libavcodec/aac.h

   1 /*
   2  * AAC definitions and structures
   3  * Copyright (c) 2005-2006 Oded Shimon ( ods15 ods15 dyndns org )
   4  * Copyright (c) 2006-2007 Maxim Gavrilov ( maxim.gavrilov gmail com )
   5  *
   6  * This file is part of Libav.
   7  *
   8  * Libav 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  * Libav 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 Libav; 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
  25  * AAC definitions and structures
  26  * @author Oded Shimon  ( ods15 ods15 dyndns org )
  27  * @author Maxim Gavrilov ( maxim.gavrilov gmail com )
  28  */
  29
  30 #ifndef AVCODEC_AAC_H
  31 #define AVCODEC_AAC_H
  32
  33 #include "libavutil/float_dsp.h"
  34 #include "avcodec.h"
  35 #include "imdct15.h"
  36 #include "fft.h"
  37 #include "mpeg4audio.h"
  38 #include "sbr.h"
  39
  40 #include <stdint.h>
  41
  42 #define MAX_CHANNELS 64
  43 #define MAX_ELEM_ID 16
  44
  45 #define TNS_MAX_ORDER 20
  46 #define MAX_LTP_LONG_SFB 40
  47
  48 enum RawDataBlockType {
  49     TYPE_SCE,
  50     TYPE_CPE,
  51     TYPE_CCE,
  52     TYPE_LFE,
  53     TYPE_DSE,
  54     TYPE_PCE,
  55     TYPE_FIL,
  56     TYPE_END,
  57 };
  58
  59 enum ExtensionPayloadID {
  60     EXT_FILL,
  61     EXT_FILL_DATA,
  62     EXT_DATA_ELEMENT,
  63     EXT_DYNAMIC_RANGE = 0xb,
  64     EXT_SBR_DATA      = 0xd,
  65     EXT_SBR_DATA_CRC  = 0xe,
  66 };
  67
  68 enum WindowSequence {
  69     ONLY_LONG_SEQUENCE,
  70     LONG_START_SEQUENCE,
  71     EIGHT_SHORT_SEQUENCE,
  72     LONG_STOP_SEQUENCE,
  73 };
  74
  75 enum BandType {
  76     ZERO_BT        = 0,     ///< Scalefactors and spectral data are all zero.
  77     FIRST_PAIR_BT  = 5,     ///< This and later band types encode two values (rather than four) with one code word.
  78     ESC_BT         = 11,    ///< Spectral data are coded with an escape sequence.
  79     NOISE_BT       = 13,    ///< Spectral data are scaled white noise not coded in the bitstream.
  80     INTENSITY_BT2  = 14,    ///< Scalefactor data are intensity stereo positions.
  81     INTENSITY_BT   = 15,    ///< Scalefactor data are intensity stereo positions.
  82 };
  83
  84 #define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
  85
  86 enum ChannelPosition {
  87     AAC_CHANNEL_OFF   = 0,
  88     AAC_CHANNEL_FRONT = 1,
  89     AAC_CHANNEL_SIDE  = 2,
  90     AAC_CHANNEL_BACK  = 3,
  91     AAC_CHANNEL_LFE   = 4,
  92     AAC_CHANNEL_CC    = 5,
  93 };
  94
  95 /**
  96  * The point during decoding at which channel coupling is applied.
