FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
ac3dec.h
Go to the documentation of this file.
1 /*
2  * Common code between the AC-3 and E-AC-3 decoders
3  * Copyright (c) 2007 Bartlomiej Wolowiec <bartek.wolowiec@gmail.com>
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 and E-AC-3 decoders.
25  *
26  * Summary of MDCT Coefficient Grouping:
27  * The individual MDCT coefficient indices are often referred to in the
28  * (E-)AC-3 specification as frequency bins. These bins are grouped together
29  * into subbands of 12 coefficients each. The subbands are grouped together
30  * into bands as defined in the bitstream by the band structures, which
31  * determine the number of bands and the size of each band. The full spectrum
32  * of 256 frequency bins is divided into 1 DC bin + 21 subbands = 253 bins.
33  * This system of grouping coefficients is used for channel bandwidth, stereo
34  * rematrixing, channel coupling, enhanced coupling, and spectral extension.
35  *
36  * +-+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+-+
37  * |1| |12| | [12|12|12|12] | | | | | | | | | | | | |3|
38  * +-+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+-+
39  * ~~~ ~~~~ ~~~~~~~~~~~~~ ~~~
40  * | | | |
41  * | | | 3 unused frequency bins--+
42  * | | |
43  * | | +--1 band containing 4 subbands
44  * | |
45  * | +--1 subband of 12 frequency bins
46  * |
47  * +--DC frequency bin
48  */
49 
50 #ifndef AVCODEC_AC3DEC_H
51 #define AVCODEC_AC3DEC_H
52 
53 #include "libavutil/float_dsp.h"
54 #include "libavutil/lfg.h"
55 #include "ac3.h"
56 #include "ac3dsp.h"
57 #include "get_bits.h"
58 #include "dsputil.h"
59 #include "fft.h"
60 #include "fmtconvert.h"
61 
62 #define AC3_OUTPUT_LFEON 8
63 
64 #define SPX_MAX_BANDS 17
65 
66 /** Large enough for maximum possible frame size when the specification limit is ignored */
67 #define AC3_FRAME_BUFFER_SIZE 32768
68 
69 typedef struct AC3DecodeContext {
70  AVClass *class; ///< class for AVOptions
71  AVCodecContext *avctx; ///< parent context
72  GetBitContext gbc; ///< bitstream reader
73 
74 ///@name Bit stream information
75 ///@{
76  int frame_type; ///< frame type (strmtyp)
77  int substreamid; ///< substream identification
78  int frame_size; ///< current frame size, in bytes
79  int bit_rate; ///< stream bit rate, in bits-per-second
80  int sample_rate; ///< sample frequency, in Hz
81  int num_blocks; ///< number of audio blocks
82  int bitstream_id; ///< bitstream id (bsid)
83  int bitstream_mode; ///< bitstream mode (bsmod)
84  int channel_mode; ///< channel mode (acmod)
85  int lfe_on; ///< lfe channel in use
86  int channel_map; ///< custom channel map
87  int preferred_downmix; ///< Preferred 2-channel downmix mode (dmixmod)
88  int center_mix_level; ///< Center mix level index
89  int center_mix_level_ltrt; ///< Center mix level index for Lt/Rt (ltrtcmixlev)
90  int surround_mix_level; ///< Surround mix level index
91  int surround_mix_level_ltrt; ///< Surround mix level index for Lt/Rt (ltrtsurmixlev)
92  int lfe_mix_level_exists; ///< indicates if lfemixlevcod is specified (lfemixlevcode)
93  int lfe_mix_level; ///< LFE mix level index (lfemixlevcod)
94  int eac3; ///< indicates if current frame is E-AC-3
95  int dolby_surround_mode; ///< dolby surround mode (dsurmod)
96  int dolby_surround_ex_mode; ///< dolby surround ex mode (dsurexmod)
97  int dolby_headphone_mode; ///< dolby headphone mode (dheadphonmod)
98 ///@}
99 
105 
106 ///@name Frame syntax parameters
107  int snr_offset_strategy; ///< SNR offset strategy (snroffststr)
108  int block_switch_syntax; ///< block switch syntax enabled (blkswe)
109  int dither_flag_syntax; ///< dither flag syntax enabled (dithflage)
110  int bit_allocation_syntax; ///< bit allocation model syntax enabled (bamode)
111  int fast_gain_syntax; ///< fast gain codes enabled (frmfgaincode)
112  int dba_syntax; ///< delta bit allocation syntax enabled (dbaflde)
113  int skip_syntax; ///< skip field syntax enabled (skipflde)
114  ///@}
115 
116 ///@name Standard coupling
117  int cpl_in_use[AC3_MAX_BLOCKS]; ///< coupling in use (cplinu)
118  int cpl_strategy_exists[AC3_MAX_BLOCKS];///< coupling strategy exists (cplstre)
119  int channel_in_cpl[AC3_MAX_CHANNELS]; ///< channel in coupling (chincpl)
120  int phase_flags_in_use; ///< phase flags in use (phsflginu)
121  int phase_flags[AC3_MAX_CPL_BANDS]; ///< phase flags (phsflg)
122  int num_cpl_bands; ///< number of coupling bands (ncplbnd)
123  uint8_t cpl_band_sizes[AC3_MAX_CPL_BANDS]; ///< number of coeffs in each coupling band
124  int firstchincpl; ///< first channel in coupling
125  int first_cpl_coords[AC3_MAX_CHANNELS]; ///< first coupling coordinates states (firstcplcos)
126  int cpl_coords[AC3_MAX_CHANNELS][AC3_MAX_CPL_BANDS]; ///< coupling coordinates (cplco)
127 ///@}
128 
129 ///@name Spectral extension
130 ///@{
131  int spx_in_use; ///< spectral extension in use (spxinu)
132  uint8_t channel_uses_spx[AC3_MAX_CHANNELS]; ///< channel uses spectral extension (chinspx)
133  int8_t spx_atten_code[AC3_MAX_CHANNELS]; ///< spx attenuation code (spxattencod)
134  int spx_src_start_freq; ///< spx start frequency bin
135  int spx_dst_end_freq; ///< spx end frequency bin
136  int spx_dst_start_freq; ///< spx starting frequency bin for copying (copystartmant)
137  ///< the copy region ends at the start of the spx region.
