xref: /third_party/ffmpeg/libavcodec/twinvq.h (revision cabdff1a)
1/*
2 * TwinVQ decoder
3 * Copyright (c) 2009 Vitor Sessak
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#ifndef AVCODEC_TWINVQ_H
23#define AVCODEC_TWINVQ_H
24
25#include <math.h>
26#include <stdint.h>
27
28#include "libavutil/common.h"
29#include "libavutil/float_dsp.h"
30#include "avcodec.h"
31#include "fft.h"
32
33enum TwinVQCodec {
34    TWINVQ_CODEC_VQF,
35    TWINVQ_CODEC_METASOUND,
36};
37
38enum TwinVQFrameType {
39    TWINVQ_FT_SHORT = 0,  ///< Short frame  (divided in n   sub-blocks)
40    TWINVQ_FT_MEDIUM,     ///< Medium frame (divided in m<n sub-blocks)
41    TWINVQ_FT_LONG,       ///< Long frame   (single sub-block + PPC)
42    TWINVQ_FT_PPC,        ///< Periodic Peak Component (part of the long frame)
43};
44
45#define TWINVQ_PPC_SHAPE_CB_SIZE 64
46#define TWINVQ_PPC_SHAPE_LEN_MAX 60
47#define TWINVQ_SUB_AMP_MAX       4500.0
48#define TWINVQ_MULAW_MU          100.0
49#define TWINVQ_GAIN_BITS         8
50#define TWINVQ_AMP_MAX           13000.0
51#define TWINVQ_SUB_GAIN_BITS     5
52#define TWINVQ_WINDOW_TYPE_BITS  4
53#define TWINVQ_PGAIN_MU          200
54#define TWINVQ_LSP_COEFS_MAX     20
55#define TWINVQ_LSP_SPLIT_MAX     4
56#define TWINVQ_CHANNELS_MAX      2
57#define TWINVQ_SUBBLOCKS_MAX     16
58#define TWINVQ_BARK_N_COEF_MAX   4
59
60#define TWINVQ_MAX_FRAMES_PER_PACKET 2
61
62/**
63 * Parameters and tables that are different for each frame type
64 */
65struct TwinVQFrameMode {
66    uint8_t         sub;      ///< Number subblocks in each frame
67    const uint16_t *bark_tab;
68
69    /** number of distinct bark scale envelope values */
70    uint8_t         bark_env_size;
71
72    const int16_t  *bark_cb;    ///< codebook for the bark scale envelope (BSE)
73    uint8_t         bark_n_coef;///< number of BSE CB coefficients to read
74    uint8_t         bark_n_bit; ///< number of bits of the BSE coefs
75
76    //@{
77    /** main codebooks for spectrum data */
78    const int16_t    *cb0;
79    const int16_t    *cb1;
80    //@}
81
82    uint8_t         cb_len_read; ///< number of spectrum coefficients to read
83};
84
85typedef struct TwinVQFrameData {
86    int     window_type;
87    enum TwinVQFrameType ftype;
88
89    uint8_t main_coeffs[1024];
90    uint8_t ppc_coeffs[TWINVQ_PPC_SHAPE_LEN_MAX];
91
92    uint8_t gain_bits[TWINVQ_CHANNELS_MAX];
93    uint8_t sub_gain_bits[TWINVQ_CHANNELS_MAX * TWINVQ_SUBBLOCKS_MAX];
94
95    uint8_t bark1[TWINVQ_CHANNELS_MAX][TWINVQ_SUBBLOCKS_MAX][TWINVQ_BARK_N_COEF_MAX];
96    uint8_t bark_use_hist[TWINVQ_CHANNELS_MAX][TWINVQ_SUBBLOCKS_MAX];
97
98    uint8_t lpc_idx1[TWINVQ_CHANNELS_MAX];
99    uint8_t lpc_idx2[TWINVQ_CHANNELS_MAX][TWINVQ_LSP_SPLIT_MAX];
100    uint8_t lpc_hist_idx[TWINVQ_CHANNELS_MAX];
101
102    int     p_coef[TWINVQ_CHANNELS_MAX];
103    int     g_coef[TWINVQ_CHANNELS_MAX];
104} TwinVQFrameData;
105
106/**
107 * Parameters and tables that are different for every combination of
108 * bitrate/sample rate
109 */
110typedef struct TwinVQModeTab {
111    struct TwinVQFrameMode fmode[3]; ///< frame type-dependent parameters
112
113    uint16_t     size;        ///< frame size in samples
114    uint8_t      n_lsp;       ///< number of lsp coefficients