  97  */
  98 enum CouplingPoint {
  99     BEFORE_TNS,
 100     BETWEEN_TNS_AND_IMDCT,
 101     AFTER_IMDCT = 3,
 102 };
 103
 104 /**
 105  * Output configuration status
 106  */
 107 enum OCStatus {
 108     OC_NONE,        ///< Output unconfigured
 109     OC_TRIAL_PCE,   ///< Output configuration under trial specified by an inband PCE
 110     OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header
 111     OC_GLOBAL_HDR,  ///< Output configuration set in a global header but not yet locked
 112     OC_LOCKED,      ///< Output configuration locked in place
 113 };
 114
 115 typedef struct OutputConfiguration {
 116     MPEG4AudioConfig m4ac;
 117     uint8_t layout_map[MAX_ELEM_ID*4][3];
 118     int layout_map_tags;
 119     int channels;
 120     uint64_t channel_layout;
 121     enum OCStatus status;
 122 } OutputConfiguration;
 123
 124 /**
 125  * Predictor State
 126  */
 127 typedef struct PredictorState {
 128     float cor0;
 129     float cor1;
 130     float var0;
 131     float var1;
 132     float r0;
 133     float r1;
 134 } PredictorState;
 135
 136 #define MAX_PREDICTORS 672
 137
 138 #define SCALE_DIV_512    36    ///< scalefactor difference that corresponds to scale difference in 512 times
 139 #define SCALE_ONE_POS   140    ///< scalefactor index that corresponds to scale=1.0
 140 #define SCALE_MAX_POS   255    ///< scalefactor index maximum value
 141 #define SCALE_MAX_DIFF   60    ///< maximum scalefactor difference allowed by standard
 142 #define SCALE_DIFF_ZERO  60    ///< codebook index corresponding to zero scalefactor indices difference
 143
 144 /**
 145  * Long Term Prediction
 146  */
 147 typedef struct LongTermPrediction {
 148     int8_t present;
 149     int16_t lag;
 150     float coef;
 151     int8_t used[MAX_LTP_LONG_SFB];
 152 } LongTermPrediction;
 153
 154 /**
 155  * Individual Channel Stream
 156  */
 157 typedef struct IndividualChannelStream {
 158     uint8_t max_sfb;            ///< number of scalefactor bands per group
 159     enum WindowSequence window_sequence[2];
 160     uint8_t use_kb_window[2];   ///< If set, use Kaiser-Bessel window, otherwise use a sine window.
 161     int num_window_groups;
 162     uint8_t group_len[8];
 163     LongTermPrediction ltp;
 164     const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
 165     const uint8_t *swb_sizes;   ///< table of scalefactor band sizes for a particular window
 166     int num_swb;                ///< number of scalefactor window bands
 167     int num_windows;
 168     int tns_max_bands;
 169     int predictor_present;
 170     int predictor_initialized;
 171     int predictor_reset_group;
 172     uint8_t prediction_used[41];
 173 } IndividualChannelStream;
 174
 175 /**
 176  * Temporal Noise Shaping
 177  */
 178 typedef struct TemporalNoiseShaping {
 179     int present;
 180     int n_filt[8];
 181     int length[8][4];
 182     int direction[8][4];
 183     int order[8][4];
 184     float coef[8][4][TNS_MAX_ORDER];
 185 } TemporalNoiseShaping;
 186
 187 /**
 188  * Dynamic Range Control - decoded from the bitstream but not processed further.
 189  */
 190 typedef struct DynamicRangeControl {
 191     int pce_instance_tag;                           ///< Indicates with which program the DRC info is associated.
 192     int dyn_rng_sgn[17];                            ///< DRC sign information; 0 - positive, 1 - negative
 193     int dyn_rng_ctl[17];                            ///< DRC magnitude information
 194     int exclude_mask[MAX_CHANNELS];                 ///< Channels to be excluded from DRC processing.
 195     int band_incr;                                  ///< Number of DRC bands greater than 1 having DRC info.
 196     int interpolation_scheme;                       ///< Indicates the interpolation scheme used in the SBR QMF domain.
 197     int band_top[17];                               ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
 198     int prog_ref_level;                             /**< A reference level for the long-term program audio level for all
 199                                                      *   channels combined.
 200                                                      */
 201 } DynamicRangeControl;
 202
 203 typedef struct Pulse {
 204     int num_pulse;
 205     int start;
 206     int pos[4];
 207     int amp[4];
 208 } Pulse;
 209
 210 /**
 211  * coupling parameters
 212  */
 213 typedef struct ChannelCoupling {
 214     enum CouplingPoint coupling_point;  ///< The point during decoding at which coupling is applied.
