?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
avformat/mxfdec: Check edit unit for overflow in mxf_set_current_edit_unit()
[FFMpeg-mirror.git] / libavfilter / af_sofalizer.c
blob3071c85bd2e426b856b56e97f2b422119e5f09d5
1 /*****************************************************************************
2 * sofalizer.c : SOFAlizer filter for virtual binaural acoustics
3 *****************************************************************************
4 * Copyright (C) 2013-2015 Andreas Fuchs, Wolfgang Hrauda,
5 * Acoustics Research Institute (ARI), Vienna, Austria
6 *
7 * Authors: Andreas Fuchs <andi.fuchs.mail@gmail.com>
8 * Wolfgang Hrauda <wolfgang.hrauda@gmx.at>
9 *
10 * SOFAlizer project coordinator at ARI, main developer of SOFA:
11 * Piotr Majdak <piotr@majdak.at>
12 *
13 * This program is free software; you can redistribute it and/or modify it
14 * under the terms of the GNU Lesser General Public License as published by
15 * the Free Software Foundation; either version 2.1 of the License, or
16 * (at your option) any later version.
17 *
18 * This program is distributed in the hope that it will be useful,
19 * but WITHOUT ANY WARRANTY; without even the implied warranty of
20 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 * GNU Lesser General Public License for more details.
22 *
23 * You should have received a copy of the GNU Lesser General Public License
24 * along with this program; if not, write to the Free Software Foundation,
25 * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
26 *****************************************************************************/
27
28 #include <math.h>
29 #include <mysofa.h>
30
31 #include "libavutil/mem.h"
32 #include "libavutil/tx.h"
33 #include "libavutil/avstring.h"
34 #include "libavutil/channel_layout.h"
35 #include "libavutil/float_dsp.h"
36 #include "libavutil/intmath.h"
37 #include "libavutil/opt.h"
38 #include "avfilter.h"
39 #include "filters.h"
40 #include "formats.h"
41 #include "audio.h"
42
43 #define TIME_DOMAIN 0
44 #define FREQUENCY_DOMAIN 1
45
46 typedef struct MySofa { /* contains data of one SOFA file */
47 struct MYSOFA_HRTF *hrtf;
48 struct MYSOFA_LOOKUP *lookup;
49 struct MYSOFA_NEIGHBORHOOD *neighborhood;
50 int ir_samples; /* length of one impulse response (IR) */
51 int n_samples; /* ir_samples to next power of 2 */
52 float *lir, *rir; /* IRs (time-domain) */
53 float *fir;
54 int max_delay;
55 } MySofa;
56
57 typedef struct VirtualSpeaker {
58 uint8_t set;
59 float azim;
60 float elev;
61 } VirtualSpeaker;
62
63 typedef struct SOFAlizerContext {
64 const AVClass *class;
65
66 char *filename; /* name of SOFA file */
67 MySofa sofa; /* contains data of the SOFA file */
68
69 int sample_rate; /* sample rate from SOFA file */
70 float *speaker_azim; /* azimuth of the virtual loudspeakers */
71 float *speaker_elev; /* elevation of the virtual loudspeakers */
72 char *speakers_pos; /* custom positions of the virtual loudspeakers */
73 float lfe_gain; /* initial gain for the LFE channel */
74 float gain_lfe; /* gain applied to LFE channel */
75 int lfe_channel; /* LFE channel position in channel layout */
76
77 int n_conv; /* number of channels to convolute */
78
79 /* buffer variables (for convolution) */
80 float *ringbuffer[2]; /* buffers input samples, length of one buffer: */
81 /* no. input ch. (incl. LFE) x buffer_length */
82 int write[2]; /* current write position to ringbuffer */
83 int buffer_length; /* is: longest IR plus max. delay in all SOFA files */
84 /* then choose next power of 2 */
85 int n_fft; /* number of samples in one FFT block */
86 int nb_samples;
87
88 /* netCDF variables */
89 int *delay[2]; /* broadband delay for each channel/IR to be convolved */
90
91 float *data_ir[2]; /* IRs for all channels to be convolved */
92 /* (this excludes the LFE) */
93 float *temp_src[2];
94 AVComplexFloat *in_fft[2]; /* Array to hold input FFT values */
95 AVComplexFloat *out_fft[2]; /* Array to hold output FFT values */
96 AVComplexFloat *temp_afft[2]; /* Array to accumulate FFT values prior to IFFT */
97
98 /* control variables */
99 float gain; /* filter gain (in dB) */
100 float rotation; /* rotation of virtual loudspeakers (in degrees) */
101 float elevation; /* elevation of virtual loudspeakers (in deg.) */
102 float radius; /* distance virtual loudspeakers to listener (in metres) */
103 int type; /* processing type */
104 int framesize; /* size of buffer */
105 int normalize; /* should all IRs be normalized upon import ? */
106 int interpolate; /* should wanted IRs be interpolated from neighbors ? */
107 int minphase; /* should all IRs be minphased upon import ? */
108 float anglestep; /* neighbor search angle step, in agles */
109 float radstep; /* neighbor search radius step, in meters */
110
111 VirtualSpeaker vspkrpos[64];
112
113 AVTXContext *fft[2], *ifft[2];
114 av_tx_fn tx_fn[2], itx_fn[2];
