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 {
146 #define MAX_PREDICTORS 672
148 #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times
149 #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0
150 #define SCALE_MAX_POS 255 ///< scalefactor index maximum value
151 #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard
152 #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference
154 #define NOISE_PRE 256 ///< preamble for NOISE_BT, put in bitstream with the first noise band
155 #define NOISE_PRE_BITS 9 ///< length of preamble
156 #define NOISE_OFFSET 90 ///< subtracted from global gain, used as offset for the preamble
159 * Long Term Prediction
161 typedef struct LongTermPrediction {
165 int8_t used[MAX_LTP_LONG_SFB];
166 } LongTermPrediction;
169 * Individual Channel Stream
171 typedef struct IndividualChannelStream {
172 uint8_t max_sfb; ///< number of scalefactor bands per group
173 enum WindowSequence window_sequence[2];
174 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sine window.
175 int num_window_groups;
176 uint8_t group_len[8];
177 LongTermPrediction ltp;
178 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
179 const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window
180 int num_swb; ///< number of scalefactor window bands
183 int predictor_present;
184 int predictor_initialized;
185 int predictor_reset_group;
186 int predictor_reset_count[31]; ///< used by encoder to count prediction resets
187 uint8_t prediction_used[41];
188 uint8_t window_clipping[8]; ///< set if a certain window is near clipping
189 float clip_avoidance_factor; ///< set if any window is near clipping to the necessary atennuation factor to avoid it
190 } IndividualChannelStream;
193 * Temporal Noise Shaping
195 typedef struct TemporalNoiseShaping {
201 int coef_idx[8][4][TNS_MAX_ORDER];
202 INTFLOAT coef[8][4][TNS_MAX_ORDER];
203 } TemporalNoiseShaping;
206 * Dynamic Range Control - decoded from the bitstream but not processed further.
208 typedef struct DynamicRangeControl {
209 int pce_instance_tag; ///< Indicates with which program the DRC info is associated.
210 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative
211 int dyn_rng_ctl[17]; ///< DRC magnitude information
212 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing.
213 int band_incr; ///< Number of DRC bands greater than 1 having DRC info.
214 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain.
215 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
216 int prog_ref_level; /**< A reference level for the long-term program audio level for all
219 } DynamicRangeControl;
221 typedef struct Pulse {
229 * coupling parameters
231 typedef struct ChannelCoupling {
232 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
233 int num_coupled; ///< number of target elements
234 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
235 int id_select[8]; ///< element id
236 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel;
237 * [2] list of gains for left channel; [3] lists of gains for both channels
239 INTFLOAT gain[16][120];
243 * Single Channel Element - used for both SCE and LFE elements.
245 typedef struct SingleChannelElement {
246 IndividualChannelStream ics;
247 TemporalNoiseShaping tns;
249 enum BandType band_type[128]; ///< band types
250 enum BandType band_alt[128]; ///< alternative band type (used by encoder)
251 int band_type_run_end[120]; ///< band type run end points
252 INTFLOAT sf[120]; ///< scalefactors
253 int sf_idx[128]; ///< scalefactor indices (used by encoder)
254 uint8_t zeroes[128]; ///< band is not coded (used by encoder)
255 float is_ener[128]; ///< Intensity stereo pos (used by encoder)
256 float pns_ener[128]; ///< Noise energy values (used by encoder)
257 DECLARE_ALIGNED(32, INTFLOAT, pcoeffs)[1024]; ///< coefficients for IMDCT, pristine
258 DECLARE_ALIGNED(32, INTFLOAT, coeffs)[1024]; ///< coefficients for IMDCT, maybe processed
259 DECLARE_ALIGNED(32, INTFLOAT, saved)[1536]; ///< overlap
260 DECLARE_ALIGNED(32, INTFLOAT, ret_buf)[2048]; ///< PCM output buffer
261 DECLARE_ALIGNED(16, INTFLOAT, ltp_state)[3072]; ///< time signal for LTP
262 DECLARE_ALIGNED(32, AAC_FLOAT, prcoeffs)[1024]; ///< Main prediction coefs (used by encoder)
263 PredictorState predictor_state[MAX_PREDICTORS];
264 INTFLOAT *ret; ///< PCM output
265 } SingleChannelElement;
268 * channel element - generic struct for SCE/CPE/CCE/LFE
270 typedef struct ChannelElement {
273 int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
274 int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder)
275 uint8_t is_mode; ///< Set if any bands have been encoded using intensity stereo (used by encoder)
276 uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band
277 uint8_t is_mask[128]; ///< Set if intensity stereo is used (used by encoder)
279 SingleChannelElement ch[2];
281 ChannelCoupling coup;
282 SpectralBandReplication sbr;
290 AVCodecContext *avctx;
293 int is_saved; ///< Set if elements have stored overlap from previous frame.
294 DynamicRangeControl che_drc;
297 * @name Channel element related data
300 ChannelElement *che[4][MAX_ELEM_ID];
301 ChannelElement *tag_che_map[4][MAX_ELEM_ID];
303 int warned_remapping_once;
307 * @name temporary aligned temporary buffers
308 * (We do not want to have these on the stack.)
311 DECLARE_ALIGNED(32, INTFLOAT, buf_mdct)[1024];
315 * @name Computed / set up during initialization
319 FFTContext mdct_small;
323 AVFixedDSPContext *fdsp;
325 IMDCT15Context *mdct480;
326 AVFloatDSPContext *fdsp;
327 #endif /* USE_FIXED */
332 * @name Members used for output
335 SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement
340 * @name Japanese DTV specific extension
343 int force_dmono_mode;///< 0->not dmono, 1->use first channel, 2->use second channel
344 int dmono_mode; ///< 0->not dmono, 1->use first channel, 2->use second channel
347 DECLARE_ALIGNED(32, INTFLOAT, temp)[128];
349 OutputConfiguration oc[2];
350 int warned_num_aac_frames;
352 /* aacdec functions pointers */
353 void (*imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce);
354 void (*apply_ltp)(AACContext *ac, SingleChannelElement *sce);
355 void (*apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns,
356 IndividualChannelStream *ics, int decode);
357 void (*windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out,
358 INTFLOAT *in, IndividualChannelStream *ics);
359 void (*update_ltp)(AACContext *ac, SingleChannelElement *sce);
360 void (*vector_pow43)(int *coefs, int len);
361 void (*subband_scale)(int *dst, int *src, int scale, int offset, int len);
365 void ff_aacdec_init_mips(AACContext *c);
367 #endif /* AVCODEC_AAC_H */