Originally committed as revision 14694 to svn://svn.ffmpeg.org/ffmpeg/trunk
Robert Swain authored on 2008/08/11 20:16:06... | ... |
@@ -99,6 +99,40 @@ static VLC vlc_scalefactors; |
99 | 99 |
static VLC vlc_spectral[11]; |
100 | 100 |
|
101 | 101 |
|
102 |
+/** |
|
103 |
+ * Decode an array of 4 bit element IDs, optionally interleaved with a stereo/mono switching bit. |
|
104 |
+ * |
|
105 |
+ * @param cpe_map Stereo (Channel Pair Element) map, NULL if stereo bit is not present. |
|
106 |
+ * @param sce_map mono (Single Channel Element) map |
|
107 |
+ * @param type speaker type/position for these channels |
|
108 |
+ */ |
|
109 |
+static void decode_channel_map(enum ChannelPosition *cpe_map, |
|
110 |
+ enum ChannelPosition *sce_map, enum ChannelPosition type, GetBitContext * gb, int n) { |
|
111 |
+ while(n--) { |
|
112 |
+ enum ChannelPosition *map = cpe_map && get_bits1(gb) ? cpe_map : sce_map; // stereo or mono map |
|
113 |
+ map[get_bits(gb, 4)] = type; |
|
114 |
+ } |
|
115 |
+} |
|
116 |
+ |
|
117 |
+/** |
|
118 |
+ * Decode program configuration element; reference: table 4.2. |
|
119 |
+ * |
|
120 |
+ * @param new_che_pos New channel position configuration - we only do something if it differs from the current one. |
|
121 |
+ * |
|
122 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
123 |
+ */ |
|
124 |
+static int decode_pce(AACContext * ac, enum ChannelPosition new_che_pos[4][MAX_ELEM_ID], |
|
125 |
+ GetBitContext * gb) { |
|
126 |
+ int num_front, num_side, num_back, num_lfe, num_assoc_data, num_cc; |
|
127 |
+ |
|
128 |
+ skip_bits(gb, 2); // object_type |
|
129 |
+ |
|
130 |
+ ac->m4ac.sampling_index = get_bits(gb, 4); |
|
131 |
+ if(ac->m4ac.sampling_index > 11) { |
|
132 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "invalid sampling rate index %d\n", ac->m4ac.sampling_index); |
|
133 |
+ return -1; |
|
134 |
+ } |
|
135 |
+ ac->m4ac.sample_rate = ff_mpeg4audio_sample_rates[ac->m4ac.sampling_index]; |
|
102 | 136 |
num_front = get_bits(gb, 4); |
103 | 137 |
num_side = get_bits(gb, 4); |
104 | 138 |
num_back = get_bits(gb, 4); |
... | ... |
@@ -130,6 +164,131 @@ static VLC vlc_spectral[11]; |
130 | 130 |
return 0; |
131 | 131 |
} |
132 | 132 |
|
133 |
+/** |
|
134 |
+ * Set up channel positions based on a default channel configuration |
|
135 |
+ * as specified in table 1.17. |
|
136 |
+ * |
|
137 |
+ * @param new_che_pos New channel position configuration - we only do something if it differs from the current one. |
|
138 |
+ * |
|
139 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
140 |
+ */ |
|
141 |
+static int set_default_channel_config(AACContext *ac, enum ChannelPosition new_che_pos[4][MAX_ELEM_ID], |
|
142 |
+ int channel_config) |
|
143 |
+{ |
|
144 |
+ if(channel_config < 1 || channel_config > 7) { |
|
145 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "invalid default channel configuration (%d)\n", |
|
146 |
+ channel_config); |
|
147 |
+ return -1; |
|
148 |
+ } |
|
149 |
+ |
|
150 |
+ /* default channel configurations: |
|
151 |
+ * |
|
152 |
+ * 1ch : front center (mono) |
|
153 |
+ * 2ch : L + R (stereo) |
|
154 |
+ * 3ch : front center + L + R |
|
155 |
+ * 4ch : front center + L + R + back center |
|
156 |
+ * 5ch : front center + L + R + back stereo |
|
157 |
