blob: 97a2df6b869d76486f6fbcfe707b85d3ba76f863
1 | /* |
2 | * AAC definitions and structures |
3 | * Copyright (c) 2005-2006 Oded Shimon ( ods15 ods15 dyndns org ) |
4 | * Copyright (c) 2006-2007 Maxim Gavrilov ( maxim.gavrilov gmail com ) |
5 | * |
6 | * This file is part of FFmpeg. |
7 | * |
8 | * FFmpeg is free software; you can redistribute it and/or |
9 | * modify it under the terms of the GNU Lesser General Public |
10 | * License as published by the Free Software Foundation; either |
11 | * version 2.1 of the License, or (at your option) any later version. |
12 | * |
13 | * FFmpeg is distributed in the hope that it will be useful, |
14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
16 | * Lesser General Public License for more details. |
17 | * |
18 | * You should have received a copy of the GNU Lesser General Public |
19 | * License along with FFmpeg; if not, write to the Free Software |
20 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
21 | */ |
22 | |
23 | /** |
24 | * @file |
25 | * AAC definitions and structures |
26 | * @author Oded Shimon ( ods15 ods15 dyndns org ) |
27 | * @author Maxim Gavrilov ( maxim.gavrilov gmail com ) |
28 | */ |
29 | |
30 | #ifndef AVCODEC_AAC_H |
31 | #define AVCODEC_AAC_H |
32 | |
33 | |
34 | #include "aac_defines.h" |
35 | #include "libavutil/float_dsp.h" |
36 | #include "libavutil/fixed_dsp.h" |
37 | #include "avcodec.h" |
38 | #if !USE_FIXED |
39 | #include "mdct15.h" |
40 | #endif |
41 | #include "fft.h" |
42 | #include "mpeg4audio.h" |
43 | #include "sbr.h" |
44 | |
45 | #include <stdint.h> |
46 | |
47 | #define MAX_CHANNELS 64 |
48 | #define MAX_ELEM_ID 16 |
49 | |
50 | #define TNS_MAX_ORDER 20 |
51 | #define MAX_LTP_LONG_SFB 40 |
52 | |
53 | #define CLIP_AVOIDANCE_FACTOR 0.95f |
54 | |
55 | enum RawDataBlockType { |
56 | TYPE_SCE, |
57 | TYPE_CPE, |
58 | TYPE_CCE, |
59 | TYPE_LFE, |
60 | TYPE_DSE, |
61 | TYPE_PCE, |
62 | TYPE_FIL, |
63 | TYPE_END, |
64 | }; |
65 | |
66 | enum ExtensionPayloadID { |
67 | EXT_FILL, |
68 | EXT_FILL_DATA, |
69 | EXT_DATA_ELEMENT, |
70 | EXT_DYNAMIC_RANGE = 0xb, |
71 | EXT_SBR_DATA = 0xd, |
72 | EXT_SBR_DATA_CRC = 0xe, |
73 | }; |
74 | |
75 | enum WindowSequence { |
76 | ONLY_LONG_SEQUENCE, |
77 | LONG_START_SEQUENCE, |
78 | EIGHT_SHORT_SEQUENCE, |
79 | LONG_STOP_SEQUENCE, |
80 | }; |
81 | |
82 | enum BandType { |
83 | ZERO_BT = 0, ///< Scalefactors and spectral data are all zero. |
84 | FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word. |
85 | ESC_BT = 11, ///< Spectral data are coded with an escape sequence. |
86 | RESERVED_BT = 12, ///< Band types following are encoded differently from others. |
87 | NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream. |
88 | INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions (out of phase). |
89 | INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions (in phase). |
90 | }; |
91 | |
92 | #define IS_CODEBOOK_UNSIGNED(x) (((x) - 1) & 10) |
93 | |
94 | enum ChannelPosition { |
95 | AAC_CHANNEL_OFF = 0, |
96 | AAC_CHANNEL_FRONT = 1, |
97 | AAC_CHANNEL_SIDE = 2, |
98 | AAC_CHANNEL_BACK = 3, |
99 | AAC_CHANNEL_LFE = 4, |
100 | AAC_CHANNEL_CC = 5, |
101 | }; |
102 | |
103 | /** |
104 | * The point during decoding at which channel coupling is applied. |
105 | */ |
106 | enum CouplingPoint { |
107 | BEFORE_TNS, |
108 | BETWEEN_TNS_AND_IMDCT, |