115    const float *lspcodebook;
116
117    /* number of bits of the different LSP CB coefficients */
118    uint8_t      lsp_bit0;
119    uint8_t      lsp_bit1;
120    uint8_t      lsp_bit2;
121
122    uint8_t      lsp_split;      ///< number of CB entries for the LSP decoding
123    const int16_t *ppc_shape_cb; ///< PPC shape CB
124
125    /** number of the bits for the PPC period value */
126    uint8_t      ppc_period_bit;
127
128    uint8_t      ppc_shape_bit;  ///< number of bits of the PPC shape CB coeffs
129    uint8_t      ppc_shape_len;  ///< size of PPC shape CB
130    uint8_t      pgain_bit;      ///< bits for PPC gain
131
132    /** constant for peak period to peak width conversion */
133    uint16_t     peak_per2wid;
134} TwinVQModeTab;
135
136typedef struct TwinVQContext {
137    AVCodecContext *avctx;
138    AVFloatDSPContext *fdsp;
139    FFTContext mdct_ctx[3];
140
141    const TwinVQModeTab *mtab;
142
143    int is_6kbps;
144
145    // history
146    float lsp_hist[2][20];           ///< LSP coefficients of the last frame
147    float bark_hist[3][2][40];       ///< BSE coefficients of last frame
148
149    // bitstream parameters
150    int16_t permut[4][4096];
151    uint8_t length[4][2];            ///< main codebook stride
152    uint8_t length_change[4];
153    uint8_t bits_main_spec[2][4][2]; ///< bits for the main codebook
154    int bits_main_spec_change[4];
155    int n_div[4];
156
157    float *spectrum;
158    float *curr_frame;               ///< non-interleaved output
159    float *prev_frame;               ///< non-interleaved previous frame
160    int last_block_pos[2];
161    int discarded_packets;
162
163    float *cos_tabs[3];
164
165    // scratch buffers
166    float *tmp_buf;
167
168    int frame_size, frames_per_packet, cur_frame;
169    TwinVQFrameData bits[TWINVQ_MAX_FRAMES_PER_PACKET];
170
171    enum TwinVQCodec codec;
172
173    int (*read_bitstream)(AVCodecContext *avctx, struct TwinVQContext *tctx,
174                          const uint8_t *buf, int buf_size);
175    void (*dec_bark_env)(struct TwinVQContext *tctx, const uint8_t *in,
176                         int use_hist, int ch, float *out, float gain,
177                         enum TwinVQFrameType ftype);
178    void (*decode_ppc)(struct TwinVQContext *tctx, int period_coef, int g_coef,
179                       const float *shape, float *speech);
180} TwinVQContext;
181
182extern const enum TwinVQFrameType ff_twinvq_wtype_to_ftype_table[];
183
184/** @note not speed critical, hence not optimized */
185static inline void twinvq_memset_float(float *buf, float val, int size)
186{
187    while (size--)
188        *buf++ = val;
189}
190
191static inline float twinvq_mulawinv(float y, float clip, float mu)
192{
193    y = av_clipf(y / clip, -1, 1);
194    return clip * FFSIGN(y) * (exp(log(1 + mu) * fabs(y)) - 1) / mu;
195}
196
197int ff_twinvq_decode_frame(AVCodecContext *avctx, AVFrame *frame,
198                           int *got_frame_ptr, AVPacket *avpkt);
199int ff_twinvq_decode_close(AVCodecContext *avctx);
200/** Requires the caller to call ff_twinvq_decode_close() upon failure. */
201int ff_twinvq_decode_init(AVCodecContext *avctx);
202
203#endif /* AVCODEC_TWINVQ_H */
204