115 AVComplexFloat *data_hrtf[2];
116
117 AVFloatDSPContext *fdsp;
118 } SOFAlizerContext;
119
120 static int close_sofa(struct MySofa *sofa)
121 {
122 if (sofa->neighborhood)
123 mysofa_neighborhood_free(sofa->neighborhood);
124 sofa->neighborhood = NULL;
125 if (sofa->lookup)
126 mysofa_lookup_free(sofa->lookup);
127 sofa->lookup = NULL;
128 if (sofa->hrtf)
129 mysofa_free(sofa->hrtf);
130 sofa->hrtf = NULL;
131 av_freep(&sofa->fir);
132
133 return 0;
134 }
135
136 static int preload_sofa(AVFilterContext *ctx, char *filename, int *samplingrate)
137 {
138 struct SOFAlizerContext *s = ctx->priv;
139 struct MYSOFA_HRTF *mysofa;
140 char *license;
141 int ret;
142
143 mysofa = mysofa_load(filename, &ret);
144 s->sofa.hrtf = mysofa;
145 if (ret || !mysofa) {
146 av_log(ctx, AV_LOG_ERROR, "Can't find SOFA-file '%s'\n", filename);
147 return AVERROR(EINVAL);
148 }
149
150 ret = mysofa_check(mysofa);
151 if (ret != MYSOFA_OK) {
152 av_log(ctx, AV_LOG_ERROR, "Selected SOFA file is invalid. Please select valid SOFA file.\n");
153 return ret;
154 }
155
156 if (s->normalize)
157 mysofa_loudness(s->sofa.hrtf);
158
159 if (s->minphase)
160 mysofa_minphase(s->sofa.hrtf, 0.01f);
161
162 mysofa_tocartesian(s->sofa.hrtf);
163
164 s->sofa.lookup = mysofa_lookup_init(s->sofa.hrtf);
165 if (s->sofa.lookup == NULL)
166 return AVERROR(EINVAL);
167
168 if (s->interpolate)
169 s->sofa.neighborhood = mysofa_neighborhood_init_withstepdefine(s->sofa.hrtf,
170 s->sofa.lookup,
171 s->anglestep,
172 s->radstep);
173
174 s->sofa.fir = av_calloc(s->sofa.hrtf->N * s->sofa.hrtf->R, sizeof(*s->sofa.fir));
175 if (!s->sofa.fir)
176 return AVERROR(ENOMEM);
177
178 if (mysofa->DataSamplingRate.elements != 1)
179 return AVERROR(EINVAL);
180 av_log(ctx, AV_LOG_DEBUG, "Original IR length: %d.\n", mysofa->N);
181 *samplingrate = mysofa->DataSamplingRate.values[0];
182 license = mysofa_getAttribute(mysofa->attributes, (char *)"License");
183 if (license)
184 av_log(ctx, AV_LOG_INFO, "SOFA license: %s\n", license);
185
186 return 0;
187 }
188
189 static int parse_channel_name(AVFilterContext *ctx, char **arg, int *rchannel)
190 {
191 int len;
192 enum AVChannel channel_id = 0;
193 char buf[8] = {0};
194
195 /* try to parse a channel name, e.g. "FL" */
196 if (av_sscanf(*arg, "%7[A-Z]%n", buf, &len)) {
197 channel_id = av_channel_from_string(buf);
198 if (channel_id < 0 || channel_id >= 64) {
199 av_log(ctx, AV_LOG_WARNING, "Failed to parse \'%s\' as channel name.\n", buf);
200 return AVERROR(EINVAL);
201 }
202
203 *rchannel = channel_id;
204 *arg += len;
205 return 0;
206 } else if (av_sscanf(*arg, "%d%n", &channel_id, &len) == 1) {
207 if (channel_id < 0 || channel_id >= 64) {
208 av_log(ctx, AV_LOG_WARNING, "Failed to parse \'%d\' as channel number.\n", channel_id);
209 return AVERROR(EINVAL);
210 }
211 *rchannel = channel_id;
212 *arg += len;
213 return 0;
214 }
215 return AVERROR(EINVAL);
216 }
217
218 static void parse_speaker_pos(AVFilterContext *ctx)
219 {
220 SOFAlizerContext *s = ctx->priv;
221 char *arg, *tokenizer, *p, *args = av_strdup(s->speakers_pos);
222
223 if (!args)
224 return;
225 p = args;
226
227 while ((arg = av_strtok(p, "|", &tokenizer))) {
228 float azim, elev;
229 int out_ch_id;
230
231 p = NULL;
232 if (parse_channel_name(ctx, &arg, &out_ch_id)) {
233 continue;
234 }
235 if (av_sscanf(arg, "%f %f", &azim, &elev) == 2) {
236 s->vspkrpos[out_ch_id].set = 1;
237 s->vspkrpos[out_ch_id].azim = azim;
238 s->vspkrpos[out_ch_id].elev = elev;
239 } else if (av_sscanf(arg, "%f", &azim) == 1) {
240 s->vspkrpos[out_ch_id].set = 1;
241 s->vspkrpos[out_ch_id].azim = azim;
242 s->vspkrpos[out_ch_id].elev = 0;
243 }
244 }
245
246 av_free(args);
247 }
248
249 static int get_speaker_pos(AVFilterContext *ctx,
250 float *speaker_azim, float *speaker_elev)
251 {
252 struct SOFAlizerContext *s = ctx->priv;
253 AVChannelLayout *channel_layout = &ctx->inputs[0]->ch_layout;
254 float azim[64] = { 0 };
255 float elev[64] = { 0 };
256 int ch, n_conv = ctx->inputs[0]->ch_layout.nb_channels; /* get no. input channels */
257
258 if (n_conv < 0 || n_conv > 64)
259 return AVERROR(EINVAL);
260
261 s->lfe_channel = -1;
262
263 if (s->speakers_pos)
264 parse_speaker_pos(ctx);
265
266 /* set speaker positions according to input channel configuration: */
267 for (ch = 0; ch < n_conv; ch++) {
268 int chan = av_channel_layout_channel_from_index(channel_layout, ch);
269
270 switch (chan) {
271 case AV_CHAN_FRONT_LEFT: azim[ch] = 30; break;
272 case AV_CHAN_FRONT_RIGHT: azim[ch] = 330; break;
273 case AV_CHAN_FRONT_CENTER: azim[ch] = 0; break;
274 case AV_CHAN_LOW_FREQUENCY:
275 case AV_CHAN_LOW_FREQUENCY_2: s->lfe_channel = ch; break;
276 case AV_CHAN_BACK_LEFT: azim[ch] = 150; break;
277 case AV_CHAN_BACK_RIGHT: azim[ch] = 210; break;
278 case AV_CHAN_BACK_CENTER: azim[ch] = 180; break;
279 case AV_CHAN_SIDE_LEFT: azim[ch] = 90; break;