+ * 6ch : front center + L + R + back stereo + LFE |
|
158 |
+ * 7ch : front center + L + R + outer front left + outer front right + back stereo + LFE |
|
159 |
+ */ |
|
160 |
+ |
|
161 |
+ if(channel_config != 2) |
|
162 |
+ new_che_pos[TYPE_SCE][0] = AAC_CHANNEL_FRONT; // front center (or mono) |
|
163 |
+ if(channel_config > 1) |
|
164 |
+ new_che_pos[TYPE_CPE][0] = AAC_CHANNEL_FRONT; // L + R (or stereo) |
|
165 |
+ if(channel_config == 4) |
|
166 |
+ new_che_pos[TYPE_SCE][1] = AAC_CHANNEL_BACK; // back center |
|
167 |
+ if(channel_config > 4) |
|
168 |
+ new_che_pos[TYPE_CPE][(channel_config == 7) + 1] |
|
169 |
+ = AAC_CHANNEL_BACK; // back stereo |
|
170 |
+ if(channel_config > 5) |
|
171 |
+ new_che_pos[TYPE_LFE][0] = AAC_CHANNEL_LFE; // LFE |
|
172 |
+ if(channel_config == 7) |
|
173 |
+ new_che_pos[TYPE_CPE][1] = AAC_CHANNEL_FRONT; // outer front left + outer front right |
|
174 |
+ |
|
175 |
+ return 0; |
|
176 |
+} |
|
177 |
+ |
|
178 |
+ return -1; |
|
179 |
+ } |
|
180 |
+ |
|
181 |
+ if (get_bits1(gb)) // dependsOnCoreCoder |
|
182 |
+ skip_bits(gb, 14); // coreCoderDelay |
|
183 |
+ extension_flag = get_bits1(gb); |
|
184 |
+ |
|
185 |
+ if(ac->m4ac.object_type == AOT_AAC_SCALABLE || |
|
186 |
+ ac->m4ac.object_type == AOT_ER_AAC_SCALABLE) |
|
187 |
+ skip_bits(gb, 3); // layerNr |
|
188 |
+ |
|
189 |
+ memset(new_che_pos, 0, 4 * MAX_ELEM_ID * sizeof(new_che_pos[0][0])); |
|
190 |
+ if (channel_config == 0) { |
|
191 |
+ skip_bits(gb, 4); // element_instance_tag |
|
192 |
+ if((ret = decode_pce(ac, new_che_pos, gb))) |
|
193 |
+ return ret; |
|
194 |
+ } else { |
|
195 |
+ if((ret = set_default_channel_config(ac, new_che_pos, channel_config))) |
|
196 |
+ return ret; |
|
197 |
+ } |
|
198 |
+ if((ret = output_configure(ac, ac->che_pos, new_che_pos))) |
|
199 |
+ return ret; |
|
200 |
+ |
|
201 |
+ if (extension_flag) { |
|
202 |
+ switch (ac->m4ac.object_type) { |
|
203 |
+ case AOT_ER_BSAC: |
|
204 |
+ skip_bits(gb, 5); // numOfSubFrame |
|
205 |
+ skip_bits(gb, 11); // layer_length |
|
206 |
+ break; |
|
207 |
+ case AOT_ER_AAC_LC: |
|
208 |
+ case AOT_ER_AAC_LTP: |
|
209 |
+ case AOT_ER_AAC_SCALABLE: |
|
210 |
+ case AOT_ER_AAC_LD: |
|
211 |
+ skip_bits(gb, 3); /* aacSectionDataResilienceFlag |
|
212 |
+ * aacScalefactorDataResilienceFlag |
|
213 |
+ * aacSpectralDataResilienceFlag |
|
214 |
+ */ |
|
215 |
+ break; |
|
216 |
+ } |
|
217 |
+ skip_bits1(gb); // extensionFlag3 (TBD in version 3) |
|
218 |
+ } |
|
219 |
+ return 0; |
|
220 |
+} |
|
221 |
+ |
|
222 |
+/** |
|
223 |
+ * Decode audio specific configuration; reference: table 1.13. |
|
224 |
+ * |
|
225 |
+ * @param data pointer to AVCodecContext extradata |
|
226 |
+ * @param data_size size of AVCCodecContext extradata |
|
227 |
+ * |
|
228 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
229 |
+ */ |
|
230 |
+static int decode_audio_specific_config(AACContext * ac, void *data, int data_size) { |
|
231 |
+ GetBitContext gb; |
|
232 |
+ int i; |
|
233 |
+ |
|
234 |
+ init_get_bits(&gb, data, data_size * 8); |
|
235 |
+ |
|
236 |
+ if((i = ff_mpeg4audio_get_config(&ac->m4ac, data, data_size)) < 0) |
|
237 |
+ return -1; |
|
238 |
+ if(ac->m4ac.sampling_index > 11) { |
|
239 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "invalid sampling rate index %d\n", ac->m4ac.sampling_index); |
