Statistics
| Branch: | Revision:

ffmpeg / libavcodec / aac.h @ fe2ff6d2

History | View | Annotate | Download (9.51 KB)

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
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 AVCODEC_AAC_H
31
#define AVCODEC_AAC_H
32

    
33
#include "avcodec.h"
34
#include "dsputil.h"
35
#include "fft.h"
36
#include "mpeg4audio.h"
37
#include "sbr.h"
38
#include "fmtconvert.h"
39

    
40
#include <stdint.h>
41

    
42
#define MAX_CHANNELS 64
43
#define MAX_ELEM_ID 16
44

    
45
#define TNS_MAX_ORDER 20
46

    
47
enum RawDataBlockType {
48
    TYPE_SCE,
49
    TYPE_CPE,
50
    TYPE_CCE,
51
    TYPE_LFE,
52
    TYPE_DSE,
53
    TYPE_PCE,
54
    TYPE_FIL,
55
    TYPE_END,
56
};
57

    
58
enum ExtensionPayloadID {
59
    EXT_FILL,
60
    EXT_FILL_DATA,
61
    EXT_DATA_ELEMENT,
62
    EXT_DYNAMIC_RANGE = 0xb,
63
    EXT_SBR_DATA      = 0xd,
64
    EXT_SBR_DATA_CRC  = 0xe,
65
};
66

    
67
enum WindowSequence {
68
    ONLY_LONG_SEQUENCE,
69
    LONG_START_SEQUENCE,
70
    EIGHT_SHORT_SEQUENCE,
71
    LONG_STOP_SEQUENCE,
72
};
73

    
74
enum BandType {
75
    ZERO_BT        = 0,     ///< Scalefactors and spectral data are all zero.
76
    FIRST_PAIR_BT  = 5,     ///< This and later band types encode two values (rather than four) with one code word.
77
    ESC_BT         = 11,    ///< Spectral data are coded with an escape sequence.
78
    NOISE_BT       = 13,    ///< Spectral data are scaled white noise not coded in the bitstream.
79
    INTENSITY_BT2  = 14,    ///< Scalefactor data are intensity stereo positions.
80
    INTENSITY_BT   = 15,    ///< Scalefactor data are intensity stereo positions.
81
};
82

    
83
#define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
84

    
85
enum ChannelPosition {
86
    AAC_CHANNEL_FRONT = 1,
87
    AAC_CHANNEL_SIDE  = 2,
88
    AAC_CHANNEL_BACK  = 3,
89
    AAC_CHANNEL_LFE   = 4,
90
    AAC_CHANNEL_CC    = 5,
91
};
92

    
93
/**
94
 * The point during decoding at which channel coupling is applied.
95
 */
96
enum CouplingPoint {
97
    BEFORE_TNS,
98
    BETWEEN_TNS_AND_IMDCT,
99
    AFTER_IMDCT = 3,
100
};
101

    
102
/**
103
 * Output configuration status
104
 */
105
enum OCStatus {
106
    OC_NONE,        //< Output unconfigured
107
    OC_TRIAL_PCE,   //< Output configuration under trial specified by an inband PCE
108
    OC_TRIAL_FRAME, //< Output configuration under trial specified by a frame header
109
    OC_GLOBAL_HDR,  //< Output configuration set in a global header but not yet locked
110
    OC_LOCKED,      //< Output configuration locked in place
111
};
112

    
113
/**
114
 * Predictor State
115
 */
116
typedef struct {
117
    float cor0;
118
    float cor1;
119
    float var0;
120
    float var1;
121
    float r0;
122
    float r1;
123
} PredictorState;
124

    
125
#define MAX_PREDICTORS 672
126

    
127
#define SCALE_DIV_512    36    ///< scalefactor difference that corresponds to scale difference in 512 times
128
#define SCALE_ONE_POS   140    ///< scalefactor index that corresponds to scale=1.0
129
#define SCALE_MAX_POS   255    ///< scalefactor index maximum value
130
#define SCALE_MAX_DIFF   60    ///< maximum scalefactor difference allowed by standard
131
#define SCALE_DIFF_ZERO  60    ///< codebook index corresponding to zero scalefactor indices difference
132

    
133
/**
134
 * Individual Channel Stream
135
 */
136
typedef struct {
137
    uint8_t max_sfb;            ///< number of scalefactor bands per group
138
    enum WindowSequence window_sequence[2];
139
    uint8_t use_kb_window[2];   ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
140
    int num_window_groups;
141
    uint8_t group_len[8];
142
    const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
143
    const uint8_t *swb_sizes;   ///< table of scalefactor band sizes for a particular window
144
    int num_swb;                ///< number of scalefactor window bands
145
    int num_windows;
146
    int tns_max_bands;
147
    int predictor_present;
148
    int predictor_initialized;
149
    int predictor_reset_group;
150
    uint8_t prediction_used[41];
151
} IndividualChannelStream;
152

    
153
/**
154
 * Temporal Noise Shaping
155
 */
156
typedef struct {
157
    int present;
158
    int n_filt[8];
159
    int length[8][4];
160
    int direction[8][4];
161
    int order[8][4];
162
    float coef[8][4][TNS_MAX_ORDER];
163
} TemporalNoiseShaping;
164