280 case AV_CHAN_SIDE_RIGHT: azim[ch] = 270; break;
281 case AV_CHAN_FRONT_LEFT_OF_CENTER: azim[ch] = 15; break;
282 case AV_CHAN_FRONT_RIGHT_OF_CENTER: azim[ch] = 345; break;
283 case AV_CHAN_TOP_CENTER: azim[ch] = 0;
284 elev[ch] = 90; break;
285 case AV_CHAN_TOP_FRONT_LEFT: azim[ch] = 30;
286 elev[ch] = 45; break;
287 case AV_CHAN_TOP_FRONT_CENTER: azim[ch] = 0;
288 elev[ch] = 45; break;
289 case AV_CHAN_TOP_FRONT_RIGHT: azim[ch] = 330;
290 elev[ch] = 45; break;
291 case AV_CHAN_TOP_BACK_LEFT: azim[ch] = 150;
292 elev[ch] = 45; break;
293 case AV_CHAN_TOP_BACK_RIGHT: azim[ch] = 210;
294 elev[ch] = 45; break;
295 case AV_CHAN_TOP_BACK_CENTER: azim[ch] = 180;
296 elev[ch] = 45; break;
297 case AV_CHAN_WIDE_LEFT: azim[ch] = 90; break;
298 case AV_CHAN_WIDE_RIGHT: azim[ch] = 270; break;
299 case AV_CHAN_SURROUND_DIRECT_LEFT: azim[ch] = 90; break;
300 case AV_CHAN_SURROUND_DIRECT_RIGHT: azim[ch] = 270; break;
301 case AV_CHAN_STEREO_LEFT: azim[ch] = 90; break;
302 case AV_CHAN_STEREO_RIGHT: azim[ch] = 270; break;
303 default:
304 return AVERROR(EINVAL);
305 }
306
307 if (s->vspkrpos[ch].set) {
308 azim[ch] = s->vspkrpos[ch].azim;
309 elev[ch] = s->vspkrpos[ch].elev;
310 }
311 }
312
313 memcpy(speaker_azim, azim, n_conv * sizeof(float));
314 memcpy(speaker_elev, elev, n_conv * sizeof(float));
315
316 return 0;
317
318 }
319
320 typedef struct ThreadData {
321 AVFrame *in, *out;
322 int *write;
323 int **delay;
324 float **ir;
325 int *n_clippings;
326 float **ringbuffer;
327 float **temp_src;
328 AVComplexFloat **in_fft;
329 AVComplexFloat **out_fft;
330 AVComplexFloat **temp_afft;
331 } ThreadData;
332
333 static int sofalizer_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
334 {
335 SOFAlizerContext *s = ctx->priv;
336 ThreadData *td = arg;
337 AVFrame *in = td->in, *out = td->out;
338 int offset = jobnr;
339 int *write = &td->write[jobnr];
340 const int *const delay = td->delay[jobnr];
341 const float *const ir = td->ir[jobnr];
342 int *n_clippings = &td->n_clippings[jobnr];
343 float *ringbuffer = td->ringbuffer[jobnr];
344 float *temp_src = td->temp_src[jobnr];
345 const int ir_samples = s->sofa.ir_samples; /* length of one IR */
346 const int n_samples = s->sofa.n_samples;
347 const int planar = in->format == AV_SAMPLE_FMT_FLTP;
348 const int mult = 1 + !planar;
349 const float *src = (const float *)in->extended_data[0]; /* get pointer to audio input buffer */
350 float *dst = (float *)out->extended_data[jobnr * planar]; /* get pointer to audio output buffer */
351 const int in_channels = s->n_conv; /* number of input channels */
352 /* ring buffer length is: longest IR plus max. delay -> next power of 2 */
353 const int buffer_length = s->buffer_length;
354 /* -1 for AND instead of MODULO (applied to powers of 2): */
355 const uint32_t modulo = (uint32_t)buffer_length - 1;
356 float *buffer[64]; /* holds ringbuffer for each input channel */
357 int wr = *write;
358 int read;
359 int i, l;
360
361 if (!planar)
362 dst += offset;
363
364 for (l = 0; l < in_channels; l++) {
365 /* get starting address of ringbuffer for each input channel */
366 buffer[l] = ringbuffer + l * buffer_length;
367 }
368
369 for (i = 0; i < in->nb_samples; i++) {
370 const float *temp_ir = ir; /* using same set of IRs for each sample */
371
372 dst[0] = 0;
373 if (planar) {
374 for (l = 0; l < in_channels; l++) {
375 const float *srcp = (const float *)in->extended_data[l];
376
377 /* write current input sample to ringbuffer (for each channel) */
378 buffer[l][wr] = srcp[i];
379 }
380 } else {
381 for (l = 0; l < in_channels; l++) {
382 /* write current input sample to ringbuffer (for each channel) */
383 buffer[l][wr] = src[l];
384 }
385 }
386
387 /* loop goes through all channels to be convolved */
388 for (l = 0; l < in_channels; l++) {
389 const float *const bptr = buffer[l];
390
391 if (l == s->lfe_channel) {
392 /* LFE is an input channel but requires no convolution */
393 /* apply gain to LFE signal and add to output buffer */
394 dst[0] += *(buffer[s->lfe_channel] + wr) * s->gain_lfe;
395 temp_ir += n_samples;
396 continue;
397 }
398
399 /* current read position in ringbuffer: input sample write position
400 * - delay for l-th ch. + diff. betw. IR length and buffer length
401 * (mod buffer length) */
402 read = (wr - delay[l] - (ir_samples - 1) + buffer_length) & modulo;
403
404 if (read + ir_samples < buffer_length) {
405 memmove(temp_src, bptr + read, ir_samples * sizeof(*temp_src));
406 } else {
407 int len = FFMIN(n_samples - (read % ir_samples), buffer_length - read);
408
409 memmove(temp_src, bptr + read, len * sizeof(*temp_src));
410 memmove(temp_src + len, bptr, (n_samples - len) * sizeof(*temp_src));
411 }
412
413 /* multiply signal and IR, and add up the results */
414 dst[0] += s->fdsp->scalarproduct_float(temp_ir, temp_src, FFALIGN(ir_samples, 32));
