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/lfg.h"
  54 #include "ac3.h"
  55 #include "ac3dsp.h"
  56 #include "get_bits.h"
  57 #include "dsputil.h"
  58 #include "fft.h"
  59 #include "fmtconvert.h"
  60
  61 #define AC3_OUTPUT_LFEON  8
  62
  63 #define SPX_MAX_BANDS    17
  64
  65 /** Large enough for maximum possible frame size when the specification limit is ignored */
  66 #define AC3_FRAME_BUFFER_SIZE 32768
  67
  68 typedef struct {
  69     AVClass        *class;                  ///< class for AVOptions
  70     AVCodecContext *avctx;                  ///< parent context
  71     AVFrame frame;                          ///< AVFrame for decoded output
  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_mode;                     ///< bitstream mode                         (bsmod)
  83     int channel_mode;                       ///< channel mode                           (acmod)
  84     int channel_layout;                     ///< channel layout
  85     int lfe_on;                             ///< lfe channel in use
  86     int channel_map;                        ///< custom channel map
  87     int center_mix_level;                   ///< Center mix level index
  88     int surround_mix_level;                 ///< Surround mix level index
  89     int eac3;                               ///< indicates if current frame is E-AC-3
  90 ///@}
  91
  92 ///@name Frame syntax parameters
  93     int snr_offset_strategy;                ///< SNR offset strategy                    (snroffststr)
  94     int block_switch_syntax;                ///< block switch syntax enabled            (blkswe)
  95     int dither_flag_syntax;                 ///< dither flag syntax enabled             (dithflage)
  96     int bit_allocation_syntax;              ///< bit allocation model syntax enabled    (bamode)
  97     int fast_gain_syntax;                   ///< fast gain codes enabled                (frmfgaincode)
  98     int dba_syntax;                         ///< delta bit allocation syntax enabled    (dbaflde)
  99     int skip_syntax;                        ///< skip field syntax enabled              (skipflde)
 100  ///@}
 101
 102 ///@name Standard coupling
 103     int cpl_in_use[AC3_MAX_BLOCKS];         ///< coupling in use                        (cplinu)
 104     int cpl_strategy_exists[AC3_MAX_BLOCKS];///< coupling strategy exists               (cplstre)
 105     int channel_in_cpl[AC3_MAX_CHANNELS];   ///< channel in coupling                    (chincpl)
 106     int phase_flags_in_use;                 ///< phase flags in use                     (phsflginu)
 107     int phase_flags[AC3_MAX_CPL_BANDS];     ///< phase flags                            (phsflg)
 108     int num_cpl_bands;                      ///< number of coupling bands               (ncplbnd)
 109     uint8_t cpl_band_sizes[AC3_MAX_CPL_BANDS]; ///< number of coeffs in each coupling band
 110     int firstchincpl;                       ///< first channel in coupling
 111     int first_cpl_coords[AC3_MAX_CHANNELS]; ///< first coupling coordinates states      (firstcplcos)
 112     int cpl_coords[AC3_MAX_CHANNELS][AC3_MAX_CPL_BANDS]; ///< coupling coordinates      (cplco)
 113 ///@}
 114
 115 ///@name Spectral extension
 116 ///@{
 117     int spx_in_use;                             ///< spectral extension in use              (spxinu)
 118     uint8_t channel_uses_spx[AC3_MAX_CHANNELS]; ///< channel uses spectral extension        (chinspx)
 119     int8_t spx_atten_code[AC3_MAX_CHANNELS];    ///< spx attenuation code                   (spxattencod)
 120     int spx_src_start_freq;                     ///< spx start frequency bin
 121     int spx_dst_end_freq;                       ///< spx end frequency bin
 122     int spx_dst_start_freq;                     ///< spx starting frequency bin for copying (copystartmant)
 123                                                 ///< the copy region ends at the start of the spx region.
