libavcodec/ac3dec.h

   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 Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; 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 "bswapdsp.h"
  58 #include "get_bits.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 eac3_frame_dependent_found;         ///< bitstream has E-AC-3 dependent frame(s)
  96     int eac3_subsbtreamid_found;            ///< bitstream has E-AC-3 additional substream(s)
  97     int dolby_surround_mode;                ///< dolby surround mode                    (dsurmod)
  98     int dolby_surround_ex_mode;             ///< dolby surround ex mode                 (dsurexmod)
  99     int dolby_headphone_mode;               ///< dolby headphone mode                   (dheadphonmod)
 100 ///@}
 101
 102 ///@name Frame syntax parameters
 103     int snr_offset_strategy;                ///< SNR offset strategy                    (snroffststr)
 104     int block_switch_syntax;                ///< block switch syntax enabled            (blkswe)
 105     int dither_flag_syntax;                 ///< dither flag syntax enabled             (dithflage)
 106     int bit_allocation_syntax;              ///< bit allocation model syntax enabled    (bamode)
 107     int fast_gain_syntax;                   ///< fast gain codes enabled                (frmfgaincode)
 108     int dba_syntax;                         ///< delta bit allocation syntax enabled    (dbaflde)
 109     int skip_syntax;                        ///< skip field syntax enabled              (skipflde)
 110  ///@}
 111
 112 ///@name Standard coupling
 113     int cpl_in_use[AC3_MAX_BLOCKS];         ///< coupling in use                        (cplinu)
 114     int cpl_strategy_exists[AC3_MAX_BLOCKS];///< coupling strategy exists               (cplstre)
 115     int channel_in_cpl[AC3_MAX_CHANNELS];   ///< channel in coupling                    (chincpl)
 116     int phase_flags_in_use;                 ///< phase flags in use                     (phsflginu)
 117     int phase_flags[AC3_MAX_CPL_BANDS];     ///< phase flags                            (phsflg)
 118     int num_cpl_bands;                      ///< number of coupling bands               (ncplbnd)
 119     uint8_t cpl_band_sizes[AC3_MAX_CPL_BANDS]; ///< number of coeffs in each coupling band
 120     int firstchincpl;                       ///< first channel in coupling
 121     int first_cpl_coords[AC3_MAX_CHANNELS]; ///< first coupling coordinates states      (firstcplcos)
 122     int cpl_coords[AC3_MAX_CHANNELS][AC3_MAX_CPL_BANDS]; ///< coupling coordinates      (cplco)
 123 ///@}
 124
 125 ///@name Spectral extension
 126 ///@{
 127     int spx_in_use;                             ///< spectral extension in use              (spxinu)
 128     uint8_t channel_uses_spx[AC3_MAX_CHANNELS]; ///< channel uses spectral extension        (chinspx)
 129     int8_t spx_atten_code[AC3_MAX_CHANNELS];    ///< spx attenuation code                   (spxattencod)
 130     int spx_src_start_freq;                     ///< spx start frequency bin
 131     int spx_dst_end_freq;                       ///< spx end frequency bin
 132     int spx_dst_start_freq;                     ///< spx starting frequency bin for copying (copystartmant)
 133                                                 ///< the copy region ends at the start of the spx region.
 134     int num_spx_bands;                          ///< number of spx bands                    (nspxbnds)
 135     uint8_t spx_band_sizes[SPX_MAX_BANDS];      ///< number of bins in each spx band
 136     uint8_t first_spx_coords[AC3_MAX_CHANNELS]; ///< first spx coordinates states           (firstspxcos)
 137     float spx_noise_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS]; ///< spx noise blending factor  (nblendfact)
 138     float spx_signal_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS];///< spx signal blending factor (sblendfact)
 139 ///@}
 140
 141 ///@name Adaptive hybrid transform
 142     int channel_uses_aht[AC3_MAX_CHANNELS];                         ///< channel AHT in use (chahtinu)
 143     int pre_mantissa[AC3_MAX_CHANNELS][AC3_MAX_COEFS][AC3_MAX_BLOCKS];  ///< pre-IDCT mantissas
 144 ///@}
 145
 146 ///@name Channel
 147     int fbw_channels;                           ///< number of full-bandwidth channels
 148     int channels;                               ///< number of total channels
 149     int lfe_ch;                                 ///< index of LFE channel
 150     float *downmix_coeffs[2];                   ///< stereo downmix coefficients
 151     int downmixed;                              ///< indicates if coeffs are currently downmixed
 152     int output_mode;                            ///< output channel configuration
 153     int out_channels;                           ///< number of output channels
 154 ///@}
 155
 156 ///@name Dynamic range