415 temp_ir += n_samples;
416 }
417
418 /* clippings counter */
419 if (fabsf(dst[0]) > 1)
420 n_clippings[0]++;
421
422 /* move output buffer pointer by +2 to get to next sample of processed channel: */
423 dst += mult;
424 src += in_channels;
425 wr = (wr + 1) & modulo; /* update ringbuffer write position */
426 }
427
428 *write = wr; /* remember write position in ringbuffer for next call */
429
430 return 0;
431 }
432
433 static int sofalizer_fast_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
434 {
435 SOFAlizerContext *s = ctx->priv;
436 ThreadData *td = arg;
437 AVFrame *in = td->in, *out = td->out;
438 int offset = jobnr;
439 int *write = &td->write[jobnr];
440 AVComplexFloat *hrtf = s->data_hrtf[jobnr]; /* get pointers to current HRTF data */
441 int *n_clippings = &td->n_clippings[jobnr];
442 float *ringbuffer = td->ringbuffer[jobnr];
443 const int ir_samples = s->sofa.ir_samples; /* length of one IR */
444 const int planar = in->format == AV_SAMPLE_FMT_FLTP;
445 const int mult = 1 + !planar;
446 float *dst = (float *)out->extended_data[jobnr * planar]; /* get pointer to audio output buffer */
447 const int in_channels = s->n_conv; /* number of input channels */
448 /* ring buffer length is: longest IR plus max. delay -> next power of 2 */
449 const int buffer_length = s->buffer_length;
450 /* -1 for AND instead of MODULO (applied to powers of 2): */
451 const uint32_t modulo = (uint32_t)buffer_length - 1;
452 AVComplexFloat *fft_in = s->in_fft[jobnr]; /* temporary array for FFT input data */
453 AVComplexFloat *fft_out = s->out_fft[jobnr]; /* temporary array for FFT output data */
454 AVComplexFloat *fft_acc = s->temp_afft[jobnr];
455 AVTXContext *ifft = s->ifft[jobnr];
456 av_tx_fn itx_fn = s->itx_fn[jobnr];
457 AVTXContext *fft = s->fft[jobnr];
458 av_tx_fn tx_fn = s->tx_fn[jobnr];
459 const int n_conv = s->n_conv;
460 const int n_fft = s->n_fft;
461 const float fft_scale = 1.0f / s->n_fft;
462 AVComplexFloat *hrtf_offset;
463 int wr = *write;
464 int n_read;
465 int i, j;
466
467 if (!planar)
468 dst += offset;
469
470 /* find minimum between number of samples and output buffer length:
471 * (important, if one IR is longer than the output buffer) */
472 n_read = FFMIN(ir_samples, in->nb_samples);
473 for (j = 0; j < n_read; j++) {
474 /* initialize output buf with saved signal from overflow buf */
475 dst[mult * j] = ringbuffer[wr];
476 ringbuffer[wr] = 0.0f; /* re-set read samples to zero */
477 /* update ringbuffer read/write position */
478 wr = (wr + 1) & modulo;
479 }
480
481 /* initialize rest of output buffer with 0 */
482 for (j = n_read; j < in->nb_samples; j++) {
483 dst[mult * j] = 0;
484 }
485
486 /* fill FFT accumulation with 0 */
487 memset(fft_acc, 0, sizeof(AVComplexFloat) * n_fft);
488
489 for (i = 0; i < n_conv; i++) {
490 const float *src = (const float *)in->extended_data[i * planar]; /* get pointer to audio input buffer */
491
492 if (i == s->lfe_channel) { /* LFE */
493 if (in->format == AV_SAMPLE_FMT_FLT) {
494 for (j = 0; j < in->nb_samples; j++) {
495 /* apply gain to LFE signal and add to output buffer */
496 dst[2 * j] += src[i + j * in_channels] * s->gain_lfe;
497 }
498 } else {
499 for (j = 0; j < in->nb_samples; j++) {
500 /* apply gain to LFE signal and add to output buffer */
501 dst[j] += src[j] * s->gain_lfe;
502 }
503 }
504 continue;
505 }
506
507 /* outer loop: go through all input channels to be convolved */
508 offset = i * n_fft; /* no. samples already processed */
509 hrtf_offset = hrtf + offset;
510
511 /* fill FFT input with 0 (we want to zero-pad) */
512 memset(fft_in, 0, sizeof(AVComplexFloat) * n_fft);
513
514 if (in->format == AV_SAMPLE_FMT_FLT) {
515 for (j = 0; j < in->nb_samples; j++) {
516 /* prepare input for FFT */
517 /* write all samples of current input channel to FFT input array */
518 fft_in[j].re = src[j * in_channels + i];
519 }
520 } else {
521 for (j = 0; j < in->nb_samples; j++) {
522 /* prepare input for FFT */
523 /* write all samples of current input channel to FFT input array */
524 fft_in[j].re = src[j];
525 }
526 }
527
528 /* transform input signal of current channel to frequency domain */
529 tx_fn(fft, fft_out, fft_in, sizeof(*fft_in));
530
531 for (j = 0; j < n_fft; j++) {
532 const AVComplexFloat *hcomplex = hrtf_offset + j;
533 const float re = fft_out[j].re;
534 const float im = fft_out[j].im;
535
536 /* complex multiplication of input signal and HRTFs */
537 /* output channel (real): */
538 fft_acc[j].re += re * hcomplex->re - im * hcomplex->im;
539 /* output channel (imag): */
540 fft_acc[j].im += re * hcomplex->im + im * hcomplex->re;
541 }
542 }
543
544 /* transform output signal of current channel back to time domain */
545 itx_fn(ifft, fft_out, fft_acc, sizeof(*fft_acc));
546
547 for (j = 0; j < in->nb_samples; j++) {
548 /* write output signal of current channel to output buffer */
