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 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 34 #include "aac_defines.h" 35 #include "libavutil/float_dsp.h" 36 #include "libavutil/fixed_dsp.h" 37 #include "libavutil/mem_internal.h" 38 #include "avcodec.h" 39 #if !USE_FIXED 40 #include "mdct15.h" 41 #endif 42 #include "fft.h" 43 #include "mpeg4audio.h" 44 #include "sbr.h" 45 46 #include <stdint.h> 47 48 #define MAX_CHANNELS 64 49 #define MAX_ELEM_ID 16 50 51 #define TNS_MAX_ORDER 20 52 #define MAX_LTP_LONG_SFB 40 53 54 #define CLIP_AVOIDANCE_FACTOR 0.95f 55 56 enum RawDataBlockType { 57 TYPE_SCE, 58 TYPE_CPE, 59 TYPE_CCE, 60 TYPE_LFE, 61 TYPE_DSE, 62 TYPE_PCE, 63 TYPE_FIL, 64 TYPE_END, 65 }; 66 67 enum ExtensionPayloadID { 68 EXT_FILL, 69 EXT_FILL_DATA, 70 EXT_DATA_ELEMENT, 71 EXT_DYNAMIC_RANGE = 0xb, 72 EXT_SBR_DATA = 0xd, 73 EXT_SBR_DATA_CRC = 0xe, 74 }; 75 76 enum WindowSequence { 77 ONLY_LONG_SEQUENCE, 78 LONG_START_SEQUENCE, 79 EIGHT_SHORT_SEQUENCE, 80 LONG_STOP_SEQUENCE, 81 }; 82 83 enum BandType { 84 ZERO_BT = 0, ///< Scalefactors and spectral data are all zero. 85 FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word. 86 ESC_BT = 11, ///< Spectral data are coded with an escape sequence. 87 RESERVED_BT = 12, ///< Band types following are encoded differently from others. 88 NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream. 89 INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions (out of phase). 90 INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions (in phase). 91 }; 92 93 #define IS_CODEBOOK_UNSIGNED(x) (((x) - 1) & 10) 94 95 enum ChannelPosition { 96 AAC_CHANNEL_OFF = 0, 97 AAC_CHANNEL_FRONT = 1, 98 AAC_CHANNEL_SIDE = 2, 99 AAC_CHANNEL_BACK = 3, 100 AAC_CHANNEL_LFE = 4, 101 AAC_CHANNEL_CC = 5, 102 }; 103 104 /** 105 * The point during decoding at which channel coupling is applied. 106 */ 107 enum CouplingPoint { 108 BEFORE_TNS, 109 BETWEEN_TNS_AND_IMDCT, 110 AFTER_IMDCT = 3, 111 }; 112 113 /** 114 * Output configuration status 115 */ 116 enum OCStatus { 117 OC_NONE, ///< Output unconfigured 118 OC_TRIAL_PCE, ///< Output configuration under trial specified by an inband PCE 119 OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header 120 OC_GLOBAL_HDR, ///< Output configuration set in a global header but not yet locked 121 OC_LOCKED, ///< Output configuration locked in place 122 }; 123 124 typedef struct OutputConfiguration { 125 MPEG4AudioConfig m4ac; 126 uint8_t layout_map[MAX_ELEM_ID*4][3]; 127 int layout_map_tags; 128 int channels; 129 uint64_t channel_layout; 130 enum OCStatus status; 131 } OutputConfiguration; 132 133 /** 134 * Predictor State 135 */ 136 typedef struct PredictorState { 137 AAC_FLOAT cor0; 138 AAC_FLOAT cor1; 139 AAC_FLOAT var0; 140 AAC_FLOAT var1; 141 AAC_FLOAT r0; 142 AAC_FLOAT r1; 143 AAC_FLOAT k1; 144 AAC_FLOAT x_est; 145 } PredictorState; 146 147 #define MAX_PREDICTORS 672 148 149 #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times 150 #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0 151 #define SCALE_MAX_POS 255 ///< scalefactor index maximum value 152 #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard 153 #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference 154 155 #define POW_SF2_ZERO 200 ///< ff_aac_pow2sf_tab index corresponding to pow(2, 0); 156 157 #define NOISE_PRE 256 ///< preamble for NOISE_BT, put in bitstream with the first noise band 158 #define NOISE_PRE_BITS 9 ///< length of preamble 159 #define NOISE_OFFSET 90 ///< subtracted from global gain, used as offset for the preamble 160 161 /** 162 * Long Term Prediction 163 */ 164 typedef struct LongTermPrediction { 165 int8_t present; 166 int16_t lag; 167 int coef_idx; 168 INTFLOAT coef; 169 int8_t used[MAX_LTP_LONG_SFB]; 170 } LongTermPrediction; 171 172 /** 173 * Individual Channel Stream 174 */ 175 typedef struct IndividualChannelStream { 176 uint8_t max_sfb; ///< number of scalefactor bands per group 177 enum WindowSequence window_sequence[2]; 178 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sine window. 