    
165
/**
166
 * Dynamic Range Control - decoded from the bitstream but not processed further.
167
 */
168
typedef struct {
169
    int pce_instance_tag;                           ///< Indicates with which program the DRC info is associated.
170
    int dyn_rng_sgn[17];                            ///< DRC sign information; 0 - positive, 1 - negative
171
    int dyn_rng_ctl[17];                            ///< DRC magnitude information
172
    int exclude_mask[MAX_CHANNELS];                 ///< Channels to be excluded from DRC processing.
173
    int band_incr;                                  ///< Number of DRC bands greater than 1 having DRC info.
174
    int interpolation_scheme;                       ///< Indicates the interpolation scheme used in the SBR QMF domain.
175
    int band_top[17];                               ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
176
    int prog_ref_level;                             /**< A reference level for the long-term program audio level for all
177
                                                     *   channels combined.
178
                                                     */
179
} DynamicRangeControl;
180

    
181
typedef struct {
182
    int num_pulse;
183
    int start;
184
    int pos[4];
185
    int amp[4];
186
} Pulse;
187

    
188
/**
189
 * coupling parameters
190
 */
191
typedef struct {
192
    enum CouplingPoint coupling_point;  ///< The point during decoding at which coupling is applied.
193
    int num_coupled;       ///< number of target elements
194
    enum RawDataBlockType type[8];   ///< Type of channel element to be coupled - SCE or CPE.
195
    int id_select[8];      ///< element id
196
    int ch_select[8];      /**< [0] shared list of gains; [1] list of gains for right channel;
197
                            *   [2] list of gains for left channel; [3] lists of gains for both channels
198
                            */
199
    float gain[16][120];
200
} ChannelCoupling;
201

    
202
/**
203
 * Single Channel Element - used for both SCE and LFE elements.
204
 */
205
typedef struct {
206
    IndividualChannelStream ics;
207
    TemporalNoiseShaping tns;
208
    Pulse pulse;
209
    enum BandType band_type[128];             ///< band types
210
    int band_type_run_end[120];               ///< band type run end points
211
    float sf[120];                            ///< scalefactors
212
    int sf_idx[128];                          ///< scalefactor indices (used by encoder)
213
    uint8_t zeroes[128];                      ///< band is not coded (used by encoder)
214
    DECLARE_ALIGNED(16, float, coeffs)[1024]; ///< coefficients for IMDCT
215
    DECLARE_ALIGNED(16, float, saved)[1024];  ///< overlap
216
    DECLARE_ALIGNED(16, float, ret)[2048];    ///< PCM output
217
    PredictorState predictor_state[MAX_PREDICTORS];
218
} SingleChannelElement;
219

    
220
/**
221
 * channel element - generic struct for SCE/CPE/CCE/LFE
222
 */
223
typedef struct {
224
    // CPE specific
225
    int common_window;        ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
226
    int     ms_mode;          ///< Signals mid/side stereo flags coding mode (used by encoder)
227
    uint8_t ms_mask[128];     ///< Set if mid/side stereo is used for each scalefactor window band
228
    // shared
229
    SingleChannelElement ch[2];
230
    // CCE specific
231
    ChannelCoupling coup;
232
    SpectralBandReplication sbr;
233
} ChannelElement;
234

    
235
/**
236
 * main AAC context
237
 */
238
typedef struct {
239
    AVCodecContext *avctx;
240

    
241
    MPEG4AudioConfig m4ac;
242

    
243
    int is_saved;                 ///< Set if elements have stored overlap from previous frame.
244
    DynamicRangeControl che_drc;
245

    
246
    /**
247
     * @defgroup elements Channel element related data.
248
     * @{
249
     */
250
    enum ChannelPosition che_pos[4][MAX_ELEM_ID]; /**< channel element channel mapping with the
251
                                                   *   first index as the first 4 raw data block types
252
                                                   */
253
    ChannelElement          *che[4][MAX_ELEM_ID];
254
    ChannelElement  *tag_che_map[4][MAX_ELEM_ID];
255
    int tags_mapped;
256
    /** @} */
257

    
258
    /**
259
     * @defgroup temporary aligned temporary buffers (We do not want to have these on the stack.)
260
     * @{
261
     */
262
    DECLARE_ALIGNED(16, float, buf_mdct)[1024];
263
    /** @} */
264

    
265
    /**
266
     * @defgroup tables   Computed / set up during initialization.
267
     * @{
268
     */
269
    FFTContext mdct;
270
    FFTContext mdct_small;
271
    DSPContext dsp;
272
    FmtConvertContext fmt_conv;
273
    int random_state;
274
    /** @} */
275

    
276
    /**
277
     * @defgroup output   Members used for output interleaving.
278
     * @{
279
     */
280
    float *output_data[MAX_CHANNELS];                 ///< Points to each element's 'ret' buffer (PCM output).
281
    float sf_scale;                                   ///< Pre-scale for correct IMDCT and dsp.float_to_int16.
282
    int sf_offset;                                    ///< offset into pow2sf_tab as appropriate for dsp.float_to_int16
283
    /** @} */
284

    
285
    DECLARE_ALIGNED(16, float, temp)[128];
286

    
287
    enum OCStatus output_configured;
288
} AACContext;
289

    
290
#endif /* AVCODEC_AAC_H */