aacdec: use float planar sample format for output
[libav.git] / libavcodec / aac.h
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 Libav.
7 *
8 * Libav 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 * Libav 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 Libav; 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 #include "libavutil/float_dsp.h"
34 #include "avcodec.h"
35 #include "dsputil.h"
36 #include "fft.h"
37 #include "mpeg4audio.h"
38 #include "sbr.h"
39 #include "fmtconvert.h"
40
41 #include <stdint.h>
42
43 #define MAX_CHANNELS 64
44 #define MAX_ELEM_ID 16
45
46 #define TNS_MAX_ORDER 20
47 #define MAX_LTP_LONG_SFB 40
48
49 enum RawDataBlockType {
50 TYPE_SCE,
51 TYPE_CPE,
52 TYPE_CCE,
53 TYPE_LFE,
54 TYPE_DSE,
55 TYPE_PCE,
56 TYPE_FIL,
57 TYPE_END,
58 };
59
60 enum ExtensionPayloadID {
61 EXT_FILL,
62 EXT_FILL_DATA,
63 EXT_DATA_ELEMENT,
64 EXT_DYNAMIC_RANGE = 0xb,
65 EXT_SBR_DATA = 0xd,
66 EXT_SBR_DATA_CRC = 0xe,
67 };
68
69 enum WindowSequence {
70 ONLY_LONG_SEQUENCE,
71 LONG_START_SEQUENCE,
72 EIGHT_SHORT_SEQUENCE,
73 LONG_STOP_SEQUENCE,
74 };
75
76 enum BandType {
77 ZERO_BT = 0, ///< Scalefactors and spectral data are all zero.
78 FIRST_PAIR_BT = 5, ///< This and later band types encode two values (rather than four) with one code word.
79 ESC_BT = 11, ///< Spectral data are coded with an escape sequence.
80 NOISE_BT = 13, ///< Spectral data are scaled white noise not coded in the bitstream.
81 INTENSITY_BT2 = 14, ///< Scalefactor data are intensity stereo positions.
82 INTENSITY_BT = 15, ///< Scalefactor data are intensity stereo positions.
83 };
84
85 #define IS_CODEBOOK_UNSIGNED(x) ((x - 1) & 10)
86
87 enum ChannelPosition {
88 AAC_CHANNEL_OFF = 0,
89 AAC_CHANNEL_FRONT = 1,
90 AAC_CHANNEL_SIDE = 2,
91 AAC_CHANNEL_BACK = 3,
92 AAC_CHANNEL_LFE = 4,
93 AAC_CHANNEL_CC = 5,
94 };
95
96 /**
97 * The point during decoding at which channel coupling is applied.
98 */
99 enum CouplingPoint {
100 BEFORE_TNS,
101 BETWEEN_TNS_AND_IMDCT,
102 AFTER_IMDCT = 3,
103 };
104
105 /**
106 * Output configuration status
107 */
108 enum OCStatus {
109 OC_NONE, ///< Output unconfigured
110 OC_TRIAL_PCE, ///< Output configuration under trial specified by an inband PCE
111 OC_TRIAL_FRAME, ///< Output configuration under trial specified by a frame header
112 OC_GLOBAL_HDR, ///< Output configuration set in a global header but not yet locked
113 OC_LOCKED, ///< Output configuration locked in place
114 };
115
116 typedef struct OutputConfiguration {
117 MPEG4AudioConfig m4ac;
118 uint8_t layout_map[MAX_ELEM_ID*4][3];
119 int layout_map_tags;
120 int channels;
121 uint64_t channel_layout;
122 enum OCStatus status;
123 } OutputConfiguration;
124
125 /**
126 * Predictor State
127 */
128 typedef struct PredictorState {
129 float cor0;
130 float cor1;
131 float var0;
132 float var1;
133 float r0;
134 float r1;
135 } PredictorState;
136
137 #define MAX_PREDICTORS 672
138
139 #define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times
140 #define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0
141 #define SCALE_MAX_POS 255 ///< scalefactor index maximum value
142 #define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard
143 #define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference
144
145 /**
146 * Long Term Prediction
147 */
148 typedef struct LongTermPrediction {
149 int8_t present;
150 int16_t lag;
151 float coef;
152 int8_t used[MAX_LTP_LONG_SFB];
153 } LongTermPrediction;
154
155 /**
156 * Individual Channel Stream
157 */
158 typedef struct IndividualChannelStream {
159 uint8_t max_sfb; ///< number of scalefactor bands per group
160 enum WindowSequence window_sequence[2];
161 uint8_t use_kb_window[2]; ///< If set, use Kaiser-Bessel window, otherwise use a sinus window.