549 dst[mult * j] += fft_out[j].re * fft_scale;
550 }
551
552 for (j = 0; j < ir_samples - 1; j++) { /* overflow length is IR length - 1 */
553 /* write the rest of output signal to overflow buffer */
554 int write_pos = (wr + j) & modulo;
555
556 *(ringbuffer + write_pos) += fft_out[in->nb_samples + j].re * fft_scale;
557 }
558
559 /* go through all samples of current output buffer: count clippings */
560 for (i = 0; i < out->nb_samples; i++) {
561 /* clippings counter */
562 if (fabsf(dst[i * mult]) > 1) { /* if current output sample > 1 */
563 n_clippings[0]++;
564 }
565 }
566
567 /* remember read/write position in ringbuffer for next call */
568 *write = wr;
569
570 return 0;
571 }
572
573 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
574 {
575 AVFilterContext *ctx = inlink->dst;
576 SOFAlizerContext *s = ctx->priv;
577 AVFilterLink *outlink = ctx->outputs[0];
578 int n_clippings[2] = { 0 };
579 ThreadData td;
580 AVFrame *out;
581
582 out = ff_get_audio_buffer(outlink, in->nb_samples);
583 if (!out) {
584 av_frame_free(&in);
585 return AVERROR(ENOMEM);
586 }
587 av_frame_copy_props(out, in);
588
589 td.in = in; td.out = out; td.write = s->write;
590 td.delay = s->delay; td.ir = s->data_ir; td.n_clippings = n_clippings;
591 td.ringbuffer = s->ringbuffer; td.temp_src = s->temp_src;
592 td.in_fft = s->in_fft;
593 td.out_fft = s->out_fft;
594 td.temp_afft = s->temp_afft;
595
596 if (s->type == TIME_DOMAIN) {
597 ff_filter_execute(ctx, sofalizer_convolute, &td, NULL, 2);
598 } else if (s->type == FREQUENCY_DOMAIN) {
599 ff_filter_execute(ctx, sofalizer_fast_convolute, &td, NULL, 2);
600 }
601
602 /* display error message if clipping occurred */
603 if (n_clippings[0] + n_clippings[1] > 0) {
604 av_log(ctx, AV_LOG_WARNING, "%d of %d samples clipped. Please reduce gain.\n",
605 n_clippings[0] + n_clippings[1], out->nb_samples * 2);
606 }
607
608 av_frame_free(&in);
609 return ff_filter_frame(outlink, out);
610 }
611
612 static int activate(AVFilterContext *ctx)
613 {
614 AVFilterLink *inlink = ctx->inputs[0];
615 AVFilterLink *outlink = ctx->outputs[0];
616 SOFAlizerContext *s = ctx->priv;
617 AVFrame *in;
618 int ret;
619
620 FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
621
622 if (s->nb_samples)
623 ret = ff_inlink_consume_samples(inlink, s->nb_samples, s->nb_samples, &in);
624 else
625 ret = ff_inlink_consume_frame(inlink, &in);
626 if (ret < 0)
627 return ret;
628 if (ret > 0)
629 return filter_frame(inlink, in);
630
631 FF_FILTER_FORWARD_STATUS(inlink, outlink);
632 FF_FILTER_FORWARD_WANTED(outlink, inlink);
633
634 return FFERROR_NOT_READY;
635 }
636
637 static int query_formats(const AVFilterContext *ctx,
638 AVFilterFormatsConfig **cfg_in,
639 AVFilterFormatsConfig **cfg_out)
640 {
641 const SOFAlizerContext *s = ctx->priv;
642 AVFilterChannelLayouts *layouts = NULL;
643 int ret, sample_rates[] = { 48000, -1 };
644 static const enum AVSampleFormat sample_fmts[] = {
645 AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
646 AV_SAMPLE_FMT_NONE
647 };
648
649 ret = ff_set_common_formats_from_list2(ctx, cfg_in, cfg_out, sample_fmts);
650 if (ret)
651 return ret;
652
653 layouts = ff_all_channel_layouts();
654 if (!layouts)
655 return AVERROR(ENOMEM);
656
657 ret = ff_channel_layouts_ref(layouts, &cfg_in[0]->channel_layouts);
658 if (ret)
659 return ret;
660
661 layouts = NULL;
662 ret = ff_add_channel_layout(&layouts, &(AVChannelLayout)AV_CHANNEL_LAYOUT_STEREO);
663 if (ret)
664 return ret;
665
666 ret = ff_channel_layouts_ref(layouts, &cfg_out[0]->channel_layouts);
667 if (ret)
668 return ret;
669
670 sample_rates[0] = s->sample_rate;
671 return ff_set_common_samplerates_from_list2(ctx, cfg_in, cfg_out, sample_rates);
672 }
673
674 static int getfilter_float(AVFilterContext *ctx, float x, float y, float z,
675 float *left, float *right,
676 float *delay_left, float *delay_right)
677 {
678 struct SOFAlizerContext *s = ctx->priv;
679 float c[3], delays[2];
680 float *fl, *fr;
681 int nearest;
682 int *neighbors;
683 float *res;
684
685 c[0] = x, c[1] = y, c[2] = z;
686 nearest = mysofa_lookup(s->sofa.lookup, c);
687 if (nearest < 0)
688 return AVERROR(EINVAL);
689
690 if (s->interpolate) {
691 neighbors = mysofa_neighborhood(s->sofa.neighborhood, nearest);
692 res = mysofa_interpolate(s->sofa.hrtf, c,
693 nearest, neighbors,
694 s->sofa.fir, delays);
695 } else {
696 if (s->sofa.hrtf->DataDelay.elements > s->sofa.hrtf->R) {
697 delays[0] = s->sofa.hrtf->DataDelay.values[nearest * s->sofa.hrtf->R];
698 delays[1] = s->sofa.hrtf->DataDelay.values[nearest * s->sofa.hrtf->R + 1];
699 } else {
700 delays[0] = s->sofa.hrtf->DataDelay.values[0];
701 delays[1] = s->sofa.hrtf->DataDelay.values[1];
702 }
703 res = s->sofa.hrtf->DataIR.values + nearest * s->sofa.hrtf->N * s->sofa.hrtf->R;
704 }
705
706 *delay_left = delays[0];
707 *delay_right = delays[1];
708
