FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
ac3.h
Go to the documentation of this file.
1 /*
2  * Common code between the AC-3 encoder and decoder
3  * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 /**
23  * @file
24  * Common code between the AC-3 encoder and decoder.
25  */
26 
27 #ifndef AVCODEC_AC3_H
28 #define AVCODEC_AC3_H
29 
30 #define AC3_MAX_CODED_FRAME_SIZE 3840 /* in bytes */
31 #define AC3_MAX_CHANNELS 7 /**< maximum number of channels, including coupling channel */
32 #define CPL_CH 0 /**< coupling channel index */
33 
34 #define AC3_MAX_COEFS 256
35 #define AC3_BLOCK_SIZE 256
36 #define AC3_MAX_BLOCKS 6
37 #define AC3_FRAME_SIZE (AC3_MAX_BLOCKS * 256)
38 #define AC3_WINDOW_SIZE (AC3_BLOCK_SIZE * 2)
39 #define AC3_CRITICAL_BANDS 50
40 #define AC3_MAX_CPL_BANDS 18
41 
42 #include "libavutil/opt.h"
43 #include "avcodec.h"
44 #include "ac3tab.h"
45 
46 /* exponent encoding strategy */
47 #define EXP_REUSE 0
48 #define EXP_NEW 1
49 
50 #define EXP_D15 1
51 #define EXP_D25 2
52 #define EXP_D45 3
53 
54 #ifndef USE_FIXED
55 #define USE_FIXED 0
56 #endif
57 
58 #if USE_FIXED
59 
60 #define FFT_FLOAT 0
61 
62 #define FIXR(a) ((int)((a) * 0 + 0.5))
63 #define FIXR12(a) ((int)((a) * 4096 + 0.5))
64 #define FIXR15(a) ((int)((a) * 32768 + 0.5))
65 #define ROUND15(x) ((x) + 16384) >> 15
66 
67 #define AC3_RENAME(x) x ## _fixed
68 #define AC3_NORM(norm) (1<<24)/(norm)
69 #define AC3_MUL(a,b) ((((int64_t) (a)) * (b))>>12)
70 #define AC3_RANGE(x) (x)
71 #define AC3_DYNAMIC_RANGE(x) (x)
72 #define AC3_SPX_BLEND(x) (x)
73 #define AC3_DYNAMIC_RANGE1 0
74 
75 #define INTFLOAT int
76 #define SHORTFLOAT int16_t
77 
78 #else /* USE_FIXED */
79 
80 #define FIXR(x) ((float)(x))
81 #define FIXR12(x) ((float)(x))
82 #define FIXR15(x) ((float)(x))
83 #define ROUND15(x) (x)
84 
85 #define AC3_RENAME(x) x
86 #define AC3_NORM(norm) (1.0f/(norm))
87 #define AC3_MUL(a,b) ((a) * (b))
88 #define AC3_RANGE(x) (dynamic_range_tab[(x)])
89 #define AC3_DYNAMIC_RANGE(x) (powf(x, s->drc_scale))
90 #define AC3_SPX_BLEND(x) (x)* (1.0f/32)
91 #define AC3_DYNAMIC_RANGE1 1.0f
92 
93 #define INTFLOAT float
94 #define SHORTFLOAT float
95 
96 #endif /* USE_FIXED */
97 
98 #define AC3_LEVEL(x) ROUND15((x) * FIXR15(0.7071067811865476))
99 
100 /* pre-defined gain values */
101 #define LEVEL_PLUS_3DB 1.4142135623730950
102 #define LEVEL_PLUS_1POINT5DB 1.1892071150027209
103 #define LEVEL_MINUS_1POINT5DB 0.8408964152537145
104 #define LEVEL_MINUS_3DB 0.7071067811865476
105 #define LEVEL_MINUS_4POINT5DB 0.5946035575013605
106 #define LEVEL_MINUS_6DB 0.5000000000000000
107 #define LEVEL_MINUS_9DB 0.3535533905932738
108 #define LEVEL_ZERO 0.0000000000000000
109 #define LEVEL_ONE 1.0000000000000000
110 
111 /** Delta bit allocation strategy */
112 typedef enum {
118 
119 /** Channel mode (audio coding mode) */
120 typedef enum {
130 
131 /** Dolby Surround mode */
132 typedef enum AC3DolbySurroundMode {
138 
139 /** Dolby Surround EX mode */
146 
147 /** Dolby Headphone mode */
148 typedef enum AC3DolbyHeadphoneMode {
154 
155 /** Preferred Stereo Downmix mode */
160  AC3_DMIXMOD_DPLII // reserved value in A/52, but used by encoders to indicate DPL2
162 
163 typedef struct AC3BitAllocParameters {
164  int sr_code;
165  int sr_shift;
169 
170 /**
171  * @struct AC3HeaderInfo
172  * Coded AC-3 header values up to the lfeon element, plus derived values.