109 | AFTER_IMDCT = 3, |
110 | }; |
111 | |
112 | /** |
113 | * Output configuration status |
114 | */ |
115 | enum OCStatus { |
116 | OC_NONE, ///< Output unconfigured |
117 | OC_TRIAL_PCE, ///< Output configuration under trial specified by an inband PCE |
118 | OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header |
119 | OC_GLOBAL_HDR, ///< Output configuration set in a global header but not yet locked |
120 | OC_LOCKED, ///< Output configuration locked in place |
121 | }; |
122 | |
123 | typedef struct OutputConfiguration { |
124 | MPEG4AudioConfig m4ac; |
125 | uint8_t layout_map[MAX_ELEM_ID*4][3]; |
126 | int layout_map_tags; |
127 | int channels; |
128 | uint64_t channel_layout; |
129 | enum OCStatus status; |
130 | } OutputConfiguration; |
131 | |
132 | /** |
133 | * Predictor State |
134 | */ |
135 | typedef struct PredictorState { |
136 | AAC_FLOAT cor0; |
137 | AAC_FLOAT cor1; |
138 | AAC_FLOAT var0; |
139 | AAC_FLOAT var1; |
140 | AAC_FLOAT r0; |
141 | AAC_FLOAT r1; |
142 | AAC_FLOAT k1; |
143 | AAC_FLOAT x_est; |
144 | } PredictorState; |
145 | |
146 | #define MAX_PREDICTORS 672 |
147 | |
148 | #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times |
149 | #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0 |
150 | #define SCALE_MAX_POS 255 ///< scalefactor index maximum value |
151 | #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard |
152 | #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference |
153 | |
154 | #define POW_SF2_ZERO 200 ///< ff_aac_pow2sf_tab index corresponding to pow(2, 0); |
155 | |
156 | #define NOISE_PRE 256 ///< preamble for NOISE_BT, put in bitstream with the first noise band |
157 | #define NOISE_PRE_BITS 9 ///< length of preamble |
158 | #define NOISE_OFFSET 90 ///< subtracted from global gain, used as offset for the preamble |
159 | |
160 | /** |
161 | * Long Term Prediction |
162 | */ |
163 | typedef struct LongTermPrediction { |
164 | int8_t present; |
165 | int16_t lag; |
166 | int coef_idx; |
167 | INTFLOAT coef; |
168 | int8_t used[MAX_LTP_LONG_SFB]; |
169 | } LongTermPrediction; |
170 | |
171 | /** |
172 | * Individual Channel Stream |
173 | */ |
174 | typedef struct IndividualChannelStream { |
175 | uint8_t max_sfb; ///< number of scalefactor bands per group |
176 | enum WindowSequence window_sequence[2]; |
177 | uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sine window. |
178 | int num_window_groups; |
179 | uint8_t group_len[8]; |
180 | LongTermPrediction ltp; |
181 | const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window |
182 | const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window |
183 | int num_swb; ///< number of scalefactor window bands |
184 | int num_windows; |
185 | int tns_max_bands; |
186 | int predictor_present; |
187 | int predictor_initialized; |
188 | int predictor_reset_group; |
189 | int predictor_reset_count[31]; ///< used by encoder to count prediction resets |
190 | uint8_t prediction_used[41]; |
191 | uint8_t window_clipping[8]; ///< set if a certain window is near clipping |
192 | float clip_avoidance_factor; ///< set if any window is near clipping to the necessary atennuation factor to avoid it |
193 | } IndividualChannelStream; |
194 | |
195 | /** |
196 | * Temporal Noise Shaping |
197 | */ |
198 | typedef struct TemporalNoiseShaping { |