709 fl = res;
710 fr = res + s->sofa.hrtf->N;
711
712 memcpy(left, fl, sizeof(float) * s->sofa.hrtf->N);
713 memcpy(right, fr, sizeof(float) * s->sofa.hrtf->N);
714
715 return 0;
716 }
717
718 static int load_data(AVFilterContext *ctx, int azim, int elev, float radius, int sample_rate)
719 {
720 struct SOFAlizerContext *s = ctx->priv;
721 int n_samples;
722 int ir_samples;
723 int n_conv = s->n_conv; /* no. channels to convolve */
724 int n_fft;
725 float delay_l; /* broadband delay for each IR */
726 float delay_r;
727 int nb_input_channels = ctx->inputs[0]->ch_layout.nb_channels; /* no. input channels */
728 float gain_lin = expf((s->gain - 3 * nb_input_channels) / 20 * M_LN10); /* gain - 3dB/channel */
729 AVComplexFloat *data_hrtf_l = NULL;
730 AVComplexFloat *data_hrtf_r = NULL;
731 AVComplexFloat *fft_out_l = NULL;
732 AVComplexFloat *fft_out_r = NULL;
733 AVComplexFloat *fft_in_l = NULL;
734 AVComplexFloat *fft_in_r = NULL;
735 float *data_ir_l = NULL;
736 float *data_ir_r = NULL;
737 int offset = 0; /* used for faster pointer arithmetics in for-loop */
738 int i, j, azim_orig = azim, elev_orig = elev;
739 int ret = 0;
740 int n_current;
741 int n_max = 0;
742
743 av_log(ctx, AV_LOG_DEBUG, "IR length: %d.\n", s->sofa.hrtf->N);
744 s->sofa.ir_samples = s->sofa.hrtf->N;
745 s->sofa.n_samples = 1 << (32 - ff_clz(s->sofa.ir_samples));
746
747 n_samples = s->sofa.n_samples;
748 ir_samples = s->sofa.ir_samples;
749
750 if (s->type == TIME_DOMAIN) {
751 s->data_ir[0] = av_calloc(n_samples, sizeof(float) * s->n_conv);
752 s->data_ir[1] = av_calloc(n_samples, sizeof(float) * s->n_conv);
753
754 if (!s->data_ir[0] || !s->data_ir[1]) {
755 ret = AVERROR(ENOMEM);
756 goto fail;
757 }
758 }
759
760 s->delay[0] = av_calloc(s->n_conv, sizeof(int));
761 s->delay[1] = av_calloc(s->n_conv, sizeof(int));
762
763 if (!s->delay[0] || !s->delay[1]) {
764 ret = AVERROR(ENOMEM);
765 goto fail;
766 }
767
768 /* get temporary IR for L and R channel */
769 data_ir_l = av_calloc(n_conv * n_samples, sizeof(*data_ir_l));
770 data_ir_r = av_calloc(n_conv * n_samples, sizeof(*data_ir_r));
771 if (!data_ir_r || !data_ir_l) {
772 ret = AVERROR(ENOMEM);
773 goto fail;
774 }
775
776 if (s->type == TIME_DOMAIN) {
777 s->temp_src[0] = av_calloc(n_samples, sizeof(float));
778 s->temp_src[1] = av_calloc(n_samples, sizeof(float));
779 if (!s->temp_src[0] || !s->temp_src[1]) {
780 ret = AVERROR(ENOMEM);
781 goto fail;
782 }
783 }
784
785 s->speaker_azim = av_calloc(s->n_conv, sizeof(*s->speaker_azim));
786 s->speaker_elev = av_calloc(s->n_conv, sizeof(*s->speaker_elev));
787 if (!s->speaker_azim || !s->speaker_elev) {
788 ret = AVERROR(ENOMEM);
789 goto fail;
790 }
791
792 /* get speaker positions */
793 if ((ret = get_speaker_pos(ctx, s->speaker_azim, s->speaker_elev)) < 0) {
794 av_log(ctx, AV_LOG_ERROR, "Couldn't get speaker positions. Input channel configuration not supported.\n");
795 goto fail;
796 }
797
798 for (i = 0; i < s->n_conv; i++) {
799 float coordinates[3];
800
801 /* load and store IRs and corresponding delays */
802 azim = (int)(s->speaker_azim[i] + azim_orig) % 360;
803 elev = (int)(s->speaker_elev[i] + elev_orig) % 90;
804
805 coordinates[0] = azim;
806 coordinates[1] = elev;
807 coordinates[2] = radius;
808
809 mysofa_s2c(coordinates);
810
811 /* get id of IR closest to desired position */
812 ret = getfilter_float(ctx, coordinates[0], coordinates[1], coordinates[2],
813 data_ir_l + n_samples * i,
814 data_ir_r + n_samples * i,
815 &delay_l, &delay_r);
816 if (ret < 0)
817 goto fail;
818
819 s->delay[0][i] = delay_l * sample_rate;
820 s->delay[1][i] = delay_r * sample_rate;
821
822 s->sofa.max_delay = FFMAX3(s->sofa.max_delay, s->delay[0][i], s->delay[1][i]);
823 }
824
825 /* get size of ringbuffer (longest IR plus max. delay) */
826 /* then choose next power of 2 for performance optimization */
827 n_current = n_samples + s->sofa.max_delay;
828 /* length of longest IR plus max. delay */
829 n_max = FFMAX(n_max, n_current);
830
831 /* buffer length is longest IR plus max. delay -> next power of 2
832 (32 - count leading zeros gives required exponent) */
833 s->buffer_length = 1 << (32 - ff_clz(n_max));
834 s->n_fft = n_fft = 1 << (32 - ff_clz(n_max + s->framesize));
835
836 if (s->type == FREQUENCY_DOMAIN) {
837 float scale = 1.f;
838
839 av_tx_uninit(&s->fft[0]);
840 av_tx_uninit(&s->fft[1]);
841 ret = av_tx_init(&s->fft[0], &s->tx_fn[0], AV_TX_FLOAT_FFT, 0, s->n_fft, &scale, 0);
842 if (ret < 0)
843 goto fail;
844 ret = av_tx_init(&s->fft[1], &s->tx_fn[1], AV_TX_FLOAT_FFT, 0, s->n_fft, &scale, 0);
845 if (ret < 0)
846 goto fail;
847 av_tx_uninit(&s->ifft[0]);
848 av_tx_uninit(&s->ifft[1]);
849 ret = av_tx_init(&s->ifft[0], &s->itx_fn[0], AV_TX_FLOAT_FFT, 1, s->n_fft, &scale, 0);
850 if (ret < 0)
851 goto fail;
852 ret = av_tx_init(&s->ifft[1], &s->itx_fn[1], AV_TX_FLOAT_FFT, 1, s->n_fft, &scale, 0);
