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 )
6 * This file is part of FFmpeg.
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.
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.
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
25 * AAC definitions and structures
26 * @author Oded Shimon ( ods15 ods15 dyndns org )
27 * @author Maxim Gavrilov ( maxim.gavrilov gmail com )
34 #include "aac_defines.h"
35 #include "libavutil/float_dsp.h"
36 #include "libavutil/fixed_dsp.h"
42 #include "mpeg4audio.h"
47 #define MAX_CHANNELS 64
48 #define MAX_ELEM_ID 16
50 #define TNS_MAX_ORDER 20
51 #define MAX_LTP_LONG_SFB 40
53 #define CLIP_AVOIDANCE_FACTOR 0.95f
55 enum RawDataBlockType {
66 enum ExtensionPayloadID {
70 EXT_DYNAMIC_RANGE = 0xb,
72 EXT_SBR_DATA_CRC = 0xe,
83 ZERO_BT = 0, ///< Scalefactors and spectral data are all zero.
84 FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word.
85 ESC_BT = 11, ///< Spectral data are coded with an escape sequence.
86 RESERVED_BT = 12, ///< Band types following are encoded differently from others.
87 NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream.
88 INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions (out of phase).
89 INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions (in phase).
92 #define IS_CODEBOOK_UNSIGNED(x) (((x) - 1) & 10)
94 enum ChannelPosition {
96 AAC_CHANNEL_FRONT = 1,
104 * The point during decoding at which channel coupling is applied.
108 BETWEEN_TNS_AND_IMDCT,
113 * Output configuration status
116 OC_NONE, ///< Output unconfigured
117 OC_TRIAL_PCE, ///< Output configuration under trial specified by an inband PCE
118 OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header
119 OC_GLOBAL_HDR, ///< Output configuration set in a global header but not yet locked
120 OC_LOCKED, ///< Output configuration locked in place
123 typedef struct OutputConfiguration {
124 MPEG4AudioConfig m4ac;
125 uint8_t layout_map[MAX_ELEM_ID*4][3];
128 uint64_t channel_layout;
129 enum OCStatus status;
130 } OutputConfiguration;
135 typedef struct PredictorState {
144 #define MAX_PREDICTORS 672
146 #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times
147 #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0
148 #define SCALE_MAX_POS 255 ///< scalefactor index maximum value
149 #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard
150 #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference
152 #define NOISE_PRE 256 ///< preamble for NOISE_BT, put in bitstream with the first noise band
153 #define NOISE_PRE_BITS 9 ///< length of preamble
154 #define NOISE_OFFSET 90 ///< subtracted from global gain, used as offset for the preamble
157 * Long Term Prediction
159 typedef struct LongTermPrediction {
163 int8_t used[MAX_LTP_LONG_SFB];
164 } LongTermPrediction;
167 * Individual Channel Stream
169 typedef struct IndividualChannelStream {
170 uint8_t max_sfb; ///< number of scalefactor bands per group
171 enum WindowSequence window_sequence[2];
172 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sine window.
173 int num_window_groups;
174 uint8_t group_len[8];
175 LongTermPrediction ltp;
176 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
177 const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window
178 int num_swb; ///< number of scalefactor window bands
181 int predictor_present;
182 int predictor_initialized;
183 int predictor_reset_group;
184 uint8_t prediction_used[41];
185 uint8_t window_clipping[8]; ///< set if a certain window is near clipping
186 float clip_avoidance_factor; ///< set if any window is near clipping to the necessary atennuation factor to avoid it
187 } IndividualChannelStream;
190 * Temporal Noise Shaping
192 typedef struct TemporalNoiseShaping {
198 INTFLOAT coef[8][4][TNS_MAX_ORDER];
199 } TemporalNoiseShaping;
202 * Dynamic Range Control - decoded from the bitstream but not processed further.
204 typedef struct DynamicRangeControl {
205 int pce_instance_tag; ///< Indicates with which program the DRC info is associated.
206 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative
207 int dyn_rng_ctl[17]; ///< DRC magnitude information
208 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing.
209 int band_incr; ///< Number of DRC bands greater than 1 having DRC info.
210 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain.