 124     int num_spx_bands;                          ///< number of spx bands                    (nspxbnds)
 125     uint8_t spx_band_sizes[SPX_MAX_BANDS];      ///< number of bins in each spx band
 126     uint8_t first_spx_coords[AC3_MAX_CHANNELS]; ///< first spx coordinates states           (firstspxcos)
 127     float spx_noise_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS]; ///< spx noise blending factor  (nblendfact)
 128     float spx_signal_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS];///< spx signal blending factor (sblendfact)
 129 ///@}
 130
 131 ///@name Adaptive hybrid transform
 132     int channel_uses_aht[AC3_MAX_CHANNELS];                         ///< channel AHT in use (chahtinu)
 133     int pre_mantissa[AC3_MAX_CHANNELS][AC3_MAX_COEFS][AC3_MAX_BLOCKS];  ///< pre-IDCT mantissas
 134 ///@}
 135
 136 ///@name Channel
 137     int fbw_channels;                           ///< number of full-bandwidth channels
 138     int channels;                               ///< number of total channels
 139     int lfe_ch;                                 ///< index of LFE channel
 140     float downmix_coeffs[AC3_MAX_CHANNELS][2];  ///< stereo downmix coefficients
 141     int downmixed;                              ///< indicates if coeffs are currently downmixed
 142     int output_mode;                            ///< output channel configuration
 143     int out_channels;                           ///< number of output channels
 144 ///@}
 145
 146 ///@name Dynamic range
 147     float dynamic_range[2];                 ///< dynamic range
 148     float drc_scale;                        ///< percentage of dynamic range compression to be applied
 149 ///@}
 150
 151 ///@name Bandwidth
 152     int start_freq[AC3_MAX_CHANNELS];       ///< start frequency bin                    (strtmant)
 153     int end_freq[AC3_MAX_CHANNELS];         ///< end frequency bin                      (endmant)
 154 ///@}
 155
 156 ///@name Rematrixing
 157     int num_rematrixing_bands;              ///< number of rematrixing bands            (nrematbnd)
 158     int rematrixing_flags[4];               ///< rematrixing flags                      (rematflg)
 159 ///@}
 160
 161 ///@name Exponents
 162     int num_exp_groups[AC3_MAX_CHANNELS];           ///< Number of exponent groups      (nexpgrp)
 163     int8_t dexps[AC3_MAX_CHANNELS][AC3_MAX_COEFS];  ///< decoded exponents
 164     int exp_strategy[AC3_MAX_BLOCKS][AC3_MAX_CHANNELS]; ///< exponent strategies        (expstr)
 165 ///@}
 166
 167 ///@name Bit allocation
 168     AC3BitAllocParameters bit_alloc_params;         ///< bit allocation parameters
 169     int first_cpl_leak;                             ///< first coupling leak state      (firstcplleak)
 170     int snr_offset[AC3_MAX_CHANNELS];               ///< signal-to-noise ratio offsets  (snroffst)
 171     int fast_gain[AC3_MAX_CHANNELS];                ///< fast gain values/SMR's         (fgain)
 172     uint8_t bap[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< bit allocation pointers
 173     int16_t psd[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< scaled exponents
 174     int16_t band_psd[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS]; ///< interpolated exponents
 175     int16_t mask[AC3_MAX_CHANNELS][AC3_CRITICAL_BANDS];     ///< masking curve values
 176     int dba_mode[AC3_MAX_CHANNELS];                 ///< delta bit allocation mode
 177     int dba_nsegs[AC3_MAX_CHANNELS];                ///< number of delta segments
 178     uint8_t dba_offsets[AC3_MAX_CHANNELS][8];       ///< delta segment offsets
 179     uint8_t dba_lengths[AC3_MAX_CHANNELS][8];       ///< delta segment lengths
 180     uint8_t dba_values[AC3_MAX_CHANNELS][8];        ///< delta values for each segment
 181 ///@}
 182
 183 ///@name Zero-mantissa dithering
 184     int dither_flag[AC3_MAX_CHANNELS];      ///< dither flags                           (dithflg)
 185     AVLFG dith_state;                       ///< for dither generation
 186 ///@}
 187
 188 ///@name IMDCT
 189     int block_switch[AC3_MAX_CHANNELS];     ///< block switch flags                     (blksw)
 190     FFTContext imdct_512;                   ///< for 512 sample IMDCT
 191     FFTContext imdct_256;                   ///< for 256 sample IMDCT
 192 ///@}
 193
 194 ///@name Optimization
 195     DSPContext dsp;                         ///< for optimization
 196     AC3DSPContext ac3dsp;
 197     FmtConvertContext fmt_conv;             ///< optimized conversion functions
 198     float mul_bias;                         ///< scaling for float_to_int16 conversion
 199 ///@}
 200
 201 ///@name Aligned arrays
 202     DECLARE_ALIGNED(16, int,   fixed_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS];       ///< fixed-point transform coefficients
 203     DECLARE_ALIGNED(32, float, transform_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS];   ///< transform coefficients
 204     DECLARE_ALIGNED(32, float, delay)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE];             ///< delay - added to the next block
 205     DECLARE_ALIGNED(32, float, window)[AC3_BLOCK_SIZE];                              ///< window coefficients
 206     DECLARE_ALIGNED(32, float, tmp_output)[AC3_BLOCK_SIZE];                          ///< temporary storage for output before windowing
 207     DECLARE_ALIGNED(32, float, output)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE];            ///< output after imdct transform and windowing
 208     DECLARE_ALIGNED(32, uint8_t, input_buffer)[AC3_FRAME_BUFFER_SIZE + FF_INPUT_BUFFER_PADDING_SIZE]; ///< temp buffer to prevent overread
 209 ///@}
 210 } AC3DecodeContext;
 211
 212 /**
 213  * Parse the E-AC-3 frame header.
 214  * This parses both the bit stream info and audio frame header.
 215  */
 216 int ff_eac3_parse_header(AC3DecodeContext *s);
 217
 218 /**
 219  * Decode mantissas in a single channel for the entire frame.
 220  * This is used when AHT mode is enabled.
 221  */
 222 void ff_eac3_decode_transform_coeffs_aht_ch(AC3DecodeContext *s, int ch);
 223
 224 void ff_ac3_downmix_c(float (*samples)[256], float (*matrix)[2],
 225                       int out_ch, int in_ch, int len);
 226
 227 /**
 228  * Apply spectral extension to each channel by copying lower frequency
 229  * coefficients to higher frequency bins and applying side information to
 230  * approximate the original high frequency signal.
 231  */
 232 void ff_eac3_apply_spectral_extension(AC3DecodeContext *s);
 233
 234 #endif /* AVCODEC_AC3DEC_H */