173  */
174 typedef struct AC3HeaderInfo {
175  /** @name Coded elements
176  * @{
177  */
178  uint16_t sync_word;
179  uint16_t crc1;
186  int substreamid; ///< substream identification
187  int center_mix_level; ///< Center mix level index
188  int surround_mix_level; ///< Surround mix level index
189  uint16_t channel_map;
190  int num_blocks; ///< number of audio blocks
191 #if AV_HAVE_INCOMPATIBLE_LIBAV_ABI
193 #endif
194  /** @} */
195 
196  /** @name Derived values
197  * @{
198  */
200  uint16_t sample_rate;
201  uint32_t bit_rate;
203  uint16_t frame_size;
204  uint64_t channel_layout;
205  /** @} */
206 #if !AV_HAVE_INCOMPATIBLE_LIBAV_ABI
208 #endif
209 } AC3HeaderInfo;
210 
211 typedef enum {
216 } EAC3FrameType;
217 
218 void ff_ac3_common_init(void);
219 
220 /**
221  * Calculate the log power-spectral density of the input signal.
222  * This gives a rough estimate of signal power in the frequency domain by using
223  * the spectral envelope (exponents). The psd is also separately grouped
224  * into critical bands for use in the calculating the masking curve.
225  * 128 units in psd = -6 dB. The dbknee parameter in AC3BitAllocParameters
226  * determines the reference level.
227  *
228  * @param[in] exp frequency coefficient exponents
229  * @param[in] start starting bin location
230  * @param[in] end ending bin location
231  * @param[out] psd signal power for each frequency bin
232  * @param[out] band_psd signal power for each critical band
233  */
234 void ff_ac3_bit_alloc_calc_psd(int8_t *exp, int start, int end, int16_t *psd,
235  int16_t *band_psd);
236 
237 /**
238  * Calculate the masking curve.
239  * First, the excitation is calculated using parameters in s and the signal
240  * power in each critical band. The excitation is compared with a predefined
241  * hearing threshold table to produce the masking curve. If delta bit
242  * allocation information is provided, it is used for adjusting the masking
243  * curve, usually to give a closer match to a better psychoacoustic model.
244  *
245  * @param[in] s adjustable bit allocation parameters
246  * @param[in] band_psd signal power for each critical band
247  * @param[in] start starting bin location
248  * @param[in] end ending bin location
249  * @param[in] fast_gain fast gain (estimated signal-to-mask ratio)
250  * @param[in] is_lfe whether or not the channel being processed is the LFE
251  * @param[in] dba_mode delta bit allocation mode (none, reuse, or new)
252  * @param[in] dba_nsegs number of delta segments
253  * @param[in] dba_offsets location offsets for each segment
254  * @param[in] dba_lengths length of each segment
255  * @param[in] dba_values delta bit allocation for each segment
256  * @param[out] mask calculated masking curve
257  * @return returns 0 for success, non-zero for error
258  */
259 int ff_ac3_bit_alloc_calc_mask(AC3BitAllocParameters *s, int16_t *band_psd,
260  int start, int end, int fast_gain, int is_lfe,
261  int dba_mode, int dba_nsegs, uint8_t *dba_offsets,
262  uint8_t *dba_lengths, uint8_t *dba_values,
263  int16_t *mask);
264 
265 #endif /* AVCODEC_AC3_H */