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 )
39 #include "libavutil/softfloat.h"
42 #define FFT_FIXED_32 1
44 #define AAC_RENAME(x) x ## _fixed
45 #define AAC_RENAME_32(x) x ## _fixed_32
46 #define AAC_FLOAT SoftFloat
48 #define SHORTFLOAT int16_t
50 #define FIXR(a) ((int)((a) * 1 + 0.5))
51 #define FIXR10(a) ((int)((a) * 1024.0 + 0.5))
52 #define Q23(a) (int)((a) * 8388608.0 + 0.5)
53 #define Q30(x) (int)((x)*1073741824.0 + 0.5)
54 #define Q31(x) (int)((x)*2147483648.0 + 0.5)
56 #define GET_GAIN(x, y) (-(y) << (x)) + 1024
57 #define AAC_MUL26(x, y) (int)(((int64_t)(x) * (y) + 0x2000000) >> 26)
58 #define AAC_MUL30(x, y) (int)(((int64_t)(x) * (y) + 0x20000000) >> 30)
59 #define AAC_MUL31(x, y) (int)(((int64_t)(x) * (y) + 0x40000000) >> 31)
64 #define FFT_FIXED_32 0
66 #define AAC_RENAME(x) x
67 #define AAC_RENAME_32(x) x
68 #define AAC_FLOAT float
69 #define INTFLOAT float
70 #define SHORTFLOAT float
71 #define AAC_SIGNE unsigned
72 #define FIXR(x) ((float)(x))
73 #define FIXR10(x) ((float)(x))
77 #define RANGE15(x) (32768.0 * (x))
78 #define GET_GAIN(x, y) powf((x), -(y))
79 #define AAC_MUL26(x, y) ((x) * (y))
80 #define AAC_MUL30(x, y) ((x) * (y))
81 #define AAC_MUL31(x, y) ((x) * (y))
83 #endif /* USE_FIXED */
85 #include "libavutil/float_dsp.h"
86 #include "libavutil/fixed_dsp.h"
92 #include "mpeg4audio.h"
97 #define MAX_CHANNELS 64
98 #define MAX_ELEM_ID 16
100 #define TNS_MAX_ORDER 20
101 #define MAX_LTP_LONG_SFB 40
103 enum RawDataBlockType {
114 enum ExtensionPayloadID {
118 EXT_DYNAMIC_RANGE = 0xb,
120 EXT_SBR_DATA_CRC = 0xe,
123 enum WindowSequence {
126 EIGHT_SHORT_SEQUENCE,
131 ZERO_BT = 0, ///< Scalefactors and spectral data are all zero.
132 FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word.
133 ESC_BT = 11, ///< Spectral data are coded with an escape sequence.
134 RESERVED_BT = 12, ///< Band types following are encoded differently from others.
135 NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream.
136 INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions (out of phase).
137 INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions (in phase).
140 #define IS_CODEBOOK_UNSIGNED(x) (((x) - 1) & 10)
142 enum ChannelPosition {
144 AAC_CHANNEL_FRONT = 1,
145 AAC_CHANNEL_SIDE = 2,
146 AAC_CHANNEL_BACK = 3,
152 * The point during decoding at which channel coupling is applied.
156 BETWEEN_TNS_AND_IMDCT,
161 * Output configuration status
164 OC_NONE, ///< Output unconfigured
165 OC_TRIAL_PCE, ///< Output configuration under trial specified by an inband PCE
166 OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header
167 OC_GLOBAL_HDR, ///< Output configuration set in a global header but not yet locked
168 OC_LOCKED, ///< Output configuration locked in place
171 typedef struct OutputConfiguration {
172 MPEG4AudioConfig m4ac;
173 uint8_t layout_map[MAX_ELEM_ID*4][3];
176 uint64_t channel_layout;
177 enum OCStatus status;
178 } OutputConfiguration;
183 typedef struct PredictorState {
192 #define MAX_PREDICTORS 672
194 #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times
195 #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0
196 #define SCALE_MAX_POS 255 ///< scalefactor index maximum value
197 #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard
198 #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference
200 #define NOISE_PRE 256 ///< preamble for NOISE_BT, put in bitstream with the first noise band
201 #define NOISE_PRE_BITS 9 ///< length of preamble
202 #define NOISE_OFFSET 90 ///< subtracted from global gain, used as offset for the preamble
205 * Long Term Prediction
207 typedef struct LongTermPrediction {
211 int8_t used[MAX_LTP_LONG_SFB];
212 } LongTermPrediction;
215 * Individual Channel Stream
217 typedef struct IndividualChannelStream {
218 uint8_t max_sfb; ///< number of scalefactor bands per group
219 enum WindowSequence window_sequence[2];
220 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sine window.
