506     amr_nb_encode_init,
 507     amr_nb_encode_frame,
 508     amr_nb_encode_close,
 509     NULL,
 510 };
 511
 512 #endif
 513
 514 /* -----------AMR wideband ------------*/
 515 #ifdef CONFIG_AMR_WB
 516
 517 #ifdef _TYPEDEF_H
 518 //To avoid duplicate typedefs from typdef in amr-nb
 519 #define typedef_h
 520 #endif
 521
 522 #include "amrwb_float/enc_if.h"
 523 #include "amrwb_float/dec_if.h"
 524
 525 /* Common code for fixed and float version*/
 526 typedef struct AMRWB_bitrates
 527 {
 528     int startrate;
 529     int stoprate;
 530     int mode;
 531 } AMRWB_bitrates;
 532
 533 static int getWBBitrateMode(int bitrate)
 534 {
 535     /* Adjusted so that all bitrates can be used from commandline where
 536        only a multiple of 1000 can be specified*/
 537     AMRWB_bitrates rates[]={ {0,7999,0}, //6.6kHz
 538                            {8000,9999,1},//8.85
 539                            {10000,13000,2},//12.65
 540                            {13001,14999,3},//14.25
 541                            {15000,17000,4},//15.85
 542                            {17001,18000,5},//18.25
 543                            {18001,22000,6},//19.85
 544                            {22001,23000,7},//23.05
 545                            {23001,24000,8},//23.85
 546                          };
 547     int i;
 548
 549     for(i=0;i<9;i++)
 550     {
 551         if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
 552         {
 553             return(rates[i].mode);
 554         }
 555     }
 556     /*Return highest possible*/
 557     return(8);
 558 }
 559
 560
 561 typedef struct AMRWBContext {
 562     int frameCount;
 563     void *state;
 564     int mode;
 565     Word16 allow_dtx;
 566 } AMRWBContext;
 567
 568 static int amr_wb_encode_init(AVCodecContext * avctx)
 569 {
 570     AMRWBContext *s = avctx->priv_data;
 571
 572     s->frameCount=0;
 573
 574     if(avctx->sample_rate!=16000)
 575     {
 576         av_log(avctx, AV_LOG_ERROR, "Only 16000Hz sample rate supported\n");
 577         return -1;
 578     }
 579
 580     if(avctx->channels!=1)
 581     {
 582         av_log(avctx, AV_LOG_ERROR, "Only mono supported\n");
 583         return -1;
 584     }
 585
 586     avctx->frame_size=320;
 587     avctx->coded_frame= avcodec_alloc_frame();
 588
 589     s->state = E_IF_init();
 590     s->mode=getWBBitrateMode(avctx->bit_rate);
 591     s->allow_dtx=0;
 592
 593     return 0;
 594 }
 595
 596 static int amr_wb_encode_close(AVCodecContext * avctx)
 597 {
 598     AMRWBContext *s = avctx->priv_data;
 599
 600     E_IF_exit(s->state);
 601     av_freep(&avctx->coded_frame);
 602     s->frameCount++;
 603     return 0;
 604 }
 605
 606 static int amr_wb_encode_frame(AVCodecContext *avctx,
 607                             unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
 608 {
 609     AMRWBContext *s = avctx->priv_data;
 610     int size;
 611
 612     s->mode=getWBBitrateMode(avctx->bit_rate);
 613     size = E_IF_encode(s->state, s->mode, data, frame, s->allow_dtx);
 614     return size;
 615 }
 616
 617 static int amr_wb_decode_init(AVCodecContext * avctx)
 618 {
 619     AMRWBContext *s = avctx->priv_data;
 620
 621     s->frameCount=0;
 622     s->state = D_IF_init();
 623
 624     amr_decode_fix_avctx(avctx);
 625
 626     if(avctx->channels > 1)
 627     {
 628         av_log(avctx, AV_LOG_ERROR, "amr_wb: multichannel decoding not supported\n");
 629         return -1;
 630     }
 631
 632     return 0;
 633 }
 634
 635 extern const UWord8 block_size[];
 636
 637 static int amr_wb_decode_frame(AVCodecContext * avctx,
 638             void *data, int *data_size,
 639             uint8_t * buf, int buf_size)
 640 {
 641     AMRWBContext *s = avctx->priv_data;
 642     uint8_t*amrData=buf;
 643     int mode;
 644     int packet_size;
 645
 646     if(buf_size==0) {
 647         /* nothing to do */
 648         return 0;
 649     }
 650
 651     mode = (amrData[0] >> 3) & 0x000F;
 652     packet_size = block_size[mode];
 653
 654     if(packet_size > buf_size) {
 655         av_log(avctx, AV_LOG_ERROR, "amr frame too short (%u, should be %u)\n", buf_size, packet_size+1);
 656         return -1;
 657     }
 658
 659     s->frameCount++;
 660     D_IF_decode( s->state, amrData, data, _good_frame);
 661     *data_size=320*2;
 662     return packet_size;
 663 }
 664
 665 static int amr_wb_decode_close(AVCodecContext * avctx)
 666 {
 667     AMRWBContext *s = avctx->priv_data;
 668
 669     D_IF_exit(s->state);
 670     return 0;
 671 }
 672
 673 AVCodec amr_wb_decoder =
 674 {
 675     "amr_wb",
 676     CODEC_TYPE_AUDIO,
 677     CODEC_ID_AMR_WB,
 678     sizeof(AMRWBContext),
 679     amr_wb_decode_init,
 680     NULL,
 681     amr_wb_decode_close,
 682     amr_wb_decode_frame,
 683 };
 684
 685 AVCodec amr_wb_encoder =
 686 {
 687     "amr_wb",
 688     CODEC_TYPE_AUDIO,
 689     CODEC_ID_AMR_WB,
 690     sizeof(AMRWBContext),
 691     amr_wb_encode_init,
 692     amr_wb_encode_frame,
 693     amr_wb_encode_close,
 694     NULL,
 695 };
 696
 697 #endif //CONFIG_AMR_WB