Statistics
| Branch: | Revision:

ffmpeg / libavcodec / aac.h @ bb2d8e9f

History | View | Annotate | Download (9.54 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

    
39
#include <stdint.h>
40

    
41
#define MAX_CHANNELS 64
42
#define MAX_ELEM_ID 16
43

    
44
#define TNS_MAX_ORDER 20
45

    
46
enum RawDataBlockType {
47
    TYPE_SCE,
48
    TYPE_CPE,
49
    TYPE_CCE,
50
    TYPE_LFE,
51
    TYPE_DSE,
52
    TYPE_PCE,
53
    TYPE_FIL,
54
    TYPE_END,
55
};
56

    
57
enum ExtensionPayloadID {
58
    EXT_FILL,
59
    EXT_FILL_DATA,
60
    EXT_DATA_ELEMENT,
61
    EXT_DYNAMIC_RANGE = 0xb,
62
    EXT_SBR_DATA      = 0xd,
63
    EXT_SBR_DATA_CRC  = 0xe,
64
};
65

    
66
enum WindowSequence {
67
    ONLY_LONG_SEQUENCE,
68
    LONG_START_SEQUENCE,
69
    EIGHT_SHORT_SEQUENCE,
70
    LONG_STOP_SEQUENCE,
71
};
72

    
73
enum BandType {
74
    ZERO_BT        = 0,     ///< Scalefactors and spectral data are all zero.
75
    FIRST_PAIR_BT  = 5,     ///< This and later band types encode two values (rather than four) with one code word.
76
    ESC_BT         = 11,    ///< Spectral data are coded with an escape sequence.
77
    NOISE_BT       = 13,    ///< Spectral data are scaled white noise not coded in the bitstream.
78
    INTENSITY_BT2  = 14,    ///< Scalefactor data are intensity stereo positions.
79
    INTENSITY_BT   = 15,    ///< Scalefactor data are intensity stereo positions.
80
};
81

    
82
#define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
83

    
84
enum ChannelPosition {
85
    AAC_CHANNEL_FRONT = 1,
86
    AAC_CHANNEL_SIDE  = 2,
87
    AAC_CHANNEL_BACK  = 3,
88
    AAC_CHANNEL_LFE   = 4,
89
    AAC_CHANNEL_CC    = 5,
90
};
91

    
92
/**
93
 * The point during decoding at which channel coupling is applied.
94
 */
95
enum CouplingPoint {
96
    BEFORE_TNS,
97
    BETWEEN_TNS_AND_IMDCT,
98
    AFTER_IMDCT = 3,
99
};
100

    
101
/**
102
 * Output configuration status
103
 */
104
enum OCStatus {
105
    OC_NONE,        //< Output unconfigured
106
    OC_TRIAL_PCE,   //< Output configuration under trial specified by an inband PCE
107
    OC_TRIAL_FRAME, //< Output configuration under trial specified by a frame header
108
    OC_GLOBAL_HDR,  //< Output configuration set in a global header but not yet locked
109
    OC_LOCKED,      //< Output configuration locked in place
110
};
111

    
112
/**
113
 * Predictor State
114
 */
115
typedef struct {
116
    float cor0;
117
    float cor1;
118
    float var0;
119
    float var1;
120
    float r0;
121
    float r1;
122
} PredictorState;
123

    
124
#define MAX_PREDICTORS 672
125

    
126
#define SCALE_DIV_512    36    ///< scalefactor difference that corresponds to scale difference in 512 times
127
#define SCALE_ONE_POS   140    ///< scalefactor index that corresponds to scale=1.0
128
#define SCALE_MAX_POS   255    ///< scalefactor index maximum value
129
#define SCALE_MAX_DIFF   60    ///< maximum scalefactor difference allowed by standard
130
#define SCALE_DIFF_ZERO  60    ///< codebook index corresponding to zero scalefactor indices difference
131

    
132
/**
133
 * Individual Channel Stream
134
 */
135
typedef struct {
136
    uint8_t max_sfb;            ///< number of scalefactor bands per group
137
    enum WindowSequence window_sequence[2];
138
    uint8_t use_kb_window[2];   ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
139
    int num_window_groups;
140
    uint8_t group_len[8];
141
    const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
142
    const uint8_t *swb_sizes;   ///< table of scalefactor band sizes for a particular window
143
    int num_swb;                ///< number of scalefactor window bands
144
    int num_windows;
145
    int tns_max_bands;
146
    int predictor_present;
147
    int predictor_initialized;
148
    int predictor_reset_group;
149
    uint8_t prediction_used[41];
150
} IndividualChannelStream;
151

    
152
/**
153
 * Temporal Noise Shaping
154
 */
155
typedef struct {
156
    int present;
157
    int n_filt[8];
158
    int length[8][4];
159
    int direction[8][4];
160
    int order[8][4];
161
    float coef[8][4][TNS_MAX_ORDER];
162
} TemporalNoiseShaping;
163