211 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
212 int prog_ref_level; /**< A reference level for the long-term program audio level for all
215 } DynamicRangeControl;
217 typedef struct Pulse {
225 * coupling parameters
227 typedef struct ChannelCoupling {
228 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
229 int num_coupled; ///< number of target elements
230 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
231 int id_select[8]; ///< element id
232 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel;
233 * [2] list of gains for left channel; [3] lists of gains for both channels
235 INTFLOAT gain[16][120];
239 * Single Channel Element - used for both SCE and LFE elements.
241 typedef struct SingleChannelElement {
242 IndividualChannelStream ics;
243 TemporalNoiseShaping tns;
245 enum BandType band_type[128]; ///< band types
246 int band_type_run_end[120]; ///< band type run end points
247 INTFLOAT sf[120]; ///< scalefactors
248 int sf_idx[128]; ///< scalefactor indices (used by encoder)
249 uint8_t zeroes[128]; ///< band is not coded (used by encoder)
250 float is_ener[128]; ///< Intensity stereo pos (used by encoder)
251 float pns_ener[128]; ///< Noise energy values (used by encoder)
252 DECLARE_ALIGNED(32, INTFLOAT, pcoeffs)[1024]; ///< coefficients for IMDCT, pristine
253 DECLARE_ALIGNED(32, INTFLOAT, coeffs)[1024]; ///< coefficients for IMDCT, maybe processed
254 DECLARE_ALIGNED(32, INTFLOAT, saved)[1536]; ///< overlap
255 DECLARE_ALIGNED(32, INTFLOAT, ret_buf)[2048]; ///< PCM output buffer
256 DECLARE_ALIGNED(16, INTFLOAT, ltp_state)[3072]; ///< time signal for LTP
257 PredictorState predictor_state[MAX_PREDICTORS];
258 INTFLOAT *ret; ///< PCM output
259 } SingleChannelElement;
262 * channel element - generic struct for SCE/CPE/CCE/LFE
264 typedef struct ChannelElement {
267 int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
268 int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder)
269 uint8_t is_mode; ///< Set if any bands have been encoded using intensity stereo (used by encoder)
270 uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band
271 uint8_t is_mask[128]; ///< Set if intensity stereo is used (used by encoder)
273 SingleChannelElement ch[2];
275 ChannelCoupling coup;
276 SpectralBandReplication sbr;
284 AVCodecContext *avctx;
287 int is_saved; ///< Set if elements have stored overlap from previous frame.
288 DynamicRangeControl che_drc;
291 * @name Channel element related data
294 ChannelElement *che[4][MAX_ELEM_ID];
295 ChannelElement *tag_che_map[4][MAX_ELEM_ID];
297 int warned_remapping_once;
301 * @name temporary aligned temporary buffers
302 * (We do not want to have these on the stack.)
305 DECLARE_ALIGNED(32, INTFLOAT, buf_mdct)[1024];
309 * @name Computed / set up during initialization
313 FFTContext mdct_small;
317 AVFixedDSPContext *fdsp;
319 IMDCT15Context *mdct480;
320 AVFloatDSPContext *fdsp;
321 #endif /* USE_FIXED */
326 * @name Members used for output
329 SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement
334 * @name Japanese DTV specific extension
337 int force_dmono_mode;///< 0->not dmono, 1->use first channel, 2->use second channel
338 int dmono_mode; ///< 0->not dmono, 1->use first channel, 2->use second channel
341 DECLARE_ALIGNED(32, INTFLOAT, temp)[128];
343 OutputConfiguration oc[2];
344 int warned_num_aac_frames;
346 /* aacdec functions pointers */
347 void (*imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce);
348 void (*apply_ltp)(AACContext *ac, SingleChannelElement *sce);
349 void (*apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns,
350 IndividualChannelStream *ics, int decode);
351 void (*windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out,
352 INTFLOAT *in, IndividualChannelStream *ics);
353 void (*update_ltp)(AACContext *ac, SingleChannelElement *sce);
354 void (*vector_pow43)(int *coefs, int len);
355 void (*subband_scale)(int *dst, int *src, int scale, int offset, int len);
359 void ff_aacdec_init_mips(AACContext *c);
361 #endif /* AVCODEC_AAC_H */