|
240 |
+ return -1; |
|
241 |
+ } |
|
242 |
+ |
|
243 |
+ skip_bits_long(&gb, i); |
|
244 |
+ |
|
245 |
+ switch (ac->m4ac.object_type) { |
|
246 |
+ case AOT_AAC_LC: |
|
247 |
+ if (decode_ga_specific_config(ac, &gb, ac->m4ac.chan_config)) |
|
248 |
+ return -1; |
|
249 |
+ break; |
|
250 |
+ default: |
|
251 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "Audio object type %s%d is not supported.\n", |
|
252 |
+ ac->m4ac.sbr == 1? "SBR+" : "", ac->m4ac.object_type); |
|
253 |
+ return -1; |
|
254 |
+ } |
|
255 |
+ return 0; |
|
256 |
+} |
|
257 |
+ |
|
133 | 258 |
static av_cold int aac_decode_init(AVCodecContext * avccontext) { |
134 | 259 |
AACContext * ac = avccontext->priv_data; |
135 | 260 |
int i; |
... | ... |
@@ -140,6 +299,7 @@ static av_cold int aac_decode_init(AVCodecContext * avccontext) { |
140 | 140 |
decode_audio_specific_config(ac, avccontext->extradata, avccontext->extradata_size)) |
141 | 141 |
return -1; |
142 | 142 |
|
143 |
+ avccontext->sample_fmt = SAMPLE_FMT_S16; |
|
143 | 144 |
avccontext->sample_rate = ac->m4ac.sample_rate; |
144 | 145 |
avccontext->frame_size = 1024; |
145 | 146 |
|
... | ... |
@@ -157,6 +317,8 @@ static av_cold int aac_decode_init(AVCodecContext * avccontext) { |
157 | 157 |
|
158 | 158 |
dsputil_init(&ac->dsp, avccontext); |
159 | 159 |
|
160 |
+ ac->random_state = 0x1f2e3d4c; |
|
161 |
+ |
|
160 | 162 |
// -1024 - Compensate wrong IMDCT method. |
161 | 163 |
// 32768 - Required to scale values to the correct range for the bias method |
162 | 164 |
// for float to int16 conversion. |
... | ... |
@@ -188,6 +350,10 @@ static av_cold int aac_decode_init(AVCodecContext * avccontext) { |
188 | 188 |
return 0; |
189 | 189 |
} |
190 | 190 |
|
191 |
+/** |
|
192 |
+ * Skip data_stream_element; reference: table 4.10. |
|
193 |
+ */ |
|
194 |
+static void skip_data_stream_element(GetBitContext * gb) { |
|
191 | 195 |
int byte_align = get_bits1(gb); |
192 | 196 |
int count = get_bits(gb, 8); |
193 | 197 |
if (count == 255) |
... | ... |
@@ -198,6 +364,27 @@ static av_cold int aac_decode_init(AVCodecContext * avccontext) { |
198 | 198 |
} |
199 | 199 |
|
200 | 200 |
/** |
201 |
+ * Decode Individual Channel Stream info; reference: table 4.6. |
|
202 |
+ * |
|
203 |
+ * @param common_window Channels have independent [0], or shared [1], Individual Channel Stream information. |
|
204 |
+ */ |
|
205 |
+static int decode_ics_info(AACContext * ac, IndividualChannelStream * ics, GetBitContext * gb, int common_window) { |
|
206 |
+ if (get_bits1(gb)) { |
|
207 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "Reserved bit set.\n"); |
|
208 |
+ memset(ics, 0, sizeof(IndividualChannelStream)); |
|
209 |
+ return -1; |
|
210 |
+ } |
|
211 |
+ ics->window_sequence[1] = ics->window_sequence[0]; |
|
212 |
+ ics->window_sequence[0] = get_bits(gb, 2); |
|
213 |
+ ics->use_kb_window[1] = ics->use_kb_window[0]; |
|
214 |
+ ics->use_kb_window[0] = get_bits1(gb); |
|
215 |
+ ics->num_window_groups = 1; |
|
216 |
+ ics->group_len[0] = 1; |
|
217 |
+ |
|
218 |
+ return 0; |
|
219 |
+} |
|
220 |
+ |
|
221 |
+/** |
|
201 | 222 |
* inverse quantization |
202 | 223 |
* |
203 | 224 |
* @param a quantized value to be dequantized |
... | ... |
@@ -210,6 +397,15 @@ static inline float ivquant(int a) { |
