More OKed sections of AAC decoder code
[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/**
24 * @file aac.h
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 FFMPEG_AAC_H
31#define FFMPEG_AAC_H
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
589ce6e6
RS
48#define IVQUANT_SIZE 1024
49
cc0591da
RS
50enum AudioObjectType {
51 AOT_NULL,
52 // Support? Name
53 AOT_AAC_MAIN, ///< Y Main
54 AOT_AAC_LC, ///< Y Low Complexity
55 AOT_AAC_SSR, ///< N (code in SoC repo) Scalable Sample Rate
56 AOT_AAC_LTP, ///< N (code in SoC repo) Long Term Prediction
57 AOT_SBR, ///< N (in progress) Spectral Band Replication
58 AOT_AAC_SCALABLE, ///< N Scalable
59 AOT_TWINVQ, ///< N Twin Vector Quantizer
60 AOT_CELP, ///< N Code Excited Linear Prediction
61 AOT_HVXC, ///< N Harmonic Vector eXcitation Coding
62 AOT_TTSI = 12, ///< N Text-To-Speech Interface
63 AOT_MAINSYNTH, ///< N Main Synthesis
64 AOT_WAVESYNTH, ///< N Wavetable Synthesis
65 AOT_MIDI, ///< N General MIDI
66 AOT_SAFX, ///< N Algorithmic Synthesis and Audio Effects
67 AOT_ER_AAC_LC, ///< N Error Resilient Low Complexity
68 AOT_ER_AAC_LTP = 19, ///< N Error Resilient Long Term Prediction
69 AOT_ER_AAC_SCALABLE, ///< N Error Resilient Scalable
70 AOT_ER_TWINVQ, ///< N Error Resilient Twin Vector Quantizer
71 AOT_ER_BSAC, ///< N Error Resilient Bit-Sliced Arithmetic Coding
72 AOT_ER_AAC_LD, ///< N Error Resilient Low Delay
73 AOT_ER_CELP, ///< N Error Resilient Code Excited Linear Prediction
74 AOT_ER_HVXC, ///< N Error Resilient Harmonic Vector eXcitation Coding
75 AOT_ER_HILN, ///< N Error Resilient Harmonic and Individual Lines plus Noise
76 AOT_ER_PARAM, ///< N Error Resilient Parametric
77 AOT_SSC, ///< N SinuSoidal Coding
78};
79
9cc04edf
RS
80enum RawDataBlockType {
81 TYPE_SCE,
82 TYPE_CPE,
83 TYPE_CCE,
84 TYPE_LFE,
85 TYPE_DSE,
86 TYPE_PCE,
87 TYPE_FIL,
88 TYPE_END,
89};
90
cc0591da
RS
91enum ExtensionPayloadID {
92 EXT_FILL,
93 EXT_FILL_DATA,
94 EXT_DATA_ELEMENT,
95 EXT_DYNAMIC_RANGE = 0xb,
96 EXT_SBR_DATA = 0xd,
97 EXT_SBR_DATA_CRC = 0xe,
98};
99
589ce6e6
RS
100enum WindowSequence {
101 ONLY_LONG_SEQUENCE,
102 LONG_START_SEQUENCE,
103 EIGHT_SHORT_SEQUENCE,
104 LONG_STOP_SEQUENCE,
105};
106
cc0591da
RS
107enum BandType {
108 ZERO_BT = 0, ///< Scalefactors and spectral data are all zero.
109 FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word.
110 ESC_BT = 11, ///< Spectral data are coded with an escape sequence.
111 NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream.
112 INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions.
113 INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions.
114};
115
116#define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
117
118enum ChannelPosition {
589ce6e6
RS
119 AAC_CHANNEL_FRONT = 1,
120 AAC_CHANNEL_SIDE = 2,
121 AAC_CHANNEL_BACK = 3,
122 AAC_CHANNEL_LFE = 4,
123 AAC_CHANNEL_CC = 5,
124};
125
9cc04edf
RS
126/**
127 * The point during decoding at which channel coupling is applied.
128 */
129enum CouplingPoint {
130 BEFORE_TNS,
131 BETWEEN_TNS_AND_IMDCT,
132 AFTER_IMDCT = 3,
133};
134
135/**
136 * Individual Channel Stream
137 */
62a57fae
RS
138typedef struct {
139 uint8_t max_sfb; ///< number of scalefactor bands per group
140 enum WindowSequence window_sequence[2];
141 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
142 int num_window_groups;
143 uint8_t group_len[8];
144 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
145 int num_swb; ///< number of scalefactor window bands
146 int num_windows;
147 int tns_max_bands;
148} IndividualChannelStream;
9cc04edf
RS
149
150/**
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;
168 int start;
169 int offset[4];
170 int amp[4];
171} Pulse;
172
173/**
174 * coupling parameters
175 */
176typedef struct {
62a57fae
RS
177 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
178 int num_coupled; ///< number of target elements
179 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
180 int id_select[8]; ///< element id
181 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for left channel;
182 * [2] list of gains for right channel; [3] lists of gains for both channels
183 */
184 float gain[16][120];
185} ChannelCoupling;
186
187/**
188 * Single Channel Element - used for both SCE and LFE elements.
189 */
190typedef struct {
191 IndividualChannelStream ics;
192 TemporalNoiseShaping tns;
193 enum BandType band_type[120]; ///< band types
194 int band_type_run_end[120]; ///< band type run end points
195 float sf[120]; ///< scalefactors
196 DECLARE_ALIGNED_16(float, coeffs[1024]); ///< coefficients for IMDCT
197 DECLARE_ALIGNED_16(float, saved[1024]); ///< overlap
198 DECLARE_ALIGNED_16(float, ret[1024]); ///< PCM output
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
RS
224 /**
225 * @defgroup elements
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
RS
231 ChannelElement * che[4][MAX_ELEM_ID];
232 /** @} */
cc0591da 233
589ce6e6
RS
234 /**
235 * @defgroup tables Computed / set up during initialization.
236 * @{
237 */
238 MDCTContext mdct;
239 MDCTContext mdct_small;
240 DSPContext dsp;
9cc04edf 241 int random_state;
589ce6e6
RS
242 /** @} */
243
244 /**
cc0591da 245 * @defgroup output Members used for output interleaving.
589ce6e6
RS
246 * @{
247 */
cc0591da 248 float *output_data[MAX_CHANNELS]; ///< Points to each element's 'ret' buffer (PCM output).
589ce6e6
RS
249 float add_bias; ///< offset for dsp.float_to_int16
250 float sf_scale; ///< Pre-scale for correct IMDCT and dsp.float_to_int16.
251 int sf_offset; ///< offset into pow2sf_tab as appropriate for dsp.float_to_int16
252 /** @} */
253
254} AACContext;
255
256#endif /* FFMPEG_AAC_H */