199 | int present; |
200 | int n_filt[8]; |
201 | int length[8][4]; |
202 | int direction[8][4]; |
203 | int order[8][4]; |
204 | int coef_idx[8][4][TNS_MAX_ORDER]; |
205 | INTFLOAT coef[8][4][TNS_MAX_ORDER]; |
206 | } TemporalNoiseShaping; |
207 | |
208 | /** |
209 | * Dynamic Range Control - decoded from the bitstream but not processed further. |
210 | */ |
211 | typedef struct DynamicRangeControl { |
212 | int pce_instance_tag; ///< Indicates with which program the DRC info is associated. |
213 | int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative |
214 | int dyn_rng_ctl[17]; ///< DRC magnitude information |
215 | int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing. |
216 | int band_incr; ///< Number of DRC bands greater than 1 having DRC info. |
217 | int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain. |
218 | int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines. |
219 | int prog_ref_level; /**< A reference level for the long-term program audio level for all |
220 | * channels combined. |
221 | */ |
222 | } DynamicRangeControl; |
223 | |
224 | typedef struct Pulse { |
225 | int num_pulse; |
226 | int start; |
227 | int pos[4]; |
228 | int amp[4]; |
229 | } Pulse; |
230 | |
231 | /** |
232 | * coupling parameters |
233 | */ |
234 | typedef struct ChannelCoupling { |
235 | enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied. |
236 | int num_coupled; ///< number of target elements |
237 | enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE. |
238 | int id_select[8]; ///< element id |
239 | int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel; |
240 | * [2] list of gains for left channel; [3] lists of gains for both channels |
241 | */ |
242 | INTFLOAT gain[16][120]; |
243 | } ChannelCoupling; |
244 | |
245 | /** |
246 | * Single Channel Element - used for both SCE and LFE elements. |
247 | */ |
248 | typedef struct SingleChannelElement { |
249 | IndividualChannelStream ics; |
250 | TemporalNoiseShaping tns; |
251 | Pulse pulse; |
252 | enum BandType band_type[128]; ///< band types |
253 | enum BandType band_alt[128]; ///< alternative band type (used by encoder) |
254 | int band_type_run_end[120]; ///< band type run end points |
255 | INTFLOAT sf[120]; ///< scalefactors |
256 | int sf_idx[128]; ///< scalefactor indices (used by encoder) |
257 | uint8_t zeroes[128]; ///< band is not coded (used by encoder) |
258 | uint8_t can_pns[128]; ///< band is allowed to PNS (informative) |
259 | float is_ener[128]; ///< Intensity stereo pos (used by encoder) |
260 | float pns_ener[128]; ///< Noise energy values (used by encoder) |
261 | DECLARE_ALIGNED(32, INTFLOAT, pcoeffs)[1024]; ///< coefficients for IMDCT, pristine |
262 | DECLARE_ALIGNED(32, INTFLOAT, coeffs)[1024]; ///< coefficients for IMDCT, maybe processed |
263 | DECLARE_ALIGNED(32, INTFLOAT, saved)[1536]; ///< overlap |
264 | DECLARE_ALIGNED(32, INTFLOAT, ret_buf)[2048]; ///< PCM output buffer |
265 | DECLARE_ALIGNED(16, INTFLOAT, ltp_state)[3072]; ///< time signal for LTP |
266 | DECLARE_ALIGNED(32, AAC_FLOAT, lcoeffs)[1024]; ///< MDCT of LTP coefficients (used by encoder) |
267 | DECLARE_ALIGNED(32, AAC_FLOAT, prcoeffs)[1024]; ///< Main prediction coefs (used by encoder) |
268 | PredictorState predictor_state[MAX_PREDICTORS]; |
269 | INTFLOAT *ret; ///< PCM output |
270 | } SingleChannelElement; |