210 | 210 |
return cbrtf(fabsf(a)) * a; |
211 | 211 |
} |
212 | 212 |
|
213 |
+/** |
|
214 |
+ * Decode band types (section_data payload); reference: table 4.46. |
|
215 |
+ * |
|
216 |
+ * @param band_type array of the used band type |
|
217 |
+ * @param band_type_run_end array of the last scalefactor band of a band type run |
|
218 |
+ * |
|
219 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
220 |
+ */ |
|
221 |
+static int decode_band_types(AACContext * ac, enum BandType band_type[120], |
|
213 | 222 |
int band_type_run_end[120], GetBitContext * gb, IndividualChannelStream * ics) { |
214 | 223 |
int g, idx = 0; |
215 | 224 |
const int bits = (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) ? 3 : 5; |
... | ... |
@@ -232,7 +428,13 @@ static inline float ivquant(int a) { |
232 | 232 |
sect_len, ics->max_sfb); |
233 | 233 |
return -1; |
234 | 234 |
} |
235 |
+ } |
|
236 |
+ } |
|
237 |
+ return 0; |
|
238 |
+} |
|
235 | 239 |
|
240 |
+/** |
|
241 |
+ * Decode scalefactors; reference: table 4.47. |
|
236 | 242 |
* |
237 | 243 |
* @param mix_gain channel gain (Not used by AAC bitstream.) |
238 | 244 |
* @param global_gain first scalefactor value as scalefactors are differentially coded |
... | ... |
@@ -314,6 +516,16 @@ static void decode_pulses(Pulse * pulse, GetBitContext * gb) { |
314 | 314 |
} |
315 | 315 |
|
316 | 316 |
/** |
317 |
+ * Decode Mid/Side data; reference: table 4.54. |
|
318 |
+ * |
|
319 |
+ * @param ms_present Indicates mid/side stereo presence. [0] mask is all 0s; |
|
320 |
+ * [1] mask is decoded from bitstream; [2] mask is all 1s; |
|
321 |
+ * [3] reserved for scalable AAC |
|
322 |
+ */ |
|
323 |
+static void decode_mid_side_stereo(ChannelElement * cpe, GetBitContext * gb, |
|
324 |
+ int ms_present) { |
|
325 |
+ |
|
326 |
+/** |
|
317 | 327 |
* Add pulses with particular amplitudes to the quantized spectral data; reference: 4.6.3.3. |
318 | 328 |
* |
319 | 329 |
* @param pulse pointer to pulse data struct |
... | ... |
@@ -330,10 +542,109 @@ static void add_pulses(int icoef[1024], const Pulse * pulse, const IndividualCha |
330 | 330 |
} |
331 | 331 |
|
332 | 332 |
/** |
333 |
- * Parse Spectral Band Replication extension data; reference: table 4.55. |
|
333 |
+ * Decode an individual_channel_stream payload; reference: table 4.44. |
|
334 |
+ * |
|
335 |
+ * @param common_window Channels have independent [0], or shared [1], Individual Channel Stream information. |
|
336 |
+ * @param scale_flag scalable [1] or non-scalable [0] AAC (Unused until scalable AAC is implemented.) |
|
337 |
+ * |
|
338 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
339 |
+ */ |
|
340 |
+static int decode_ics(AACContext * ac, SingleChannelElement * sce, GetBitContext * gb, int common_window, int scale_flag) { |
|
341 |
+ int icoeffs[1024]; |
|
342 |
+ Pulse pulse; |
|
343 |
+ TemporalNoiseShaping * tns = &sce->tns; |
|
344 |
+ IndividualChannelStream * ics = &sce->ics; |
|
345 |
+ float * out = sce->coeffs; |
|
346 |
+ int global_gain, pulse_present = 0; |
|
347 |
+ |
|
348 |
+ /* These two assignments are to silence some GCC warnings about the |
|
349 |
+ * variables being used uninitialised when in fact they always are. |
|
350 |
+ */ |
|
351 |
+ pulse.num_pulse = 0; |
|
352 |
+ pulse.start = 0; |
|
353 |
+ |
