blob: bc0aa65b1af3df4654423a79cc67f5e98479d1ee
| 1 | /* |
| 2 | * Copyright (C) 2016 foo86 |
| 3 | * |
| 4 | * This file is part of FFmpeg. |
| 5 | * |
| 6 | * FFmpeg is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU Lesser General Public |
| 8 | * License as published by the Free Software Foundation; either |
| 9 | * version 2.1 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | * FFmpeg is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * Lesser General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU Lesser General Public |
| 17 | * License along with FFmpeg; if not, write to the Free Software |
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 19 | */ |
| 20 | |
| 21 | #ifndef AVCODEC_DCA_XLL_H |
| 22 | #define AVCODEC_DCA_XLL_H |
| 23 | |
| 24 | #include "libavutil/common.h" |
| 25 | #include "libavutil/mem.h" |
| 26 | |
| 27 | #include "avcodec.h" |
| 28 | #include "internal.h" |
| 29 | #include "get_bits.h" |
| 30 | #include "dca.h" |
| 31 | #include "dcadsp.h" |
| 32 | #include "dca_exss.h" |
| 33 | |
| 34 | #define DCA_XLL_CHSETS_MAX 3 |
| 35 | #define DCA_XLL_CHANNELS_MAX 8 |
| 36 | #define DCA_XLL_BANDS_MAX 2 |
| 37 | #define DCA_XLL_ADAPT_PRED_ORDER_MAX 16 |
| 38 | #define DCA_XLL_DECI_HISTORY_MAX 8 |
| 39 | #define DCA_XLL_DMIX_SCALES_MAX ((DCA_XLL_CHSETS_MAX - 1) * DCA_XLL_CHANNELS_MAX) |
| 40 | #define DCA_XLL_DMIX_COEFFS_MAX (DCA_XLL_DMIX_SCALES_MAX * DCA_XLL_CHANNELS_MAX) |
| 41 | #define DCA_XLL_PBR_BUFFER_MAX (240 << 10) |
| 42 | #define DCA_XLL_SAMPLE_BUFFERS_MAX 3 |
| 43 | |
| 44 | typedef struct DCAXllBand { |
| 45 | int decor_enabled; ///< Pairwise channel decorrelation flag |
| 46 | int orig_order[DCA_XLL_CHANNELS_MAX]; ///< Original channel order |
| 47 | int decor_coeff[DCA_XLL_CHANNELS_MAX / 2]; ///< Pairwise channel coefficients |
| 48 | |
| 49 | int adapt_pred_order[DCA_XLL_CHANNELS_MAX]; ///< Adaptive predictor order |
| 50 | int highest_pred_order; ///< Highest adaptive predictor order |
| 51 | int fixed_pred_order[DCA_XLL_CHANNELS_MAX]; ///< Fixed predictor order |
| 52 | int adapt_refl_coeff[DCA_XLL_CHANNELS_MAX][DCA_XLL_ADAPT_PRED_ORDER_MAX]; ///< Adaptive predictor reflection coefficients |
| 53 | |
| 54 | int dmix_embedded; ///< Downmix performed by encoder in frequency band |
| 55 | |
| 56 | int lsb_section_size; ///< Size of LSB section in any segment |
| 57 | int nscalablelsbs[DCA_XLL_CHANNELS_MAX]; ///< Number of bits to represent the samples in LSB part |
| 58 | int bit_width_adjust[DCA_XLL_CHANNELS_MAX]; ///< Number of bits discarded by authoring |
| 59 | |
| 60 | int32_t *msb_sample_buffer[DCA_XLL_CHANNELS_MAX]; ///< MSB sample buffer pointers |
| 61 | int32_t *lsb_sample_buffer[DCA_XLL_CHANNELS_MAX]; ///< LSB sample buffer pointers or NULL |
| 62 | } DCAXllBand; |
| 63 | |
| 64 | typedef struct DCAXllChSet { |
| 65 | // Channel set header |
| 66 | int nchannels; ///< Number of channels in the channel set (N) |
| 67 | int residual_encode; ///< Residual encoding mask (0 - residual, 1 - full channel) |
| 68 | int pcm_bit_res; ///< PCM bit resolution (variable) |
| 69 | int storage_bit_res; ///< Storage bit resolution (16 or 24) |
| 70 | int freq; ///< Original sampling frequency (max. 96000 Hz) |
| 71 | |
| 72 | int primary_chset; ///< Primary channel set flag |
| 73 | int dmix_coeffs_present; ///< Downmix coefficients present in stream |
| 74 | int dmix_embedded; ///< Downmix already performed by encoder |
| 75 | int dmix_type; ///< Primary channel set downmix type |
| 76 | int hier_chset; ///< Whether the channel set is part of a hierarchy |
| 77 | int hier_ofs; ///< Number of preceding channels in a hierarchy (M) |