179 int num_window_groups; 180 uint8_t group_len[8]; 181 LongTermPrediction ltp; 182 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window 183 const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window 184 int num_swb; ///< number of scalefactor window bands 185 int num_windows; 186 int tns_max_bands; 187 int predictor_present; 188 int predictor_initialized; 189 int predictor_reset_group; 190 int predictor_reset_count[31]; ///< used by encoder to count prediction resets 191 uint8_t prediction_used[41]; 192 uint8_t window_clipping[8]; ///< set if a certain window is near clipping 193 float clip_avoidance_factor; ///< set if any window is near clipping to the necessary atennuation factor to avoid it 194 } IndividualChannelStream; 195 196 /** 197 * Temporal Noise Shaping 198 */ 199 typedef struct TemporalNoiseShaping { 200 int present; 201 int n_filt[8]; 202 int length[8][4]; 203 int direction[8][4]; 204 int order[8][4]; 205 int coef_idx[8][4][TNS_MAX_ORDER]; 206 INTFLOAT coef[8][4][TNS_MAX_ORDER]; 207 } TemporalNoiseShaping; 208 209 /** 210 * Dynamic Range Control - decoded from the bitstream but not processed further. 211 */ 212 typedef struct DynamicRangeControl { 213 int pce_instance_tag; ///< Indicates with which program the DRC info is associated. 214 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative 215 int dyn_rng_ctl[17]; ///< DRC magnitude information 216 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing. 217 int band_incr; ///< Number of DRC bands greater than 1 having DRC info. 218 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain. 219 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines. 220 int prog_ref_level; /**< A reference level for the long-term program audio level for all 221 * channels combined. 222 */ 223 } DynamicRangeControl; 224 225 typedef struct Pulse { 226 int num_pulse; 227 int start; 228 int pos[4]; 229 int amp[4]; 230 } Pulse; 231 232 /** 233 * coupling parameters 234 */ 235 typedef struct ChannelCoupling { 236 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied. 237 int num_coupled; ///< number of target elements 238 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE. 239 int id_select[8]; ///< element id 240 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel; 241 * [2] list of gains for left channel; [3] lists of gains for both channels 242 */ 243 INTFLOAT gain[16][120]; 244 } ChannelCoupling; 245 246 /** 247 * Single Channel Element - used for both SCE and LFE elements. 248 */ 249 typedef struct SingleChannelElement { 250 IndividualChannelStream ics; 251 TemporalNoiseShaping tns; 252 Pulse pulse; 253 enum BandType band_type[128]; ///< band types 254 enum BandType band_alt[128]; ///< alternative band type (used by encoder) 255 int band_type_run_end[120]; ///< band type run end points 256 INTFLOAT sf[120]; ///< scalefactors 257 int sf_idx[128]; ///< scalefactor indices (used by encoder) 258 uint8_t zeroes[128]; ///< band is not coded (used by encoder) 259 uint8_t can_pns[128]; ///< band is allowed to PNS (informative) 260 float is_ener[128]; ///< Intensity stereo pos (used by encoder) 261 float pns_ener[128]; ///< Noise energy values (used by encoder) 262 DECLARE_ALIGNED(32, INTFLOAT, pcoeffs)[1024]; ///< coefficients for IMDCT, pristine 263 