853 if (ret < 0)
854 goto fail;
855 }
856
857 if (s->type == TIME_DOMAIN) {
858 s->ringbuffer[0] = av_calloc(s->buffer_length, sizeof(float) * nb_input_channels);
859 s->ringbuffer[1] = av_calloc(s->buffer_length, sizeof(float) * nb_input_channels);
860 } else if (s->type == FREQUENCY_DOMAIN) {
861 /* get temporary HRTF memory for L and R channel */
862 data_hrtf_l = av_malloc_array(n_fft, sizeof(*data_hrtf_l) * n_conv);
863 data_hrtf_r = av_malloc_array(n_fft, sizeof(*data_hrtf_r) * n_conv);
864 if (!data_hrtf_r || !data_hrtf_l) {
865 ret = AVERROR(ENOMEM);
866 goto fail;
867 }
868
869 s->ringbuffer[0] = av_calloc(s->buffer_length, sizeof(float));
870 s->ringbuffer[1] = av_calloc(s->buffer_length, sizeof(float));
871 s->in_fft[0] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
872 s->in_fft[1] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
873 s->out_fft[0] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
874 s->out_fft[1] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
875 s->temp_afft[0] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
876 s->temp_afft[1] = av_malloc_array(s->n_fft, sizeof(AVComplexFloat));
877 if (!s->in_fft[0] || !s->in_fft[1] ||
878 !s->out_fft[0] || !s->out_fft[1] ||
879 !s->temp_afft[0] || !s->temp_afft[1]) {
880 ret = AVERROR(ENOMEM);
881 goto fail;
882 }
883 }
884
885 if (!s->ringbuffer[0] || !s->ringbuffer[1]) {
886 ret = AVERROR(ENOMEM);
887 goto fail;
888 }
889
890 if (s->type == FREQUENCY_DOMAIN) {
891 fft_out_l = av_calloc(n_fft, sizeof(*fft_out_l));
892 fft_out_r = av_calloc(n_fft, sizeof(*fft_out_r));
893 fft_in_l = av_calloc(n_fft, sizeof(*fft_in_l));
894 fft_in_r = av_calloc(n_fft, sizeof(*fft_in_r));
895 if (!fft_in_l || !fft_in_r ||
896 !fft_out_l || !fft_out_r) {
897 ret = AVERROR(ENOMEM);
898 goto fail;
899 }
900 }
901
902 for (i = 0; i < s->n_conv; i++) {
903 float *lir, *rir;
904
905 offset = i * n_samples; /* no. samples already written */
906
907 lir = data_ir_l + offset;
908 rir = data_ir_r + offset;
909
910 if (s->type == TIME_DOMAIN) {
911 for (j = 0; j < ir_samples; j++) {
912 /* load reversed IRs of the specified source position
913 * sample-by-sample for left and right ear; and apply gain */
914 s->data_ir[0][offset + j] = lir[ir_samples - 1 - j] * gain_lin;
915 s->data_ir[1][offset + j] = rir[ir_samples - 1 - j] * gain_lin;
916 }
917 } else if (s->type == FREQUENCY_DOMAIN) {
918 memset(fft_in_l, 0, n_fft * sizeof(*fft_in_l));
919 memset(fft_in_r, 0, n_fft * sizeof(*fft_in_r));
920
921 offset = i * n_fft; /* no. samples already written */
922 for (j = 0; j < ir_samples; j++) {
923 /* load non-reversed IRs of the specified source position
924 * sample-by-sample and apply gain,
925 * L channel is loaded to real part, R channel to imag part,
926 * IRs are shifted by L and R delay */
927 fft_in_l[s->delay[0][i] + j].re = lir[j] * gain_lin;
928 fft_in_r[s->delay[1][i] + j].re = rir[j] * gain_lin;
929 }
930
931 /* actually transform to frequency domain (IRs -> HRTFs) */
932 s->tx_fn[0](s->fft[0], fft_out_l, fft_in_l, sizeof(*fft_in_l));
933 memcpy(data_hrtf_l + offset, fft_out_l, n_fft * sizeof(*fft_out_l));
934 s->tx_fn[1](s->fft[1], fft_out_r, fft_in_r, sizeof(*fft_in_r));
935 memcpy(data_hrtf_r + offset, fft_out_r, n_fft * sizeof(*fft_out_r));
936 }
937 }
938
939 if (s->type == FREQUENCY_DOMAIN) {
940 s->data_hrtf[0] = av_malloc_array(n_fft * s->n_conv, sizeof(AVComplexFloat));
941 s->data_hrtf[1] = av_malloc_array(n_fft * s->n_conv, sizeof(AVComplexFloat));
942 if (!s->data_hrtf[0] || !s->data_hrtf[1]) {
943 ret = AVERROR(ENOMEM);
944 goto fail;
945 }
946
947 memcpy(s->data_hrtf[0], data_hrtf_l, /* copy HRTF data to */
948 sizeof(AVComplexFloat) * n_conv * n_fft); /* filter struct */
949 memcpy(s->data_hrtf[1], data_hrtf_r,
950 sizeof(AVComplexFloat) * n_conv * n_fft);
951 }
952
953 fail:
954 av_freep(&data_hrtf_l); /* free temporary HRTF memory */
955 av_freep(&data_hrtf_r);
956
957 av_freep(&data_ir_l); /* free temprary IR memory */
958 av_freep(&data_ir_r);
959
960 av_freep(&fft_out_l); /* free temporary FFT memory */
961 av_freep(&fft_out_r);
962
963 av_freep(&fft_in_l); /* free temporary FFT memory */
964 av_freep(&fft_in_r);
965
966 return ret;
967 }
968
969 static av_cold int init(AVFilterContext *ctx)
970 {
971 SOFAlizerContext *s = ctx->priv;
972 int ret;
973
974 if (!s->filename) {
975 av_log(ctx, AV_LOG_ERROR, "Valid SOFA filename must be set.\n");
976 return AVERROR(EINVAL);
977 }
978
979 /* preload SOFA file, */
980 ret = preload_sofa(ctx, s->filename, &s->sample_rate);
981 if (ret) {
982 /* file loading error */
983 av_log(ctx, AV_LOG_ERROR, "Error while loading SOFA file: '%s'\n", s->filename);
984 } else { /* no file loading error, resampling not required */
985 av_log(ctx, AV_LOG_DEBUG, "File '%s' loaded.\n", s->filename);
986 }
987
988 if (ret) {