| 78 | int dmix_coeff[DCA_XLL_DMIX_COEFFS_MAX]; ///< Downmixing coefficients |
| 79 | int dmix_scale[DCA_XLL_DMIX_SCALES_MAX]; ///< Downmixing scales |
| 80 | int dmix_scale_inv[DCA_XLL_DMIX_SCALES_MAX]; ///< Inverse downmixing scales |
| 81 | int ch_mask; ///< Channel mask for set |
| 82 | int ch_remap[DCA_XLL_CHANNELS_MAX]; ///< Channel to speaker map |
| 83 | |
| 84 | int nfreqbands; ///< Number of frequency bands (1 or 2) |
| 85 | int nabits; ///< Number of bits to read bit allocation coding parameter |
| 86 | |
| 87 | DCAXllBand bands[DCA_XLL_BANDS_MAX]; ///< Frequency bands |
| 88 | |
| 89 | // Frequency band coding parameters |
| 90 | int seg_common; ///< Segment type |
| 91 | int rice_code_flag[DCA_XLL_CHANNELS_MAX]; ///< Rice coding flag |
| 92 | int bitalloc_hybrid_linear[DCA_XLL_CHANNELS_MAX]; ///< Binary code length for isolated samples |
| 93 | int bitalloc_part_a[DCA_XLL_CHANNELS_MAX]; ///< Coding parameter for part A of segment |
| 94 | int bitalloc_part_b[DCA_XLL_CHANNELS_MAX]; ///< Coding parameter for part B of segment |
| 95 | int nsamples_part_a[DCA_XLL_CHANNELS_MAX]; ///< Number of samples in part A of segment |
| 96 | |
| 97 | // Decimator history |
| 98 | DECLARE_ALIGNED(32, int32_t, deci_history)[DCA_XLL_CHANNELS_MAX][DCA_XLL_DECI_HISTORY_MAX]; ///< Decimator history for frequency band 1 |
| 99 | |
| 100 | // Sample buffers |
| 101 | unsigned int sample_size[DCA_XLL_SAMPLE_BUFFERS_MAX]; |
| 102 | int32_t *sample_buffer[DCA_XLL_SAMPLE_BUFFERS_MAX]; |
| 103 | } DCAXllChSet; |
| 104 | |
| 105 | typedef struct DCAXllDecoder { |
| 106 | AVCodecContext *avctx; |
| 107 | GetBitContext gb; |
| 108 | |
| 109 | int frame_size; ///< Number of bytes in a lossless frame |
| 110 | int nchsets; ///< Number of channels sets per frame |
| 111 | int nframesegs; ///< Number of segments per frame |
| 112 | int nsegsamples_log2; ///< log2(nsegsamples) |
| 113 | int nsegsamples; ///< Samples in segment per one frequency band |
| 114 | int nframesamples_log2; ///< log2(nframesamples) |
| 115 | int nframesamples; ///< Samples in frame per one frequency band |
| 116 | int seg_size_nbits; ///< Number of bits used to read segment size |
| 117 | int band_crc_present; ///< Presence of CRC16 within each frequency band |
| 118 | int scalable_lsbs; ///< MSB/LSB split flag |
| 119 | int ch_mask_nbits; ///< Number of bits used to read channel mask |
| 120 | int fixed_lsb_width; ///< Fixed LSB width |
| 121 | |
| 122 | DCAXllChSet chset[DCA_XLL_CHSETS_MAX]; ///< Channel sets |
| 123 | |
| 124 | int *navi; ///< NAVI table |
| 125 | unsigned int navi_size; |
| 126 | |
| 127 | int nfreqbands; ///< Highest number of frequency bands |
| 128 | int nchannels; ///< Total number of channels in a hierarchy |
| 129 | int nreschsets; ///< Number of channel sets that have residual encoded channels |
| 130 | int nactivechsets; ///< Number of active channel sets to decode |
| 131 | |
| 132 | int hd_stream_id; ///< Previous DTS-HD stream ID for detecting changes |
| 133 | |
| 134 | uint8_t *pbr_buffer; ///< Peak bit rate (PBR) smoothing buffer |
| 135 | int pbr_length; ///< Length in bytes of data currently buffered |
| 136 | int pbr_delay; ///< Delay in frames before decoding buffered data |
| 137 | |
| 138 | DCADSPContext *dcadsp; |
| 139 | |
| 140 | int output_mask; |
| 141 | int32_t *output_samples[DCA_SPEAKER_COUNT]; |
| 142 | } DCAXllDecoder; |
| 143 | |
| 144 | int ff_dca_xll_parse(DCAXllDecoder *s, uint8_t *data, DCAExssAsset *asset); |
| 145 | int ff_dca_xll_filter_frame(DCAXllDecoder *s, AVFrame *frame); |
| 146 | av_cold void ff_dca_xll_flush(DCAXllDecoder *s); |
| 147 | av_cold void ff_dca_xll_close(DCAXllDecoder *s); |
| 148 | |
| 149 | #endif |
| 150 |