DECLARE_ALIGNED(32, INTFLOAT, coeffs)[1024]; ///< coefficients for IMDCT, maybe processed 264 DECLARE_ALIGNED(32, INTFLOAT, saved)[1536]; ///< overlap 265 DECLARE_ALIGNED(32, INTFLOAT, ret_buf)[2048]; ///< PCM output buffer 266 DECLARE_ALIGNED(16, INTFLOAT, ltp_state)[3072]; ///< time signal for LTP 267 DECLARE_ALIGNED(32, AAC_FLOAT, lcoeffs)[1024]; ///< MDCT of LTP coefficients (used by encoder) 268 DECLARE_ALIGNED(32, AAC_FLOAT, prcoeffs)[1024]; ///< Main prediction coefs (used by encoder) 269 PredictorState predictor_state[MAX_PREDICTORS]; 270 INTFLOAT *ret; ///< PCM output 271 } SingleChannelElement; 272 273 /** 274 * channel element - generic struct for SCE/CPE/CCE/LFE 275 */ 276 typedef struct ChannelElement { 277 int present; 278 // CPE specific 279 int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream. 280 int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder) 281 uint8_t is_mode; ///< Set if any bands have been encoded using intensity stereo (used by encoder) 282 uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band 283 uint8_t is_mask[128]; ///< Set if intensity stereo is used (used by encoder) 284 // shared 285 SingleChannelElement ch[2]; 286 // CCE specific 287 ChannelCoupling coup; 288 SpectralBandReplication sbr; 289 } ChannelElement; 290 291 /** 292 * main AAC context 293 */ 294 struct AACContext { 295 AVClass *class; 296 AVCodecContext *avctx; 297 AVFrame *frame; 298 299 int is_saved; ///< Set if elements have stored overlap from previous frame. 300 DynamicRangeControl che_drc; 301 302 /** 303 * @name Channel element related data 304 * @{ 305 */ 306 ChannelElement *che[4][MAX_ELEM_ID]; 307 ChannelElement *tag_che_map[4][MAX_ELEM_ID]; 308 int tags_mapped; 309 int warned_remapping_once; 310 /** @} */ 311 312 /** 313 * @name temporary aligned temporary buffers 314 * (We do not want to have these on the stack.) 315 * @{ 316 */ 317 DECLARE_ALIGNED(32, INTFLOAT, buf_mdct)[1024]; 318 /** @} */ 319 320 /** 321 * @name Computed / set up during initialization 322 * @{ 323 */ 324 FFTContext mdct; 325 FFTContext mdct_small; 326 FFTContext mdct_ld; 327 FFTContext mdct_ltp; 328 #if USE_FIXED 329 AVFixedDSPContext *fdsp; 330 #else 331 MDCT15Context *mdct120; 332 MDCT15Context *mdct480; 333 MDCT15Context *mdct960; 334 AVFloatDSPContext *fdsp; 335 #endif /* USE_FIXED */ 336 int random_state; 337 /** @} */ 338 339 /** 340 * @name Members used for output 341 * @{ 342 */ 343 SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement 344 /** @} */ 345 346 347 /** 348 * @name Japanese DTV specific extension 349 * @{ 350 */ 351 int force_dmono_mode;///< 0->not dmono, 1->use first channel, 2->use second channel 352 int dmono_mode; ///< 0->not dmono, 1->use first channel, 2->use second channel 353 /** @} */ 354 355 DECLARE_ALIGNED(32, INTFLOAT, temp)[128]; 356 357 OutputConfiguration oc[2]; 358 int warned_num_aac_frames; 359 int warned_960_sbr; 360 unsigned warned_71_wide; 361 int warned_gain_control; 362 363 /* aacdec functions pointers */ 364 void (*imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce); 365 void (*apply_ltp)(AACContext *ac, SingleChannelElement *sce); 366 void (*apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns, 367 IndividualChannelStream *ics, int decode); 368 void (*windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out, 369 INTFLOAT *in, IndividualChannelStream *ics); 370 void (*update_ltp)(AACContext *ac, SingleChannelElement *sce); 371 void (*vector_pow43)(int *coefs, int len); 372 void (*subband_scale)(int *dst, int *src, int scale, int offset, int len, void *log_context); 373 374 }; 375 376 void ff_aacdec_init_mips(AACContext *c); 377 378 #endif /* AVCODEC_AAC_H */ 379