|
354 |
+ global_gain = get_bits(gb, 8); |
|
355 |
+ |
|
356 |
+ if (!common_window && !scale_flag) { |
|
357 |
+ if (decode_ics_info(ac, ics, gb, 0) < 0) |
|
358 |
+ return -1; |
|
359 |
+ } |
|
360 |
+ |
|
361 |
+ if (decode_band_types(ac, sce->band_type, sce->band_type_run_end, gb, ics) < 0) |
|
362 |
+ return -1; |
|
363 |
+ if (decode_scalefactors(ac, sce->sf, gb, global_gain, ics, sce->band_type, sce->band_type_run_end) < 0) |
|
364 |
+ return -1; |
|
365 |
+ |
|
366 |
+ pulse_present = 0; |
|
367 |
+ if (!scale_flag) { |
|
368 |
+ if ((pulse_present = get_bits1(gb))) { |
|
369 |
+ if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) { |
|
370 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "Pulse tool not allowed in eight short sequence.\n"); |
|
371 |
+ return -1; |
|
372 |
+ } |
|
373 |
+ decode_pulses(&pulse, gb); |
|
374 |
+ } |
|
375 |
+ if ((tns->present = get_bits1(gb)) && decode_tns(ac, tns, gb, ics)) |
|
376 |
+ return -1; |
|
377 |
+ if (get_bits1(gb)) { |
|
378 |
+ av_log_missing_feature(ac->avccontext, "SSR", 1); |
|
379 |
+ return -1; |
|
380 |
+ } |
|
381 |
+ } |
|
382 |
+ |
|
383 |
+ if (decode_spectrum(ac, icoeffs, gb, ics, sce->band_type) < 0) |
|
384 |
+ return -1; |
|
385 |
+ if (pulse_present) |
|
386 |
+ add_pulses(icoeffs, &pulse, ics); |
|
387 |
+ dequant(ac, out, icoeffs, sce->sf, ics, sce->band_type); |
|
388 |
+ return 0; |
|
389 |
+} |
|
390 |
+ |
|
391 |
+/** |
|
392 |
+ * Decode a channel_pair_element; reference: table 4.4. |
|
393 |
+ * |
|
394 |
+ * @param elem_id Identifies the instance of a syntax element. |
|
395 |
+ * |
|
396 |
+ * @return Returns error status. 0 - OK, !0 - error |
|
397 |
+ */ |
|
398 |
+static int decode_cpe(AACContext * ac, GetBitContext * gb, int elem_id) { |
|
399 |
+ int i, ret, common_window, ms_present = 0; |
|
400 |
+ ChannelElement * cpe; |
|
401 |
+ |
|
402 |
+ cpe = ac->che[TYPE_CPE][elem_id]; |
|
403 |
+ common_window = get_bits1(gb); |
|
404 |
+ if (common_window) { |
|
405 |
+ if (decode_ics_info(ac, &cpe->ch[0].ics, gb, 1)) |
|
406 |
+ return -1; |
|
407 |
+ i = cpe->ch[1].ics.use_kb_window[0]; |
|
408 |
+ cpe->ch[1].ics = cpe->ch[0].ics; |
|
409 |
+ cpe->ch[1].ics.use_kb_window[1] = i; |
|
410 |
+ ms_present = get_bits(gb, 2); |
|
411 |
+ if(ms_present == 3) { |
|
412 |
+ av_log(ac->avccontext, AV_LOG_ERROR, "ms_present = 3 is reserved.\n"); |
|
413 |
+ return -1; |
|
414 |
+ } else if(ms_present) |
|
415 |
+ decode_mid_side_stereo(cpe, gb, ms_present); |
|
416 |
+ } |
|
417 |
+ if ((ret = decode_ics(ac, &cpe->ch[0], gb, common_window, 0))) |
|
418 |
+ return ret; |
|
419 |
+ if ((ret = decode_ics(ac, &cpe->ch[1], gb, common_window, 0))) |
|
420 |
+ return ret; |
|
421 |
+ |
|
422 |
+ if (common_window && ms_present) |
|
423 |
+ apply_mid_side_stereo(cpe); |
|
424 |
+ |
|
425 |
+ if (cpe->ch[1].ics.intensity_present) |
|
426 |
+ apply_intensity_stereo(cpe, ms_present); |
|
427 |
+ return 0; |
|
428 |
+} |
|
429 |
+ |
|
430 |
+/** |
|
431 |
+ * Decode Spectral Band Replication extension data; reference: table 4.55. |
|
334 | 432 |
* |
335 | 433 |
* @param crc flag indicating the presence of CRC checksum |
336 | 434 |
* @param cnt length of TYPE_FIL syntactic element in bytes |
435 |
+ * |
|
337 | 436 |
* @return Returns number of bytes consumed from the TYPE_FIL element. |