138  int num_spx_bands; ///< number of spx bands (nspxbnds)
139  uint8_t spx_band_sizes[SPX_MAX_BANDS]; ///< number of bins in each spx band
140  uint8_t first_spx_coords[AC3_MAX_CHANNELS]; ///< first spx coordinates states (firstspxcos)
141  float spx_noise_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS]; ///< spx noise blending factor (nblendfact)
142  float spx_signal_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS];///< spx signal blending factor (sblendfact)
143 ///@}
144 
145 ///@name Adaptive hybrid transform
146  int channel_uses_aht[AC3_MAX_CHANNELS]; ///< channel AHT in use (chahtinu)
148 ///@}
149 
150 ///@name Channel
151  int fbw_channels; ///< number of full-bandwidth channels
152  int channels; ///< number of total channels
153  int lfe_ch; ///< index of LFE channel
154  float downmix_coeffs[AC3_MAX_CHANNELS][2]; ///< stereo downmix coefficients
155  int downmixed; ///< indicates if coeffs are currently downmixed
156  int output_mode; ///< output channel configuration
157  int out_channels; ///< number of output channels
158 ///@}
159 
160 ///@name Dynamic range
161  float dynamic_range[2]; ///< dynamic range
162  float drc_scale; ///< percentage of dynamic range compression to be applied
163 ///@}
164 
165 ///@name Bandwidth
166  int start_freq[AC3_MAX_CHANNELS]; ///< start frequency bin (strtmant)
167  int end_freq[AC3_MAX_CHANNELS]; ///< end frequency bin (endmant)
168 ///@}
169 
170 ///@name Rematrixing
171  int num_rematrixing_bands; ///< number of rematrixing bands (nrematbnd)
172  int rematrixing_flags[4]; ///< rematrixing flags (rematflg)
173 ///@}
174 
175 ///@name Exponents
176  int num_exp_groups[AC3_MAX_CHANNELS]; ///< Number of exponent groups (nexpgrp)
177  int8_t dexps[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< decoded exponents
178  int exp_strategy[AC3_MAX_BLOCKS][AC3_MAX_CHANNELS]; ///< exponent strategies (expstr)
179 ///@}
180 
181 ///@name Bit allocation
182  AC3BitAllocParameters bit_alloc_params; ///< bit allocation parameters
183  int first_cpl_leak; ///< first coupling leak state (firstcplleak)
184  int snr_offset[AC3_MAX_CHANNELS]; ///< signal-to-noise ratio offsets (snroffst)
185  int fast_gain[AC3_MAX_CHANNELS]; ///< fast gain values/SMR's (fgain)
186  uint8_t bap[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< bit allocation pointers
187  int16_t psd[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< scaled exponents
188  int16_t band_psd[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS]; ///< interpolated exponents
189  int16_t mask[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS]; ///< masking curve values
190  int dba_mode[AC3_MAX_CHANNELS]; ///< delta bit allocation mode
191  int dba_nsegs[AC3_MAX_CHANNELS]; ///< number of delta segments
192  uint8_t dba_offsets[AC3_MAX_CHANNELS][8]; ///< delta segment offsets
193  uint8_t dba_lengths[AC3_MAX_CHANNELS][8]; ///< delta segment lengths
194  uint8_t dba_values[AC3_MAX_CHANNELS][8]; ///< delta values for each segment
195 ///@}
196 
197 ///@name Zero-mantissa dithering
198  int dither_flag[AC3_MAX_CHANNELS]; ///< dither flags (dithflg)
199  AVLFG dith_state; ///< for dither generation
200 ///@}
201 
202 ///@name IMDCT
203  int block_switch[AC3_MAX_CHANNELS]; ///< block switch flags (blksw)
204  FFTContext imdct_512; ///< for 512 sample IMDCT
205  FFTContext imdct_256; ///< for 256 sample IMDCT
206 ///@}
207 
208 ///@name Optimization
209  DSPContext dsp; ///< for optimization
212  FmtConvertContext fmt_conv; ///< optimized conversion functions
213 ///@}
214 
218 
219 ///@name Aligned arrays
220  DECLARE_ALIGNED(16, int32_t, fixed_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< fixed-point transform coefficients
221  DECLARE_ALIGNED(32, float, transform_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< transform coefficients
222  DECLARE_ALIGNED(32, float, delay)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< delay - added to the next block
223  DECLARE_ALIGNED(32, float, window)[AC3_BLOCK_SIZE]; ///< window coefficients
224  DECLARE_ALIGNED(32, float, tmp_output)[AC3_BLOCK_SIZE]; ///< temporary storage for output before windowing
225  DECLARE_ALIGNED(32, float, output)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< output after imdct transform and windowing
227 ///@}
229 
230 /**
231  * Parse the E-AC-3 frame header.
232  * This parses both the bit stream info and audio frame header.
233  */
235 
236 /**
237  * Decode mantissas in a single channel for the entire frame.
238  * This is used when AHT mode is enabled.
239  */
241 
242 /**
243  * Apply spectral extension to each channel by copying lower frequency
244  * coefficients to higher frequency bins and applying side information to
245  * approximate the original high frequency signal.
246  */
248 
249 #endif /* AVCODEC_AC3DEC_H */