162 int num_window_groups;
163 uint8_t group_len[8];
164 LongTermPrediction ltp;
165 const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window
166 const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window
167 int num_swb; ///< number of scalefactor window bands
168 int num_windows;
169 int tns_max_bands;
170 int predictor_present;
171 int predictor_initialized;
172 int predictor_reset_group;
173 uint8_t prediction_used[41];
174 } IndividualChannelStream;
175
176 /**
177 * Temporal Noise Shaping
178 */
179 typedef struct TemporalNoiseShaping {
180 int present;
181 int n_filt[8];
182 int length[8][4];
183 int direction[8][4];
184 int order[8][4];
185 float coef[8][4][TNS_MAX_ORDER];
186 } TemporalNoiseShaping;
187
188 /**
189 * Dynamic Range Control - decoded from the bitstream but not processed further.
190 */
191 typedef struct DynamicRangeControl {
192 int pce_instance_tag; ///< Indicates with which program the DRC info is associated.
193 int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative
194 int dyn_rng_ctl[17]; ///< DRC magnitude information
195 int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing.
196 int band_incr; ///< Number of DRC bands greater than 1 having DRC info.
197 int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain.
198 int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines.
199 int prog_ref_level; /**< A reference level for the long-term program audio level for all
200 * channels combined.
201 */
202 } DynamicRangeControl;
203
204 typedef struct Pulse {
205 int num_pulse;
206 int start;
207 int pos[4];
208 int amp[4];
209 } Pulse;
210
211 /**
212 * coupling parameters
213 */
214 typedef struct ChannelCoupling {
215 enum CouplingPoint coupling_point; ///< The point during decoding at which coupling is applied.
216 int num_coupled; ///< number of target elements
217 enum RawDataBlockType type[8]; ///< Type of channel element to be coupled - SCE or CPE.
218 int id_select[8]; ///< element id
219 int ch_select[8]; /**< [0] shared list of gains; [1] list of gains for right channel;
220 * [2] list of gains for left channel; [3] lists of gains for both channels
221 */
222 float gain[16][120];
223 } ChannelCoupling;
224
225 /**
226 * Single Channel Element - used for both SCE and LFE elements.
227 */
228 typedef struct SingleChannelElement {
229 IndividualChannelStream ics;
230 TemporalNoiseShaping tns;
231 Pulse pulse;
232 enum BandType band_type[128]; ///< band types
233 int band_type_run_end[120]; ///< band type run end points
234 float sf[120]; ///< scalefactors
235 int sf_idx[128]; ///< scalefactor indices (used by encoder)
236 uint8_t zeroes[128]; ///< band is not coded (used by encoder)
237 DECLARE_ALIGNED(32, float, coeffs)[1024]; ///< coefficients for IMDCT
238 DECLARE_ALIGNED(32, float, saved)[1024]; ///< overlap
239 DECLARE_ALIGNED(32, float, ret_buf)[2048]; ///< PCM output buffer
240 DECLARE_ALIGNED(16, float, ltp_state)[3072]; ///< time signal for LTP
241 PredictorState predictor_state[MAX_PREDICTORS];
242 float *ret; ///< PCM output
243 } SingleChannelElement;
244
245 /**
246 * channel element - generic struct for SCE/CPE/CCE/LFE
247 */
248 typedef struct ChannelElement {
249 // CPE specific
250 int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream.
251 int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder)
252 uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band
253 // shared
254 SingleChannelElement ch[2];
255 // CCE specific
256 ChannelCoupling coup;
257 SpectralBandReplication sbr;
258 } ChannelElement;
259
260 /**
261 * main AAC context
262 */
263 typedef struct AACContext {
264 AVCodecContext *avctx;
265 AVFrame frame;
266
267 int is_saved; ///< Set if elements have stored overlap from previous frame.
268 DynamicRangeControl che_drc;
269
270 /**
271 * @name Channel element related data
272 * @{
273 */
274 ChannelElement *che[4][MAX_ELEM_ID];
275 ChannelElement *tag_che_map[4][MAX_ELEM_ID];
276 int tags_mapped;
277 /** @} */
278
279 /**
280 * @name temporary aligned temporary buffers
281 * (We do not want to have these on the stack.)
282 * @{
283 */
284 DECLARE_ALIGNED(32, float, buf_mdct)[1024];
285 /** @} */
286
287 /**
288 * @name Computed / set up during initialization
289 * @{
290 */
291 FFTContext mdct;
292 FFTContext mdct_small;
293 FFTContext mdct_ltp;
294 DSPContext dsp;
295 FmtConvertContext fmt_conv;
296 AVFloatDSPContext fdsp;
297 int random_state;
298 /** @} */
299
300 /**
301 * @name Members used for output
302 * @{
303 */
304 SingleChannelElement *output_element[MAX_CHANNELS]; ///< Points to each SingleChannelElement
305 /** @} */
306
307 DECLARE_ALIGNED(32, float, temp)[128];
308
309 OutputConfiguration oc[2];
310 } AACContext;
311
312 #endif /* AVCODEC_AAC_H */