The internal header libavutil/internal.h should not be used outside libavutil.
[libav.git] / libavcodec / aac.h
CommitLineData
589ce6e6
RS
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/**
bad5537e 24 * @file libavcodec/aac.h
589ce6e6
RS
25 * AAC definitions and structures
26 * @author Oded Shimon ( ods15 ods15 dyndns org )
27 * @author Maxim Gavrilov ( maxim.gavrilov gmail com )
28 */
29
98790382
SS
30#ifndef AVCODEC_AAC_H
31#define AVCODEC_AAC_H
589ce6e6
RS
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
cc0591da 45#define MAX_CHANNELS 64
9cc04edf 46#define MAX_ELEM_ID 16
cc0591da 47
9ffd5c1c 48#define TNS_MAX_ORDER 20
9ffd5c1c 49
9cc04edf
RS
50enum 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
cc0591da
RS
61enum 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
589ce6e6
RS
70enum WindowSequence {
71 ONLY_LONG_SEQUENCE,
72 LONG_START_SEQUENCE,
73 EIGHT_SHORT_SEQUENCE,
74 LONG_STOP_SEQUENCE,
75};
76
cc0591da
RS
77enum 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
88enum ChannelPosition {
589ce6e6
RS
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
9cc04edf
RS
96/**
97 * The point during decoding at which channel coupling is applied.
98 */
99enum CouplingPoint {
100 BEFORE_TNS,
101 BETWEEN_TNS_AND_IMDCT,
102 AFTER_IMDCT = 3,
103};
104
105/**
7633a041
AC
106 * Predictor State
107 */
108typedef struct {
109 float cor0;
110 float cor1;
111 float var0;
112 float var1;
113 float r0;
114 float r1;
115} PredictorState;
116
117#define MAX_PREDICTORS 672
118
119/**
9cc04edf
RS
120 * Individual Channel Stream
121 */
62a57fae
RS
122typedef struct {
123 uint8_t max_sfb; ///< number of scalefactor bands per group
124 enum WindowSequence window_sequence[2];
125 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
126 int num_window_groups;
127 uint8_t group_len[8];
128 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
129 int num_swb; ///< number of scalefactor window bands
130 int num_windows;
131 int tns_max_bands;
7633a041
AC
132 int predictor_present;
133 int predictor_initialized;
134 int predictor_reset_group;
135 uint8_t prediction_used[41];
62a57fae 136} IndividualChannelStream;
9cc04edf
RS
137
138/**
1dece0d2
RS
139 * Temporal Noise Shaping
140 */
141typedef struct {
142 int present;
143 int n_filt[8];
144 int length[8][4];
145 int direction[8][4];
146 int order[8][4];
147 float coef[8][4][TNS_MAX_ORDER];
148} TemporalNoiseShaping;
149
150/**
9cc04edf
RS
151 * Dynamic Range Control - decoded from the bitstream but not processed further.
152 */
153typedef struct {
154 int pce_instance_tag; ///< Indicates with which program the DRC info is associated.
155 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative
156 int dyn_rng_ctl[17]; ///< DRC magnitude information
157 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing.
158 int band_incr; ///< Number of DRC bands greater than 1 having DRC info.
159 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain.
160 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
161 int prog_ref_level; /**< A reference level for the long-term program audio level for all
162 * channels combined.
163 */
164} DynamicRangeControl;
165
cc0591da
RS
166typedef struct {
167 int num_pulse;
848a5815 168 int pos[4];
cc0591da
RS
169 int amp[4];
170} Pulse;
171
172/**
173 * coupling parameters
174 */
175typedef struct {
62a57fae
RS
176 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
177 int num_coupled; ///< number of target elements
178 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
179 int id_select[8]; ///< element id
5c2cb172
AC
180 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel;
181 * [2] list of gains for left channel; [3] lists of gains for both channels
62a57fae
RS
182 */
183 float gain[16][120];
184} ChannelCoupling;
185
186/**
187 * Single Channel Element - used for both SCE and LFE elements.
188 */
189typedef struct {
190 IndividualChannelStream ics;
191 TemporalNoiseShaping tns;
192 enum BandType band_type[120]; ///< band types
193 int band_type_run_end[120]; ///< band type run end points
194 float sf[120]; ///< scalefactors
195 DECLARE_ALIGNED_16(float, coeffs[1024]); ///< coefficients for IMDCT
b0f5852a 196 DECLARE_ALIGNED_16(float, saved[512]); ///< overlap
62a57fae 197 DECLARE_ALIGNED_16(float, ret[1024]); ///< PCM output
aab54133 198 PredictorState predictor_state[MAX_PREDICTORS];
62a57fae
RS
199} SingleChannelElement;
200
201/**
202 * channel element - generic struct for SCE/CPE/CCE/LFE
203 */
204typedef struct {
205 // CPE specific
206 uint8_t ms_mask[120]; ///< Set if mid/side stereo is used for each scalefactor window band
207 // shared
208 SingleChannelElement ch[2];
209 // CCE specific
210 ChannelCoupling coup;
211} ChannelElement;
cc0591da 212
589ce6e6
RS
213/**
214 * main AAC context
215 */
216typedef struct {
217 AVCodecContext * avccontext;
218
cc0591da
RS
219 MPEG4AudioConfig m4ac;
220
221 int is_saved; ///< Set if elements have stored overlap from previous frame.
222 DynamicRangeControl che_drc;
223
9cc04edf 224 /**
b17f1a07 225 * @defgroup elements Channel element related data.
9cc04edf
RS
226 * @{
227 */
cc0591da
RS
228 enum ChannelPosition che_pos[4][MAX_ELEM_ID]; /**< channel element channel mapping with the
229 * first index as the first 4 raw data block types
230 */
9cc04edf 231 ChannelElement * che[4][MAX_ELEM_ID];
bb5c0988
AC
232 ChannelElement * tag_che_map[4][MAX_ELEM_ID];
233 int tags_mapped;
9cc04edf 234 /** @} */
cc0591da 235
589ce6e6 236 /**
c7f4d983
RS
237 * @defgroup temporary aligned temporary buffers (We do not want to have these on the stack.)
238 * @{
239 */
b0f5852a 240 DECLARE_ALIGNED_16(float, buf_mdct[1024]);
c7f4d983
RS
241 /** @} */
242
243 /**
589ce6e6
RS
244 * @defgroup tables Computed / set up during initialization.
245 * @{
246 */
247 MDCTContext mdct;
248 MDCTContext mdct_small;
249 DSPContext dsp;
9cc04edf 250 int random_state;
589ce6e6
RS
251 /** @} */
252
253 /**
cc0591da 254 * @defgroup output Members used for output interleaving.
589ce6e6
RS
255 * @{
256 */
cc0591da 257 float *output_data[MAX_CHANNELS]; ///< Points to each element's 'ret' buffer (PCM output).
589ce6e6
RS
258 float add_bias; ///< offset for dsp.float_to_int16
259 float sf_scale; ///< Pre-scale for correct IMDCT and dsp.float_to_int16.
260 int sf_offset; ///< offset into pow2sf_tab as appropriate for dsp.float_to_int16
261 /** @} */
262
8153f144 263 DECLARE_ALIGNED(16, float, temp[128]);
589ce6e6
RS
264} AACContext;
265
98790382 266#endif /* AVCODEC_AAC_H */