338 | 437 |
*/ |
339 | 438 |
static int decode_sbr_extension(AACContext * ac, GetBitContext * gb, int crc, int cnt) { |
... | ... |
@@ -343,6 +654,66 @@ static int decode_sbr_extension(AACContext * ac, GetBitContext * gb, int crc, in |
343 | 343 |
return cnt; |
344 | 344 |
} |
345 | 345 |
|
346 |
+/** |
|
347 |
+ * Decode dynamic range information; reference: table 4.52. |
|
348 |
+ * |
|
349 |
+ * @param cnt length of TYPE_FIL syntactic element in bytes |
|
350 |
+ * |
|
351 |
+ * @return Returns number of bytes consumed. |
|
352 |
+ */ |
|
353 |
+static int decode_dynamic_range(DynamicRangeControl *che_drc, GetBitContext * gb, int cnt) { |
|
354 |
+ int n = 1; |
|
355 |
+ int drc_num_bands = 1; |
|
356 |
+ int i; |
|
357 |
+ |
|
358 |
+ /* pce_tag_present? */ |
|
359 |
+ if(get_bits1(gb)) { |
|
360 |
+ che_drc->pce_instance_tag = get_bits(gb, 4); |
|
361 |
+ skip_bits(gb, 4); // tag_reserved_bits |
|
362 |
+ n++; |
|
363 |
+ } |
|
364 |
+ |
|
365 |
+ /* excluded_chns_present? */ |
|
366 |
+ if(get_bits1(gb)) { |
|
367 |
+ n += decode_drc_channel_exclusions(che_drc, gb); |
|
368 |
+ } |
|
369 |
+ |
|
370 |
+ /* drc_bands_present? */ |
|
371 |
+ if (get_bits1(gb)) { |
|
372 |
+ che_drc->band_incr = get_bits(gb, 4); |
|
373 |
+ che_drc->interpolation_scheme = get_bits(gb, 4); |
|
374 |
+ n++; |
|
375 |
+ drc_num_bands += che_drc->band_incr; |
|
376 |
+ for (i = 0; i < drc_num_bands; i++) { |
|
377 |
+ che_drc->band_top[i] = get_bits(gb, 8); |
|
378 |
+ n++; |
|
379 |
+ } |
|
380 |
+ } |
|
381 |
+ |
|
382 |
+ /* prog_ref_level_present? */ |
|
383 |
+ if (get_bits1(gb)) { |
|
384 |
+ che_drc->prog_ref_level = get_bits(gb, 7); |
|
385 |
+ skip_bits1(gb); // prog_ref_level_reserved_bits |
|
386 |
+ n++; |
|
387 |
+ } |
|
388 |
+ |
|
389 |
+ for (i = 0; i < drc_num_bands; i++) { |
|
390 |
+ che_drc->dyn_rng_sgn[i] = get_bits1(gb); |
|
391 |
+ che_drc->dyn_rng_ctl[i] = get_bits(gb, 7); |
|
392 |
+ n++; |
|
393 |
+ } |
|
394 |
+ |
|
395 |
+ return n; |
|
396 |
+} |
|
397 |
+ |
|
398 |
+/** |
|
399 |
+ * Decode extension data (incomplete); reference: table 4.51. |
|
400 |
+ * |
|
401 |
+ * @param cnt length of TYPE_FIL syntactic element in bytes |
|
402 |
+ * |
|
403 |
+ * @return Returns number of bytes consumed |
|
404 |
+ */ |
|
405 |
+static int decode_extension_payload(AACContext * ac, GetBitContext * gb, int cnt) { |
|
346 | 406 |
int crc_flag = 0; |
347 | 407 |
int res = cnt; |
348 | 408 |
switch (get_bits(gb, 4)) { // extension type |
... | ... |
@@ -365,6 +736,21 @@ static int decode_sbr_extension(AACContext * ac, GetBitContext * gb, int crc, in |
365 | 365 |
} |
366 | 366 |
|
367 | 367 |
/** |
368 |
+ * Conduct IMDCT and windowing. |
|
369 |
+ */ |
|
370 |
+static void imdct_and_windowing(AACContext * ac, SingleChannelElement * sce) { |
|
371 |
+ IndividualChannelStream * ics = &sce->ics; |
|
372 |
+ float * in = sce->coeffs; |
|
373 |
+ float * out = sce->ret; |
|
374 |
+ float * saved = sce->saved; |
|
375 |
+ const float * lwindow = ics->use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_aac_sine_long_1024; |
|
376 |
+ const float * swindow = ics->use_kb_window[0] ? ff_aac_kbd_short_128 : ff_aac_sine_short_128; |
|
377 |
+ const float * lwindow_prev = ics->use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_aac_sine_long_1024; |