221 int num_window_groups;
222 uint8_t group_len[8];
223 LongTermPrediction ltp;
224 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
225 const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window
226 int num_swb; ///< number of scalefactor window bands
229 int predictor_present;
230 int predictor_initialized;
231 int predictor_reset_group;
232 uint8_t prediction_used[41];
233 } IndividualChannelStream;
236 * Temporal Noise Shaping
238 typedef struct TemporalNoiseShaping {
244 INTFLOAT coef[8][4][TNS_MAX_ORDER];
245 } TemporalNoiseShaping;
248 * Dynamic Range Control - decoded from the bitstream but not processed further.
250 typedef struct DynamicRangeControl {
251 int pce_instance_tag; ///< Indicates with which program the DRC info is associated.
252 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative
253 int dyn_rng_ctl[17]; ///< DRC magnitude information
254 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing.
255 int band_incr; ///< Number of DRC bands greater than 1 having DRC info.
256 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain.
257 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
258 int prog_ref_level; /**< A reference level for the long-term program audio level for all
261 } DynamicRangeControl;
263 typedef struct Pulse {
271 * coupling parameters
273 typedef struct ChannelCoupling {
274 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
275 int num_coupled; ///< number of target elements
276 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
277 int id_select[8]; ///< element id
278 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel;
279 * [2] list of gains for left channel; [3] lists of gains for both channels
281 INTFLOAT gain[16][120];
285 * Single Channel Element - used for both SCE and LFE elements.
287 typedef struct SingleChannelElement {
288 IndividualChannelStream ics;
289 TemporalNoiseShaping tns;
291 enum BandType band_type[128]; ///< band types
292 int band_type_run_end[120]; ///< band type run end points
293 INTFLOAT sf[120]; ///< scalefactors
294 int sf_idx[128]; ///< scalefactor indices (used by encoder)
295 uint8_t zeroes[128]; ///< band is not coded (used by encoder)
296 float is_ener[128]; ///< Intensity stereo pos (used by encoder)
297 float pns_ener[128]; ///< Noise energy values (used by encoder)
298 DECLARE_ALIGNED(32, INTFLOAT, pcoeffs)[1024]; ///< coefficients for IMDCT, pristine
299 DECLARE_ALIGNED(32, INTFLOAT, coeffs)[1024]; ///< coefficients for IMDCT, maybe processed
300 DECLARE_ALIGNED(32, INTFLOAT, saved)[1536]; ///< overlap
301 DECLARE_ALIGNED(32, INTFLOAT, ret_buf)[2048]; ///< PCM output buffer
302 DECLARE_ALIGNED(16, INTFLOAT, ltp_state)[3072]; ///< time signal for LTP
303 PredictorState predictor_state[MAX_PREDICTORS];
304 INTFLOAT *ret; ///< PCM output
305 } SingleChannelElement;
308 * channel element - generic struct for SCE/CPE/CCE/LFE
310 typedef struct ChannelElement {
313 int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
314 int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder)
315 uint8_t is_mode; ///< Set if any bands have been encoded using intensity stereo (used by encoder)
316 uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band
317 uint8_t is_mask[128]; ///< Set if intensity stereo is used (used by encoder)
319 SingleChannelElement ch[2];
321 ChannelCoupling coup;
322 SpectralBandReplication sbr;
330 AVCodecContext *avctx;
333 int is_saved; ///< Set if elements have stored overlap from previous frame.
334 DynamicRangeControl che_drc;
337 * @name Channel element related data
340 ChannelElement *che[4][MAX_ELEM_ID];
341 ChannelElement *tag_che_map[4][MAX_ELEM_ID];
343 int warned_remapping_once;
347 * @name temporary aligned temporary buffers
348 * (We do not want to have these on the stack.)
351 DECLARE_ALIGNED(32, INTFLOAT, buf_mdct)[1024];
355 * @name Computed / set up during initialization
359 FFTContext mdct_small;
363 AVFixedDSPContext *fdsp;
365 IMDCT15Context *mdct480;
366 AVFloatDSPContext *fdsp;
367 #endif /* USE_FIXED */
372 * @name Members used for output
375 SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement
380 * @name Japanese DTV specific extension
383 int force_dmono_mode;///< 0->not dmono, 1->use first channel, 2->use second channel
384 int dmono_mode; ///< 0->not dmono, 1->use first channel, 2->use second channel
387 DECLARE_ALIGNED(32, INTFLOAT, temp)[128];
389 OutputConfiguration oc[2];
390 int warned_num_aac_frames;
392 /* aacdec functions pointers */
393 void (*imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce);
394 void (*apply_ltp)(AACContext *ac, SingleChannelElement *sce);
395 void (*apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns,
396 IndividualChannelStream *ics, int decode);
397 void (*windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out,
398 INTFLOAT *in, IndividualChannelStream *ics);
399 void (*update_ltp)(AACContext *ac, SingleChannelElement *sce);
400 void (*vector_pow43)(int *coefs, int len);
401 void (*subband_scale)(int *dst, int *src, int scale, int offset, int len);
405 void ff_aacdec_init_mips(AACContext *c);
407 #endif /* AVCODEC_AAC_H */