 157     float dynamic_range[2];                 ///< dynamic range
 158     float drc_scale;                        ///< percentage of dynamic range compression to be applied
 159 ///@}
 160
 161 ///@name Bandwidth
 162     int start_freq[AC3_MAX_CHANNELS];       ///< start frequency bin                    (strtmant)
 163     int end_freq[AC3_MAX_CHANNELS];         ///< end frequency bin                      (endmant)
 164 ///@}
 165
 166 ///@name Rematrixing
 167     int num_rematrixing_bands;              ///< number of rematrixing bands            (nrematbnd)
 168     int rematrixing_flags[4];               ///< rematrixing flags                      (rematflg)
 169 ///@}
 170
 171 ///@name Exponents
 172     int num_exp_groups[AC3_MAX_CHANNELS];           ///< Number of exponent groups      (nexpgrp)
 173     int8_t dexps[AC3_MAX_CHANNELS][AC3_MAX_COEFS];  ///< decoded exponents
 174     int exp_strategy[AC3_MAX_BLOCKS][AC3_MAX_CHANNELS]; ///< exponent strategies        (expstr)
 175 ///@}
 176
 177 ///@name Bit allocation
 178     AC3BitAllocParameters bit_alloc_params;         ///< bit allocation parameters
 179     int first_cpl_leak;                             ///< first coupling leak state      (firstcplleak)
 180     int snr_offset[AC3_MAX_CHANNELS];               ///< signal-to-noise ratio offsets  (snroffst)
 181     int fast_gain[AC3_MAX_CHANNELS];                ///< fast gain values/SMR's         (fgain)
 182     uint8_t bap[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< bit allocation pointers
 183     int16_t psd[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< scaled exponents
 184     int16_t band_psd[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS]; ///< interpolated exponents
 185     int16_t mask[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS];     ///< masking curve values
 186     int dba_mode[AC3_MAX_CHANNELS];                 ///< delta bit allocation mode
 187     int dba_nsegs[AC3_MAX_CHANNELS];                ///< number of delta segments
 188     uint8_t dba_offsets[AC3_MAX_CHANNELS][8];       ///< delta segment offsets
 189     uint8_t dba_lengths[AC3_MAX_CHANNELS][8];       ///< delta segment lengths
 190     uint8_t dba_values[AC3_MAX_CHANNELS][8];        ///< delta values for each segment
 191 ///@}
 192
 193 ///@name Zero-mantissa dithering
 194     int dither_flag[AC3_MAX_CHANNELS];      ///< dither flags                           (dithflg)
 195     AVLFG dith_state;                       ///< for dither generation
 196 ///@}
 197
 198 ///@name IMDCT
 199     int block_switch[AC3_MAX_CHANNELS];     ///< block switch flags                     (blksw)
 200     FFTContext imdct_512;                   ///< for 512 sample IMDCT
 201     FFTContext imdct_256;                   ///< for 256 sample IMDCT
 202 ///@}
 203
 204 ///@name Optimization
 205     BswapDSPContext bdsp;
 206     AVFloatDSPContext fdsp;
 207     AC3DSPContext ac3dsp;
 208     FmtConvertContext fmt_conv;             ///< optimized conversion functions
 209 ///@}
 210
 211     float *outptr[AC3_MAX_CHANNELS];
 212     float *xcfptr[AC3_MAX_CHANNELS];
 213     float *dlyptr[AC3_MAX_CHANNELS];
 214
 215 ///@name Aligned arrays
 216     DECLARE_ALIGNED(16, int32_t, fixed_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS];     ///< fixed-point transform coefficients
 217     DECLARE_ALIGNED(32, float, transform_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< transform coefficients
 218     DECLARE_ALIGNED(32, float, delay)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE];             ///< delay - added to the next block
 219     DECLARE_ALIGNED(32, float, window)[AC3_BLOCK_SIZE];                              ///< window coefficients
 220     DECLARE_ALIGNED(32, float, tmp_output)[AC3_BLOCK_SIZE];                          ///< temporary storage for output before windowing
 221     DECLARE_ALIGNED(32, float, output)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE];            ///< output after imdct transform and windowing
 222     DECLARE_ALIGNED(32, uint8_t, input_buffer)[AC3_FRAME_BUFFER_SIZE + AV_INPUT_BUFFER_PADDING_SIZE]; ///< temp buffer to prevent overread
 223 ///@}
 224 } AC3DecodeContext;
 225
 226 /**
 227  * Parse the E-AC-3 frame header.
 228  * This parses both the bit stream info and audio frame header.
 229  */
 230 int ff_eac3_parse_header(AC3DecodeContext *s);
 231
 232 /**
 233  * Decode mantissas in a single channel for the entire frame.
 234  * This is used when AHT mode is enabled.
 235  */
 236 void ff_eac3_decode_transform_coeffs_aht_ch(AC3DecodeContext *s, int ch);
 237
 238 /**
 239  * Apply spectral extension to each channel by copying lower frequency
 240  * coefficients to higher frequency bins and applying side information to
 241  * approximate the original high frequency signal.
 242  */
 243 void ff_eac3_apply_spectral_extension(AC3DecodeContext *s);
 244
 245 #endif /* AVCODEC_AC3DEC_H */