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 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/aac.h
  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 "avcodec.h"
  34 #include "dsputil.h"
  35 #include "mpeg4audio.h"
  36
  37 #include <stdint.h>
  38
  39 #define AAC_INIT_VLC_STATIC(num, size) \
  40     INIT_VLC_STATIC(&vlc_spectral[num], 6, ff_aac_spectral_sizes[num], \
  41          ff_aac_spectral_bits[num], sizeof( ff_aac_spectral_bits[num][0]), sizeof( ff_aac_spectral_bits[num][0]), \
  42         ff_aac_spectral_codes[num], sizeof(ff_aac_spectral_codes[num][0]), sizeof(ff_aac_spectral_codes[num][0]), \
  43         size);
  44
  45 #define MAX_CHANNELS 64
  46 #define MAX_ELEM_ID 16
  47
  48 #define TNS_MAX_ORDER 20
  49
  50 enum RawDataBlockType {
  51     TYPE_SCE,
  52     TYPE_CPE,
  53     TYPE_CCE,
  54     TYPE_LFE,
  55     TYPE_DSE,
  56     TYPE_PCE,
  57     TYPE_FIL,
  58     TYPE_END,
  59 };
  60
  61 enum ExtensionPayloadID {
  62     EXT_FILL,
  63     EXT_FILL_DATA,
  64     EXT_DATA_ELEMENT,
  65     EXT_DYNAMIC_RANGE = 0xb,
  66     EXT_SBR_DATA      = 0xd,
  67     EXT_SBR_DATA_CRC  = 0xe,
  68 };
  69
  70 enum WindowSequence {
  71     ONLY_LONG_SEQUENCE,
  72     LONG_START_SEQUENCE,
  73     EIGHT_SHORT_SEQUENCE,
  74     LONG_STOP_SEQUENCE,
  75 };
  76
  77 enum BandType {
  78     ZERO_BT        = 0,     ///< Scalefactors and spectral data are all zero.
  79     FIRST_PAIR_BT  = 5,     ///< This and later band types encode two values (rather than four) with one code word.
  80     ESC_BT         = 11,    ///< Spectral data are coded with an escape sequence.
  81     NOISE_BT       = 13,    ///< Spectral data are scaled white noise not coded in the bitstream.
  82     INTENSITY_BT2  = 14,    ///< Scalefactor data are intensity stereo positions.
  83     INTENSITY_BT   = 15,    ///< Scalefactor data are intensity stereo positions.
  84 };
  85
  86 #define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
  87
  88 enum ChannelPosition {
  89     AAC_CHANNEL_FRONT = 1,
  90     AAC_CHANNEL_SIDE  = 2,
  91     AAC_CHANNEL_BACK  = 3,
  92     AAC_CHANNEL_LFE   = 4,
  93     AAC_CHANNEL_CC    = 5,
  94 };
  95
  96 /**
  97  * The point during decoding at which channel coupling is applied.
  98  */
  99 enum CouplingPoint {
 100     BEFORE_TNS,
 101     BETWEEN_TNS_AND_IMDCT,
 102     AFTER_IMDCT = 3,
 103 };
 104
 105 /**
 106  * Output configuration status
 107  */
 108 enum OCStatus {
 109     OC_NONE,        //< Output unconfigured
 110     OC_TRIAL_PCE,   //< Output configuration under trial specified by an inband PCE
 111     OC_TRIAL_FRAME, //< Output configuration under trial specified by a frame header
 112     OC_LOCKED,      //< Output configuration locked in place
 113 };
 114
 115 /**
 116  * Predictor State
 117  */
 118 typedef struct {
 119     float cor0;
 120     float cor1;
 121     float var0;
 122     float var1;
 123     float r0;
 124     float r1;
 125 } PredictorState;
 126
 127 #define MAX_PREDICTORS 672
 128
 129 #define SCALE_DIV_512    36    ///< scalefactor difference that corresponds to scale difference in 512 times
 130 #define SCALE_ONE_POS   140    ///< scalefactor index that corresponds to scale=1.0
 131 #define SCALE_MAX_POS   255    ///< scalefactor index maximum value
 132 #define SCALE_MAX_DIFF   60    ///< maximum scalefactor difference allowed by standard
 133 #define SCALE_DIFF_ZERO  60    ///< codebook index corresponding to zero scalefactor indices difference
 134
 135 /**
 136  * Individual Channel Stream
 137  */
 138 typedef struct {
 139     uint8_t max_sfb;            ///< number of scalefactor bands per group
 140     enum WindowSequence window_sequence[2];
 141     uint8_t use_kb_window[2];   ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
 142     int num_window_groups;
 143     uint8_t group_len[8];
 144     const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
 145     const uint8_t *swb_sizes;   ///< table of scalefactor band sizes for a particular window
 146     int num_swb;                ///< number of scalefactor window bands
 147     int num_windows;
 148     int tns_max_bands;
 149     int predictor_present;
 150     int predictor_initialized;
 151     int predictor_reset_group;
 152     uint8_t prediction_used[41];
 153 } IndividualChannelStream;
 154
 155 /**
 156  * Temporal Noise Shaping
 157  */
 158 typedef struct {
 159     int present;
 160     int n_filt[8];
 161     int length[8][4];
 162     int direction[8][4];
 163     int order[8][4];
 164     float coef[8][4][TNS_MAX_ORDER];
 165 } TemporalNoiseShaping;
 166
 167 /**
 168  * Dynamic Range Control - decoded from the bitstream but not processed further.