989 av_log(ctx, AV_LOG_ERROR, "No valid SOFA file could be loaded. Please specify valid SOFA file.\n");
990 return ret;
991 }
992
993 s->fdsp = avpriv_float_dsp_alloc(0);
994 if (!s->fdsp)
995 return AVERROR(ENOMEM);
996
997 return 0;
998 }
999
1000 static int config_input(AVFilterLink *inlink)
1001 {
1002 AVFilterContext *ctx = inlink->dst;
1003 SOFAlizerContext *s = ctx->priv;
1004 int ret;
1005
1006 if (s->type == FREQUENCY_DOMAIN)
1007 s->nb_samples = s->framesize;
1008
1009 /* gain -3 dB per channel */
1010 s->gain_lfe = expf((s->gain - 3 * inlink->ch_layout.nb_channels + s->lfe_gain) / 20 * M_LN10);
1011
1012 s->n_conv = inlink->ch_layout.nb_channels;
1013
1014 /* load IRs to data_ir[0] and data_ir[1] for required directions */
1015 if ((ret = load_data(ctx, s->rotation, s->elevation, s->radius, inlink->sample_rate)) < 0)
1016 return ret;
1017
1018 av_log(ctx, AV_LOG_DEBUG, "Samplerate: %d Channels to convolute: %d, Length of ringbuffer: %d x %d\n",
1019 inlink->sample_rate, s->n_conv, inlink->ch_layout.nb_channels, s->buffer_length);
1020
1021 return 0;
1022 }
1023
1024 static av_cold void uninit(AVFilterContext *ctx)
1025 {
1026 SOFAlizerContext *s = ctx->priv;
1027
1028 close_sofa(&s->sofa);
1029 av_tx_uninit(&s->ifft[0]);
1030 av_tx_uninit(&s->ifft[1]);
1031 av_tx_uninit(&s->fft[0]);
1032 av_tx_uninit(&s->fft[1]);
1033 s->ifft[0] = NULL;
1034 s->ifft[1] = NULL;
1035 s->fft[0] = NULL;
1036 s->fft[1] = NULL;
1037 av_freep(&s->delay[0]);
1038 av_freep(&s->delay[1]);
1039 av_freep(&s->data_ir[0]);
1040 av_freep(&s->data_ir[1]);
1041 av_freep(&s->ringbuffer[0]);
1042 av_freep(&s->ringbuffer[1]);
1043 av_freep(&s->speaker_azim);
1044 av_freep(&s->speaker_elev);
1045 av_freep(&s->temp_src[0]);
1046 av_freep(&s->temp_src[1]);
1047 av_freep(&s->temp_afft[0]);
1048 av_freep(&s->temp_afft[1]);
1049 av_freep(&s->in_fft[0]);
1050 av_freep(&s->in_fft[1]);
1051 av_freep(&s->out_fft[0]);
1052 av_freep(&s->out_fft[1]);
1053 av_freep(&s->data_hrtf[0]);
1054 av_freep(&s->data_hrtf[1]);
1055 av_freep(&s->fdsp);
1056 }
1057
1058 #define OFFSET(x) offsetof(SOFAlizerContext, x)
1059 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
1060
1061 static const AVOption sofalizer_options[] = {
1062 { "sofa", "sofa filename", OFFSET(filename), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
1063 { "gain", "set gain in dB", OFFSET(gain), AV_OPT_TYPE_FLOAT, {.dbl=0}, -20, 40, .flags = FLAGS },
1064 { "rotation", "set rotation" , OFFSET(rotation), AV_OPT_TYPE_FLOAT, {.dbl=0}, -360, 360, .flags = FLAGS },
1065 { "elevation", "set elevation", OFFSET(elevation), AV_OPT_TYPE_FLOAT, {.dbl=0}, -90, 90, .flags = FLAGS },
1066 { "radius", "set radius", OFFSET(radius), AV_OPT_TYPE_FLOAT, {.dbl=1}, 0, 5, .flags = FLAGS },
1067 { "type", "set processing", OFFSET(type), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, .flags = FLAGS, .unit = "type" },
1068 { "time", "time domain", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, .flags = FLAGS, .unit = "type" },
1069 { "freq", "frequency domain", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, .flags = FLAGS, .unit = "type" },
1070 { "speakers", "set speaker custom positions", OFFSET(speakers_pos), AV_OPT_TYPE_STRING, {.str=0}, 0, 0, .flags = FLAGS },
1071 { "lfegain", "set lfe gain", OFFSET(lfe_gain), AV_OPT_TYPE_FLOAT, {.dbl=0}, -20,40, .flags = FLAGS },
1072 { "framesize", "set frame size", OFFSET(framesize), AV_OPT_TYPE_INT, {.i64=1024},1024,96000, .flags = FLAGS },
1073 { "normalize", "normalize IRs", OFFSET(normalize), AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, .flags = FLAGS },
1074 { "interpolate","interpolate IRs from neighbors", OFFSET(interpolate),AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, .flags = FLAGS },
1075 { "minphase", "minphase IRs", OFFSET(minphase), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, .flags = FLAGS },
1076 { "anglestep", "set neighbor search angle step", OFFSET(anglestep), AV_OPT_TYPE_FLOAT, {.dbl=.5}, 0.01, 10, .flags = FLAGS },
1077 { "radstep", "set neighbor search radius step", OFFSET(radstep), AV_OPT_TYPE_FLOAT, {.dbl=.01}, 0.01, 1, .flags = FLAGS },
1078 { NULL }
1079 };
1080
1081 AVFILTER_DEFINE_CLASS(sofalizer);
1082
1083 static const AVFilterPad inputs[] = {
1084 {
1085 .name = "default",
1086 .type = AVMEDIA_TYPE_AUDIO,
1087 .config_props = config_input,
1088 },
1089 };
1090
1091 const FFFilter ff_af_sofalizer = {
1092 .p.name = "sofalizer",
1093 .p.description = NULL_IF_CONFIG_SMALL("SOFAlizer (Spatially Oriented Format for Acoustics)."),
1094 .p.priv_class = &sofalizer_class,
1095 .p.flags = AVFILTER_FLAG_SLICE_THREADS,
1096 .priv_size = sizeof(SOFAlizerContext),
1097 .init = init,
1098 .activate = activate,
1099 .uninit = uninit,
1100 FILTER_INPUTS(inputs),
1101 FILTER_OUTPUTS(ff_audio_default_filterpad),
1102 FILTER_QUERY_FUNC2(query_formats),
1103 };
Mirror of the official FFMpeg Git repo