FFmpeg  4.0
af_sofalizer.c
Go to the documentation of this file.
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 "libavcodec/avfft.h"
32 #include "libavutil/avstring.h"
34 #include "libavutil/float_dsp.h"
35 #include "libavutil/intmath.h"
36 #include "libavutil/opt.h"
37 #include "avfilter.h"
38 #include "internal.h"
39 #include "audio.h"
40 
41 #define TIME_DOMAIN 0
42 #define FREQUENCY_DOMAIN 1
43 
44 typedef struct MySofa { /* contains data of one SOFA file */
45  struct MYSOFA_EASY *easy;
46  int n_samples; /* length of one impulse response (IR) */
47  float *lir, *rir; /* IRs (time-domain) */
48  int max_delay;
49 } MySofa;
50 
51 typedef struct VirtualSpeaker {
52  uint8_t set;
53  float azim;
54  float elev;
56 
57 typedef struct SOFAlizerContext {
58  const AVClass *class;
59 
60  char *filename; /* name of SOFA file */
61  MySofa sofa; /* contains data of the SOFA file */
62 
63  int sample_rate; /* sample rate from SOFA file */
64  float *speaker_azim; /* azimuth of the virtual loudspeakers */
65  float *speaker_elev; /* elevation of the virtual loudspeakers */
66  char *speakers_pos; /* custom positions of the virtual loudspeakers */
67  float lfe_gain; /* initial gain for the LFE channel */
68  float gain_lfe; /* gain applied to LFE channel */
69  int lfe_channel; /* LFE channel position in channel layout */
70 
71  int n_conv; /* number of channels to convolute */
72 
73  /* buffer variables (for convolution) */
74  float *ringbuffer[2]; /* buffers input samples, length of one buffer: */
75  /* no. input ch. (incl. LFE) x buffer_length */
76  int write[2]; /* current write position to ringbuffer */
77  int buffer_length; /* is: longest IR plus max. delay in all SOFA files */
78  /* then choose next power of 2 */
79  int n_fft; /* number of samples in one FFT block */
80 
81  /* netCDF variables */
82  int *delay[2]; /* broadband delay for each channel/IR to be convolved */
83 
84  float *data_ir[2]; /* IRs for all channels to be convolved */
85  /* (this excludes the LFE) */
86  float *temp_src[2];
87  FFTComplex *temp_fft[2];
88 
89  /* control variables */
90  float gain; /* filter gain (in dB) */
91  float rotation; /* rotation of virtual loudspeakers (in degrees) */
92  float elevation; /* elevation of virtual loudspeakers (in deg.) */
93  float radius; /* distance virtual loudspeakers to listener (in metres) */
94  int type; /* processing type */
95 
96  VirtualSpeaker vspkrpos[64];
97 
98  FFTContext *fft[2], *ifft[2];
99  FFTComplex *data_hrtf[2];
100 
103 
104 static int close_sofa(struct MySofa *sofa)
105 {
106  mysofa_close(sofa->easy);
107  sofa->easy = NULL;
108 
109  return 0;
110 }
111 
112 static int preload_sofa(AVFilterContext *ctx, char *filename, int *samplingrate)
113 {
114  struct SOFAlizerContext *s = ctx->priv;
115  struct MYSOFA_HRTF *mysofa;
116  int ret;
117 
118  mysofa = mysofa_load(filename, &ret);
119  if (ret || !mysofa) {
120  av_log(ctx, AV_LOG_ERROR, "Can't find SOFA-file '%s'\n", filename);
121  return AVERROR(EINVAL);
122  }
123 
124  if (mysofa->DataSamplingRate.elements != 1)
125  return AVERROR(EINVAL);
126  *samplingrate = mysofa->DataSamplingRate.values[0];
127  s->sofa.n_samples = mysofa->N;
128  mysofa_free(mysofa);
129 
130  return 0;
131 }
132 
133 static int parse_channel_name(char **arg, int *rchannel, char *buf)