    
164
/**
165
 * Dynamic Range Control - decoded from the bitstream but not processed further.
166
 */
167
typedef struct {
168
    int pce_instance_tag;                           ///< Indicates with which program the DRC info is associated.
169
    int dyn_rng_sgn[17];                            ///< DRC sign information; 0 - positive, 1 - negative
170
    int dyn_rng_ctl[17];                            ///< DRC magnitude information
171
    int exclude_mask[MAX_CHANNELS];                 ///< Channels to be excluded from DRC processing.
172
    int band_incr;                                  ///< Number of DRC bands greater than 1 having DRC info.
173
    int interpolation_scheme;                       ///< Indicates the interpolation scheme used in the SBR QMF domain.
174
    int band_top[17];                               ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
175
    int prog_ref_level;                             /**< A reference level for the long-term program audio level for all
176
                                                     *   channels combined.
177
                                                     */
178
} DynamicRangeControl;
179

    
180
typedef struct {
181
    int num_pulse;
182
    int start;
183
    int pos[4];
184
    int amp[4];
185
} Pulse;
186

    
187
/**
188
 * coupling parameters
189
 */
190
typedef struct {
191
    enum CouplingPoint coupling_point;  ///< The point during decoding at which coupling is applied.
192
    int num_coupled;       ///< number of target elements
193
    enum RawDataBlockType type[8];   ///< Type of channel element to be coupled - SCE or CPE.
194
    int id_select[8];      ///< element id
195
    int ch_select[8];      /**< [0] shared list of gains; [1] list of gains for right channel;
196
                            *   [2] list of gains for left channel; [3] lists of gains for both channels
197
                            */
198
    float gain[16][120];
199
} ChannelCoupling;
200

    
201
/**
202
 * Single Channel Element - used for both SCE and LFE elements.
203
 */
204
typedef struct {
205
    IndividualChannelStream ics;
206
    TemporalNoiseShaping tns;
207
    Pulse pulse;
208
    enum BandType band_type[128];             ///< band types
209
    int band_type_run_end[120];               ///< band type run end points
210
    float sf[120];                            ///< scalefactors
211
    int sf_idx[128];                          ///< scalefactor indices (used by encoder)
212
    uint8_t zeroes[128];                      ///< band is not coded (used by encoder)
213
    DECLARE_ALIGNED(16, float, coeffs)[1024]; ///< coefficients for IMDCT
214
    DECLARE_ALIGNED(16, float, saved)[1024];  ///< overlap
215
    DECLARE_ALIGNED(16, float, ret)[2048];    ///< PCM output
216
    PredictorState predictor_state[MAX_PREDICTORS];
217
} SingleChannelElement;
218

    
219
/**
220
 * channel element - generic struct for SCE/CPE/CCE/LFE
221
 */
222
typedef struct {
223
    // CPE specific
224
    int common_window;        ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
225
    int     ms_mode;          ///< Signals mid/side stereo flags coding mode (used by encoder)
226
    uint8_t ms_mask[128];     ///< Set if mid/side stereo is used for each scalefactor window band
227
    // shared
228
    SingleChannelElement ch[2];
229
    // CCE specific
230
    ChannelCoupling coup;
231
    SpectralBandReplication sbr;
232
} ChannelElement;
233

    
234
/**
235
 * main AAC context
236
 */
237
typedef struct {
238
    AVCodecContext *avctx;
239

    
240
    MPEG4AudioConfig m4ac;
241

    
242
    int is_saved;                 ///< Set if elements have stored overlap from previous frame.
243
    DynamicRangeControl che_drc;
244

    
245
    /**
246
     * @defgroup elements Channel element related data.
247
     * @{
248
     */
249
    enum ChannelPosition che_pos[4][MAX_ELEM_ID]; /**< channel element channel mapping with the
250
                                                   *   first index as the first 4 raw data block types
251
                                                   */
252
    ChannelElement          *che[4][MAX_ELEM_ID];
253
    ChannelElement  *tag_che_map[4][MAX_ELEM_ID];
254
    int tags_mapped;
255
    /** @} */
256

    
257
    /**
258
     * @defgroup temporary aligned temporary buffers (We do not want to have these on the stack.)
259
     * @{
260
     */
261
    DECLARE_ALIGNED(16, float, buf_mdct)[1024];
262
    /** @} */
263

    
264
    /**
265
     * @defgroup tables   Computed / set up during initialization.
266
     * @{
267
     */
268
    FFTContext mdct;
269
    FFTContext mdct_small;
270
    DSPContext dsp;
271
    int random_state;
272
    /** @} */
273

    
274
    /**
275
     * @defgroup output   Members used for output interleaving.
276
     * @{
277
     */
278
    float *output_data[MAX_CHANNELS];                 ///< Points to each element's 'ret' buffer (PCM output).
279
    float add_bias;                                   ///< offset for dsp.float_to_int16
280
    float sf_scale;                                   ///< Pre-scale for correct IMDCT and dsp.float_to_int16.
281
    int sf_offset;                                    ///< offset into pow2sf_tab as appropriate for dsp.float_to_int16
282
    /** @} */
283

    
284
    DECLARE_ALIGNED(16, float, temp)[128];
285

    
286
    enum OCStatus output_configured;
287
} AACContext;
288

    
289
#endif /* AVCODEC_AAC_H */