|
378 |
+ const float * swindow_prev = ics->use_kb_window[1] ? ff_aac_kbd_short_128 : ff_aac_sine_short_128; |
|
379 |
+ float * buf = ac->buf_mdct; |
|
380 |
+ int i; |
|
381 |
+ |
|
382 |
+/** |
|
368 | 383 |
* Apply dependent channel coupling (applied before IMDCT). |
369 | 384 |
* |
370 | 385 |
* @param index index into coupling gain array |
... | ... |
@@ -409,6 +795,26 @@ static void apply_independent_coupling(AACContext * ac, SingleChannelElement * s |
409 | 409 |
sce->ret[i] += gain * (cc->ch[0].ret[i] - ac->add_bias); |
410 | 410 |
} |
411 | 411 |
|
412 |
+ if (!ac->is_saved) { |
|
413 |
+ ac->is_saved = 1; |
|
414 |
+ *data_size = 0; |
|
415 |
+ return 0; |
|
416 |
+ } |
|
417 |
+ |
|
418 |
+ data_size_tmp = 1024 * avccontext->channels * sizeof(int16_t); |
|
419 |
+ if(*data_size < data_size_tmp) { |
|
420 |
+ av_log(avccontext, AV_LOG_ERROR, |
|
421 |
+ "Output buffer too small (%d) or trying to output too many samples (%d) for this frame.\n", |
|
422 |
+ *data_size, data_size_tmp); |
|
423 |
+ return -1; |
|
424 |
+ } |
|
425 |
+ *data_size = data_size_tmp; |
|
426 |
+ |
|
427 |
+ ac->dsp.float_to_int16_interleave(data, (const float **)ac->output_data, 1024, avccontext->channels); |
|
428 |
+ |
|
429 |
+ return buf_size; |
|
430 |
+} |
|
431 |
+ |
|
412 | 432 |
static av_cold int aac_decode_close(AVCodecContext * avccontext) { |
413 | 433 |
AACContext * ac = avccontext->priv_data; |
414 | 434 |
int i, j; |
... | ... |
@@ -43,6 +43,7 @@ |
43 | 43 |
size); |
44 | 44 |
|
45 | 45 |
#define MAX_CHANNELS 64 |
46 |
+#define MAX_ELEM_ID 16 |
|
46 | 47 |
|
47 | 48 |
#define IVQUANT_SIZE 1024 |
48 | 49 |
|
... | ... |
@@ -76,6 +77,17 @@ enum AudioObjectType { |
76 | 76 |
AOT_SSC, ///< N SinuSoidal Coding |
77 | 77 |
}; |
78 | 78 |
|
79 |
+enum RawDataBlockType { |
|
80 |
+ TYPE_SCE, |
|
81 |
+ TYPE_CPE, |
|
82 |
+ TYPE_CCE, |
|
83 |
+ TYPE_LFE, |
|
84 |
+ TYPE_DSE, |
|
85 |
+ TYPE_PCE, |
|
86 |
+ TYPE_FIL, |
|
87 |
+ TYPE_END, |
|
88 |
+}; |
|
89 |
+ |
|
79 | 90 |
enum ExtensionPayloadID { |
80 | 91 |
EXT_FILL, |
81 | 92 |
EXT_FILL_DATA, |
... | ... |
@@ -111,6 +123,35 @@ enum ChannelPosition { |
111 | 111 |
AAC_CHANNEL_CC = 5, |
112 | 112 |
}; |
113 | 113 |
|
114 |
+/** |
|
115 |
+ * The point during decoding at which channel coupling is applied. |
|
116 |
+ */ |
|
117 |
+enum CouplingPoint { |
|
118 |
+ BEFORE_TNS, |
|
119 |
+ BETWEEN_TNS_AND_IMDCT, |
|
120 |
+ AFTER_IMDCT = 3, |
|
121 |
+}; |
|
122 |
+ |
|
123 |
+/** |
|
124 |
+ * Individual Channel Stream |
|
125 |
+ */ |
|
126 |
+ |
|
127 |
+/** |
|
128 |
+ * Dynamic Range Control - decoded from the bitstream but not processed further. |
|
129 |
+ */ |
|
130 |
+typedef struct { |
|
131 |
+ int pce_instance_tag; ///< Indicates with which program the DRC info is associated. |
|
132 |
+ int dyn_rng_sgn[17]; ///< DRC sign information; 0 - positive, 1 - negative |
|
133 |
+ int dyn_rng_ctl[17]; ///< DRC magnitude information |
|
134 |
+ int exclude_mask[MAX_CHANNELS]; ///< Channels to be excluded from DRC processing. |
|
135 |
+ int band_incr; ///< Number of DRC bands greater than 1 having DRC info. |
|
136 |
+ int interpolation_scheme; ///< Indicates the interpolation scheme used in the SBR QMF domain. |
|
137 |
