blob: 23c20891cdf27c7052b310c95c8b8dd6f85bc3eb
1 | /* |
2 | * Opus decoder/demuxer common functions |
3 | * Copyright (c) 2012 Andrew D'Addesio |
4 | * Copyright (c) 2013-2014 Mozilla Corporation |
5 | * Copyright (c) 2016 Rostislav Pehlivanov <atomnuker@gmail.com> |
6 | * |
7 | * This file is part of FFmpeg. |
8 | * |
9 | * FFmpeg is free software; you can redistribute it and/or |
10 | * modify it under the terms of the GNU Lesser General Public |
11 | * License as published by the Free Software Foundation; either |
12 | * version 2.1 of the License, or (at your option) any later version. |
13 | * |
14 | * FFmpeg is distributed in the hope that it will be useful, |
15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
17 | * Lesser General Public License for more details. |
18 | * |
19 | * You should have received a copy of the GNU Lesser General Public |
20 | * License along with FFmpeg; if not, write to the Free Software |
21 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
22 | */ |
23 | |
24 | #ifndef AVCODEC_OPUS_CELT_H |
25 | #define AVCODEC_OPUS_CELT_H |
26 | |
27 | #include <float.h> |
28 | |
29 | #include "opus.h" |
30 | |
31 | #include "mdct15.h" |
32 | #include "libavutil/float_dsp.h" |
33 | #include "libavutil/libm.h" |
34 | |
35 | #define CELT_VECTORS 11 |
36 | #define CELT_ALLOC_STEPS 6 |
37 | #define CELT_FINE_OFFSET 21 |
38 | #define CELT_MAX_FINE_BITS 8 |
39 | #define CELT_NORM_SCALE 16384 |
40 | #define CELT_QTHETA_OFFSET 4 |
41 | #define CELT_QTHETA_OFFSET_TWOPHASE 16 |
42 | #define CELT_EMPH_COEFF 0.85000610f |
43 | #define CELT_POSTFILTER_MINPERIOD 15 |
44 | #define CELT_ENERGY_SILENCE (-28.0f) |
45 | |
46 | enum CeltSpread { |
47 | CELT_SPREAD_NONE, |
48 | CELT_SPREAD_LIGHT, |
49 | CELT_SPREAD_NORMAL, |
50 | CELT_SPREAD_AGGRESSIVE |
51 | }; |
52 | |
53 | enum CeltBlockSize { |
54 | CELT_BLOCK_120, |
55 | CELT_BLOCK_240, |
56 | CELT_BLOCK_480, |
57 | CELT_BLOCK_960, |
58 | |
59 | CELT_BLOCK_NB |
60 | }; |
61 | |
62 | typedef struct CeltBlock { |
63 | float energy[CELT_MAX_BANDS]; |
64 | float lin_energy[CELT_MAX_BANDS]; |
65 | float error_energy[CELT_MAX_BANDS]; |
66 | float prev_energy[2][CELT_MAX_BANDS]; |
67 | |
68 | uint8_t collapse_masks[CELT_MAX_BANDS]; |
69 | |
70 | int band_bins[CELT_MAX_BANDS]; /* MDCT bins per band */ |
71 | float *band_coeffs[CELT_MAX_BANDS]; |
72 | |
73 | /* buffer for mdct output + postfilter */ |
74 | DECLARE_ALIGNED(32, float, buf)[2048]; |
75 | DECLARE_ALIGNED(32, float, coeffs)[CELT_MAX_FRAME_SIZE]; |
76 | |
77 | /* Used by the encoder */ |
78 | DECLARE_ALIGNED(32, float, overlap)[120]; |
79 | DECLARE_ALIGNED(32, float, samples)[CELT_MAX_FRAME_SIZE]; |
80 | |
81 | /* postfilter parameters */ |
82 | int pf_period_new; |
83 | float pf_gains_new[3]; |
84 | int pf_period; |
85 | float pf_gains[3]; |
86 | int pf_period_old; |
87 | float pf_gains_old[3]; |
88 | |
89 | float emph_coeff; |
90 | } CeltBlock; |
91 | |
92 | struct CeltFrame { |
93 | // constant values that do not change during context lifetime |
94 | AVCodecContext *avctx; |
95 | MDCT15Context *imdct[4]; |
96 | AVFloatDSPContext *dsp; |
97 | CeltBlock block[2]; |
98 | int channels; |
99 | int output_channels; |
100 | |
101 | enum CeltBlockSize size; |
102 | int start_band; |
103 | int end_band; |
104 | int coded_bands; |
105 | int transient; |
106 | int intra; |
107 | int pfilter; |
108 | int skip_band_floor; |
109 | int tf_select; |
110 | int alloc_trim; |
111 | int alloc_boost[CELT_MAX_BANDS]; |
112 | int blocks; /* number of iMDCT blocks in the frame, depends on transient */ |
113 | int blocksize; /* size of each block */ |
114 | int silence; /* Frame is filled with silence */ |
115 | int anticollapse_needed; /* Whether to expect an anticollapse bit */ |
116 | int anticollapse; /* Encoded anticollapse bit */ |
117 | int intensity_stereo; |
118 | int dual_stereo; |
119 | int flushed; |
120 | uint32_t seed; |
121 | enum CeltSpread spread; |
122 | |
123 | /* Bit allocation */ |
124 | int framebits; |
125 | int remaining; |
126 | int remaining2; |
127 | int caps [CELT_MAX_BANDS]; |
128 | int fine_bits [CELT_MAX_BANDS]; |
129 | int fine_priority[CELT_MAX_BANDS]; |
130 | int pulses [CELT_MAX_BANDS]; |
131 | int tf_change [CELT_MAX_BANDS]; |
132 | |
133 | DECLARE_ALIGNED(32, float, scratch)[22 * 8]; // MAX(ff_celt_freq_range) * 1<<CELT_MAX_LOG_BLOCKS |
134 | }; |
135 | |
136 | /* LCG for noise generation */ |
137 | static av_always_inline uint32_t celt_rng(CeltFrame *f) |
138 | { |
139 | f->seed = 1664525 * f->seed + 1013904223; |
140 | return f->seed; |
141 | } |
142 | |
143 | static av_always_inline void celt_renormalize_vector(float *X, int N, float gain) |
144 | { |
145 | int i; |
146 | float g = 1e-15f; |
147 | for (i = 0; i < N; i++) |
148 | g += X[i] * X[i]; |
149 | g = gain / sqrtf(g); |
150 | |
151 | for (i = 0; i < N; i++) |
152 | X[i] *= g; |
153 | } |
154 | |
155 | int ff_celt_init(AVCodecContext *avctx, CeltFrame **f, int output_channels); |
156 | |
157 | void ff_celt_free(CeltFrame **f); |
158 | |
159 | void ff_celt_flush(CeltFrame *f); |
160 | |
161 | int ff_celt_decode_frame(CeltFrame *f, OpusRangeCoder *rc, float **output, |
162 | int coded_channels, int frame_size, int startband, int endband); |
163 | |
164 | #endif /* AVCODEC_OPUS_CELT_H */ |
165 |