134 {
135  int len, i, channel_id = 0;
136  int64_t layout, layout0;
137 
138  /* try to parse a channel name, e.g. "FL" */
139  if (sscanf(*arg, "%7[A-Z]%n", buf, &len)) {
140  layout0 = layout = av_get_channel_layout(buf);
141  /* channel_id <- first set bit in layout */
142  for (i = 32; i > 0; i >>= 1) {
143  if (layout >= 1LL << i) {
144  channel_id += i;
145  layout >>= i;
146  }
147  }
148  /* reject layouts that are not a single channel */
149  if (channel_id >= 64 || layout0 != 1LL << channel_id)
150  return AVERROR(EINVAL);
151  *rchannel = channel_id;
152  *arg += len;
153  return 0;
154  }
155  return AVERROR(EINVAL);
156 }
157 
158 static void parse_speaker_pos(AVFilterContext *ctx, int64_t in_channel_layout)
159 {
160  SOFAlizerContext *s = ctx->priv;
161  char *arg, *tokenizer, *p, *args = av_strdup(s->speakers_pos);
162 
163  if (!args)
164  return;
165  p = args;
166 
167  while ((arg = av_strtok(p, "|", &tokenizer))) {
168  char buf[8];
169  float azim, elev;
170  int out_ch_id;
171 
172  p = NULL;
173  if (parse_channel_name(&arg, &out_ch_id, buf)) {
174  av_log(ctx, AV_LOG_WARNING, "Failed to parse \'%s\' as channel name.\n", buf);
175  continue;
176  }
177  if (sscanf(arg, "%f %f", &azim, &elev) == 2) {
178  s->vspkrpos[out_ch_id].set = 1;
179  s->vspkrpos[out_ch_id].azim = azim;
180  s->vspkrpos[out_ch_id].elev = elev;
181  } else if (sscanf(arg, "%f", &azim) == 1) {
182  s->vspkrpos[out_ch_id].set = 1;
183  s->vspkrpos[out_ch_id].azim = azim;
184  s->vspkrpos[out_ch_id].elev = 0;
185  }
186  }
187 
188  av_free(args);
189 }
190 
192  float *speaker_azim, float *speaker_elev)
193 {
194  struct SOFAlizerContext *s = ctx->priv;
195  uint64_t channels_layout = ctx->inputs[0]->channel_layout;
196  float azim[16] = { 0 };
197  float elev[16] = { 0 };
198  int m, ch, n_conv = ctx->inputs[0]->channels; /* get no. input channels */
199 
200  if (n_conv > 16)
201  return AVERROR(EINVAL);
202 
203  s->lfe_channel = -1;
204 
205  if (s->speakers_pos)
206  parse_speaker_pos(ctx, channels_layout);
207 
208  /* set speaker positions according to input channel configuration: */
209  for (m = 0, ch = 0; ch < n_conv && m < 64; m++) {
210  uint64_t mask = channels_layout & (1ULL << m);
211 
212  switch (mask) {
213  case AV_CH_FRONT_LEFT: azim[ch] = 30; break;
214  case AV_CH_FRONT_RIGHT: azim[ch] = 330; break;
215  case AV_CH_FRONT_CENTER: azim[ch] = 0; break;
216  case AV_CH_LOW_FREQUENCY:
217  case AV_CH_LOW_FREQUENCY_2: s->lfe_channel = ch; break;
218  case AV_CH_BACK_LEFT: azim[ch] = 150; break;
219  case AV_CH_BACK_RIGHT: azim[ch] = 210; break;
220  case AV_CH_BACK_CENTER: azim[ch] = 180; break;
221  case AV_CH_SIDE_LEFT: azim[ch] = 90; break;
222  case AV_CH_SIDE_RIGHT: azim[ch] = 270; break;
223  case AV_CH_FRONT_LEFT_OF_CENTER: azim[ch] = 15; break;
224  case AV_CH_FRONT_RIGHT_OF_CENTER: azim[ch] = 345; break;
225  case AV_CH_TOP_CENTER: azim[ch] = 0;
226  elev[ch] = 90; break;
227  case AV_CH_TOP_FRONT_LEFT: azim[ch] = 30;
228  elev[ch] = 45; break;
229  case AV_CH_TOP_FRONT_CENTER: azim[ch] = 0;
230  elev[ch] = 45; break;
231  case AV_CH_TOP_FRONT_RIGHT: azim[ch] = 330;
232  elev[ch] = 45; break;
233  case AV_CH_TOP_BACK_LEFT: azim[ch] = 150;
234  elev[ch] = 45; break;
235  case AV_CH_TOP_BACK_RIGHT: azim[ch] = 210;
236  elev[ch] = 45; break;