+ int band_top[17]; ///< Indicates the top of the i-th DRC band in units of 4 spectral lines. |
|
138 |
+ int prog_ref_level; /**< A reference level for the long-term program audio level for all |
|
139 |
+ * channels combined. |
|
140 |
+ */ |
|
141 |
+} DynamicRangeControl; |
|
142 |
+ |
|
114 | 143 |
typedef struct { |
115 | 144 |
int num_pulse; |
116 | 145 |
int start; |
... | ... |
@@ -134,9 +175,15 @@ typedef struct { |
134 | 134 |
int is_saved; ///< Set if elements have stored overlap from previous frame. |
135 | 135 |
DynamicRangeControl che_drc; |
136 | 136 |
|
137 |
+ /** |
|
138 |
+ * @defgroup elements |
|
139 |
+ * @{ |
|
140 |
+ */ |
|
137 | 141 |
enum ChannelPosition che_pos[4][MAX_ELEM_ID]; /**< channel element channel mapping with the |
138 | 142 |
* first index as the first 4 raw data block types |
139 | 143 |
*/ |
144 |
+ ChannelElement * che[4][MAX_ELEM_ID]; |
|
145 |
+ /** @} */ |
|
140 | 146 |
|
141 | 147 |
/** |
142 | 148 |
* @defgroup tables Computed / set up during initialization. |
... | ... |
@@ -145,6 +192,7 @@ typedef struct { |
145 | 145 |
MDCTContext mdct; |
146 | 146 |
MDCTContext mdct_small; |
147 | 147 |
DSPContext dsp; |
148 |
+ int random_state; |
|
148 | 149 |
/** @} */ |
149 | 150 |
|
150 | 151 |
/** |
... | ... |
@@ -32,6 +32,14 @@ |
32 | 32 |
|
33 | 33 |
#include <stdint.h> |
34 | 34 |
|
35 |
+const uint8_t ff_aac_num_swb_1024[] = { |
|
36 |
+ 41, 41, 47, 49, 49, 51, 47, 47, 43, 43, 43, 40 |
|
37 |
+}; |
|
38 |
+ |
|
39 |
+const uint8_t ff_aac_num_swb_128[] = { |
|
40 |
+ 12, 12, 12, 14, 14, 14, 15, 15, 15, 15, 15, 15 |
|
41 |
+}; |
|
42 |
+ |
|
35 | 43 |
const uint32_t ff_aac_scalefactor_code[121] = { |
36 | 44 |
0x3ffe8, 0x3ffe6, 0x3ffe7, 0x3ffe5, 0x7fff5, 0x7fff1, 0x7ffed, 0x7fff6, |
37 | 45 |
0x7ffee, 0x7ffef, 0x7fff0, 0x7fffc, 0x7fffd, 0x7ffff, 0x7fffe, 0x7fff7, |
... | ... |
@@ -796,6 +804,13 @@ const float ff_aac_ivquant_tab[IVQUANT_SIZE] = { |
796 | 796 |
4064.0312908, 4074.6805676, 4085.3368071, 4096.0000000, |
797 | 797 |
}; |
798 | 798 |
|
799 |
+/** |
|
800 |
+ * Table of pow(2, (i - 200)/4.) used for different purposes depending on the |
|
801 |
+ * range of indices to the table: |
|
802 |
+ * [ 0, 255] scale factor decoding when using C dsp.float_to_int16 |
|
803 |
+ * [60, 315] scale factor decoding when using SIMD dsp.float_to_int16 |
|
804 |
+ * [45, 300] intensity stereo position decoding mapped in reverse order i.e. 0->300, 1->299, ..., 254->46, 255->45 |
|
805 |
+ */ |
|
799 | 806 |
const float ff_aac_pow2sf_tab[316] = { |
800 | 807 |
8.88178420e-16, 1.05622810e-15, 1.25607397e-15, 1.49373210e-15, |
801 | 808 |
1.77635684e-15, 2.11245619e-15, 2.51214793e-15, 2.98746420e-15, |
... | ... |
@@ -35,6 +35,18 @@ |
35 | 35 |
|
36 | 36 |
#include <stdint.h> |
37 | 37 |
|
38 |
+/* NOTE: |
|
39 |
+ * Tables in this file are used by the AAC decoder and will be used by the AAC |
|
40 |
+ * encoder. |
|
41 |
+ */ |
|
42 |
+ |
|
43 |
+/* @name number of scalefactor window bands for long and short transform windows respectively |
|
44 |
+ * @{ |
|
45 |
+ */ |
|
46 |
+extern const uint8_t ff_aac_num_swb_1024[]; |
|
47 |
+extern const uint8_t ff_aac_num_swb_128 []; |
|
48 |
+// @} |
|
49 |
+ |
|
38 | 50 |
extern const uint32_t ff_aac_scalefactor_code[121]; |
39 | 51 |
extern const uint8_t ff_aac_scalefactor_bits[121]; |
40 | 52 |
|