 169  */
 170 typedef struct {
 171     int pce_instance_tag;                           ///< Indicates with which program the DRC info is associated.
 172     int dyn_rng_sgn[17];                            ///< DRC sign information; 0 - positive, 1 - negative
 173     int dyn_rng_ctl[17];                            ///< DRC magnitude information
 174     int exclude_mask[MAX_CHANNELS];                 ///< Channels to be excluded from DRC processing.
 175     int band_incr;                                  ///< Number of DRC bands greater than 1 having DRC info.
 176     int interpolation_scheme;                       ///< Indicates the interpolation scheme used in the SBR QMF domain.
 177     int band_top[17];                               ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
 178     int prog_ref_level;                             /**< A reference level for the long-term program audio level for all
 179                                                      *   channels combined.
 180                                                      */
 181 } DynamicRangeControl;
 182
 183 typedef struct {
 184     int num_pulse;
 185     int start;
 186     int pos[4];
 187     int amp[4];
 188 } Pulse;
 189
 190 /**
 191  * coupling parameters
 192  */
 193 typedef struct {
 194     enum CouplingPoint coupling_point;  ///< The point during decoding at which coupling is applied.
 195     int num_coupled;       ///< number of target elements
 196     enum RawDataBlockType type[8];   ///< Type of channel element to be coupled - SCE or CPE.
 197     int id_select[8];      ///< element id
 198     int ch_select[8];      /**< [0] shared list of gains; [1] list of gains for right channel;
 199                             *   [2] list of gains for left channel; [3] lists of gains for both channels
 200                             */
 201     float gain[16][120];
 202 } ChannelCoupling;
 203
 204 /**
 205  * Single Channel Element - used for both SCE and LFE elements.
 206  */
 207 typedef struct {
 208     IndividualChannelStream ics;
 209     TemporalNoiseShaping tns;
 210     Pulse pulse;
 211     enum BandType band_type[128];             ///< band types
 212     int band_type_run_end[120];               ///< band type run end points
 213     float sf[120];                            ///< scalefactors
 214     int sf_idx[128];                          ///< scalefactor indices (used by encoder)
 215     uint8_t zeroes[128];                      ///< band is not coded (used by encoder)
 216     DECLARE_ALIGNED_16(float, coeffs[1024]);  ///< coefficients for IMDCT
 217     DECLARE_ALIGNED_16(float, saved[1024]);   ///< overlap
 218     DECLARE_ALIGNED_16(float, ret[1024]);     ///< PCM output
 219     PredictorState predictor_state[MAX_PREDICTORS];
 220 } SingleChannelElement;
 221
 222 /**
 223  * channel element - generic struct for SCE/CPE/CCE/LFE
 224  */
 225 typedef struct {
 226     // CPE specific
 227     int common_window;        ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
 228     int     ms_mode;          ///< Signals mid/side stereo flags coding mode (used by encoder)
 229     uint8_t ms_mask[128];     ///< Set if mid/side stereo is used for each scalefactor window band
 230     // shared
 231     SingleChannelElement ch[2];
 232     // CCE specific
 233     ChannelCoupling coup;
 234 } ChannelElement;
 235
 236 /**
 237  * main AAC context
 238  */
 239 typedef struct {
 240     AVCodecContext * avccontext;
 241
 242     MPEG4AudioConfig m4ac;
 243
 244     int is_saved;                 ///< Set if elements have stored overlap from previous frame.
 245     DynamicRangeControl che_drc;
 246
 247     /**
 248      * @defgroup elements Channel element related data.
 249      * @{
 250      */
 251     enum ChannelPosition che_pos[4][MAX_ELEM_ID]; /**< channel element channel mapping with the
 252                                                    *   first index as the first 4 raw data block types
 253                                                    */
 254     ChannelElement * che[4][MAX_ELEM_ID];
 255     ChannelElement * tag_che_map[4][MAX_ELEM_ID];
 256     int tags_mapped;
 257     /** @} */
 258
 259     /**
 260      * @defgroup temporary aligned temporary buffers (We do not want to have these on the stack.)
 261      * @{
 262      */
 263     DECLARE_ALIGNED_16(float, buf_mdct[1024]);
 264     /** @} */
 265
 266     /**
 267      * @defgroup tables   Computed / set up during initialization.
 268      * @{
 269      */
 270     FFTContext mdct;
 271     FFTContext mdct_small;
 272     DSPContext dsp;
 273     int random_state;
 274     /** @} */
 275
 276     /**
 277      * @defgroup output   Members used for output interleaving.
 278      * @{
 279      */
 280     float *output_data[MAX_CHANNELS];                 ///< Points to each element's 'ret' buffer (PCM output).
 281     float add_bias;                                   ///< offset for dsp.float_to_int16
 282     float sf_scale;                                   ///< Pre-scale for correct IMDCT and dsp.float_to_int16.
 283     int sf_offset;                                    ///< offset into pow2sf_tab as appropriate for dsp.float_to_int16
 284     /** @} */
 285
 286     DECLARE_ALIGNED(16, float, temp[128]);
 287
 288     enum OCStatus output_configured;
 289 } AACContext;
 290
 291 #endif /* AVCODEC_AAC_H */