 215     int num_coupled;       ///< number of target elements
 216     enum RawDataBlockType type[8];   ///< Type of channel element to be coupled - SCE or CPE.
 217     int id_select[8];      ///< element id
 218     int ch_select[8];      /**< [0] shared list of gains; [1] list of gains for right channel;
 219                             *   [2] list of gains for left channel; [3] lists of gains for both channels
 220                             */
 221     float gain[16][120];
 222 } ChannelCoupling;
 223
 224 /**
 225  * Single Channel Element - used for both SCE and LFE elements.
 226  */
 227 typedef struct SingleChannelElement {
 228     IndividualChannelStream ics;
 229     TemporalNoiseShaping tns;
 230     Pulse pulse;
 231     enum BandType band_type[128];                   ///< band types
 232     int band_type_run_end[120];                     ///< band type run end points
 233     float sf[120];                                  ///< scalefactors
 234     int sf_idx[128];                                ///< scalefactor indices (used by encoder)
 235     uint8_t zeroes[128];                            ///< band is not coded (used by encoder)
 236     DECLARE_ALIGNED(32, float,   coeffs)[1024];     ///< coefficients for IMDCT
 237     DECLARE_ALIGNED(32, float,   saved)[1536];      ///< overlap
 238     DECLARE_ALIGNED(32, float,   ret_buf)[2048];    ///< PCM output buffer
 239     DECLARE_ALIGNED(16, float,   ltp_state)[3072];  ///< time signal for LTP
 240     PredictorState predictor_state[MAX_PREDICTORS];
 241     float *ret;                                     ///< PCM output
 242 } SingleChannelElement;
 243
 244 /**
 245  * channel element - generic struct for SCE/CPE/CCE/LFE
 246  */
 247 typedef struct ChannelElement {
 248     // CPE specific
 249     int common_window;        ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
 250     int     ms_mode;          ///< Signals mid/side stereo flags coding mode (used by encoder)
 251     uint8_t ms_mask[128];     ///< Set if mid/side stereo is used for each scalefactor window band
 252     // shared
 253     SingleChannelElement ch[2];
 254     // CCE specific
 255     ChannelCoupling coup;
 256     SpectralBandReplication sbr;
 257 } ChannelElement;
 258
 259 /**
 260  * main AAC context
 261  */
 262 typedef struct AACContext {
 263     AVCodecContext *avctx;
 264     AVFrame *frame;
 265
 266     int is_saved;                 ///< Set if elements have stored overlap from previous frame.
 267     DynamicRangeControl che_drc;
 268
 269     /**
 270      * @name Channel element related data
 271      * @{
 272      */
 273     ChannelElement          *che[4][MAX_ELEM_ID];
 274     ChannelElement  *tag_che_map[4][MAX_ELEM_ID];
 275     int tags_mapped;
 276     /** @} */
 277
 278     /**
 279      * @name temporary aligned temporary buffers
 280      * (We do not want to have these on the stack.)
 281      * @{
 282      */
 283     DECLARE_ALIGNED(32, float, buf_mdct)[1024];
 284     /** @} */
 285
 286     /**
 287      * @name Computed / set up during initialization
 288      * @{
 289      */
 290     FFTContext mdct;
 291     FFTContext mdct_small;
 292     FFTContext mdct_ld;
 293     FFTContext mdct_ltp;
 294     IMDCT15Context *mdct480;
 295     AVFloatDSPContext fdsp;
 296     int random_state;
 297     /** @} */
 298
 299     /**
 300      * @name Members used for output
 301      * @{
 302      */
 303     SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement
 304     /** @} */
 305
 306     DECLARE_ALIGNED(32, float, temp)[128];
 307
 308     OutputConfiguration oc[2];
 309 } AACContext;
 310
 311 #endif /* AVCODEC_AAC_H */