271 | |
272 | /** |
273 | * channel element - generic struct for SCE/CPE/CCE/LFE |
274 | */ |
275 | typedef struct ChannelElement { |
276 | int present; |
277 | // CPE specific |
278 | int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream. |
279 | int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder) |
280 | uint8_t is_mode; ///< Set if any bands have been encoded using intensity stereo (used by encoder) |
281 | uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band |
282 | uint8_t is_mask[128]; ///< Set if intensity stereo is used (used by encoder) |
283 | // shared |
284 | SingleChannelElement ch[2]; |
285 | // CCE specific |
286 | ChannelCoupling coup; |
287 | SpectralBandReplication sbr; |
288 | } ChannelElement; |
289 | |
290 | /** |
291 | * main AAC context |
292 | */ |
293 | struct AACContext { |
294 | AVClass *class; |
295 | AVCodecContext *avctx; |
296 | AVFrame *frame; |
297 | |
298 | int is_saved; ///< Set if elements have stored overlap from previous frame. |
299 | DynamicRangeControl che_drc; |
300 | |
301 | /** |
302 | * @name Channel element related data |
303 | * @{ |
304 | */ |
305 | ChannelElement *che[4][MAX_ELEM_ID]; |
306 | ChannelElement *tag_che_map[4][MAX_ELEM_ID]; |
307 | int tags_mapped; |
308 | int warned_remapping_once; |
309 | /** @} */ |
310 | |
311 | /** |
312 | * @name temporary aligned temporary buffers |
313 | * (We do not want to have these on the stack.) |
314 | * @{ |
315 | */ |
316 | DECLARE_ALIGNED(32, INTFLOAT, buf_mdct)[1024]; |
317 | /** @} */ |
318 | |
319 | /** |
320 | * @name Computed / set up during initialization |
321 | * @{ |
322 | */ |
323 | FFTContext mdct; |
324 | FFTContext mdct_small; |
325 | FFTContext mdct_ld; |
326 | FFTContext mdct_ltp; |
327 | #if USE_FIXED |
328 | AVFixedDSPContext *fdsp; |
329 | #else |
330 | MDCT15Context *mdct480; |
331 | AVFloatDSPContext *fdsp; |
332 | #endif /* USE_FIXED */ |
333 | int random_state; |
334 | /** @} */ |
335 | |
336 | /** |
337 | * @name Members used for output |
338 | * @{ |
339 | */ |
340 | SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement |
341 | /** @} */ |
342 | |
343 | |
344 | /** |
345 | * @name Japanese DTV specific extension |
346 | * @{ |
347 | */ |
348 | int force_dmono_mode;///< 0->not dmono, 1->use first channel, 2->use second channel |
349 | int dmono_mode; ///< 0->not dmono, 1->use first channel, 2->use second channel |
350 | /** @} */ |
351 | |
352 | DECLARE_ALIGNED(32, INTFLOAT, temp)[128]; |
353 | |
354 | OutputConfiguration oc[2]; |
355 | int warned_num_aac_frames; |
356 | |
357 | /* aacdec functions pointers */ |
358 | void (*imdct_and_windowing)(AACContext *ac, SingleChannelElement *sce); |
359 | void (*apply_ltp)(AACContext *ac, SingleChannelElement *sce); |
360 | void (*apply_tns)(INTFLOAT coef[1024], TemporalNoiseShaping *tns, |
361 | IndividualChannelStream *ics, int decode); |
362 | void (*windowing_and_mdct_ltp)(AACContext *ac, INTFLOAT *out, |
363 | INTFLOAT *in, IndividualChannelStream *ics); |
364 | void (*update_ltp)(AACContext *ac, SingleChannelElement *sce); |
365 | void (*vector_pow43)(int *coefs, int len); |
366 | void (*subband_scale)(int *dst, int *src, int scale, int offset, int len); |
367 | |
368 | }; |
369 | |
370 | void ff_aacdec_init_mips(AACContext *c); |
371 | |
372 | #endif /* AVCODEC_AAC_H */ |
373 |