237  case AV_CH_TOP_BACK_CENTER: azim[ch] = 180;
238  elev[ch] = 45; break;
239  case AV_CH_WIDE_LEFT: azim[ch] = 90; break;
240  case AV_CH_WIDE_RIGHT: azim[ch] = 270; break;
241  case AV_CH_SURROUND_DIRECT_LEFT: azim[ch] = 90; break;
242  case AV_CH_SURROUND_DIRECT_RIGHT: azim[ch] = 270; break;
243  case AV_CH_STEREO_LEFT: azim[ch] = 90; break;
244  case AV_CH_STEREO_RIGHT: azim[ch] = 270; break;
245  case 0: break;
246  default:
247  return AVERROR(EINVAL);
248  }
249 
250  if (s->vspkrpos[m].set) {
251  azim[ch] = s->vspkrpos[m].azim;
252  elev[ch] = s->vspkrpos[m].elev;
253  }
254 
255  if (mask)
256  ch++;
257  }
258 
259  memcpy(speaker_azim, azim, n_conv * sizeof(float));
260  memcpy(speaker_elev, elev, n_conv * sizeof(float));
261 
262  return 0;
263 
264 }
265 
266 typedef struct ThreadData {
267  AVFrame *in, *out;
268  int *write;
269  int **delay;
270  float **ir;
271  int *n_clippings;
272  float **ringbuffer;
273  float **temp_src;
274  FFTComplex **temp_fft;
275 } ThreadData;
276 
277 static int sofalizer_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
278 {
279  SOFAlizerContext *s = ctx->priv;
280  ThreadData *td = arg;
281  AVFrame *in = td->in, *out = td->out;
282  int offset = jobnr;
283  int *write = &td->write[jobnr];
284  const int *const delay = td->delay[jobnr];
285  const float *const ir = td->ir[jobnr];
286  int *n_clippings = &td->n_clippings[jobnr];
287  float *ringbuffer = td->ringbuffer[jobnr];
288  float *temp_src = td->temp_src[jobnr];
289  const int n_samples = s->sofa.n_samples; /* length of one IR */
290  const float *src = (const float *)in->data[0]; /* get pointer to audio input buffer */
291  float *dst = (float *)out->data[0]; /* get pointer to audio output buffer */
292  const int in_channels = s->n_conv; /* number of input channels */
293  /* ring buffer length is: longest IR plus max. delay -> next power of 2 */
294  const int buffer_length = s->buffer_length;
295  /* -1 for AND instead of MODULO (applied to powers of 2): */
296  const uint32_t modulo = (uint32_t)buffer_length - 1;
297  float *buffer[16]; /* holds ringbuffer for each input channel */
298  int wr = *write;
299  int read;
300  int i, l;
301 
302  dst += offset;
303  for (l = 0; l < in_channels; l++) {
304  /* get starting address of ringbuffer for each input channel */
305  buffer[l] = ringbuffer + l * buffer_length;
306  }
307 
308  for (i = 0; i < in->nb_samples; i++) {
309  const float *temp_ir = ir; /* using same set of IRs for each sample */
310 
311  dst[0] = 0;
312  for (l = 0; l < in_channels; l++) {
313  /* write current input sample to ringbuffer (for each channel) */
314  buffer[l][wr] = src[l];
315  }
316 
317  /* loop goes through all channels to be convolved */
318  for (l = 0; l < in_channels; l++) {
319  const float *const bptr = buffer[l];
320 
321  if (l == s->lfe_channel) {
322  /* LFE is an input channel but requires no convolution */
323  /* apply gain to LFE signal and add to output buffer */
324  *dst += *(buffer[s->lfe_channel] + wr) * s->gain_lfe;
325  temp_ir += FFALIGN(n_samples, 32);
326  continue;
327  }
328 
329  /* current read position in ringbuffer: input sample write position
330  * - delay for l-th ch. + diff. betw. IR length and buffer length
331  * (mod buffer length) */
332  read = (wr - delay[l] - (n_samples - 1) + buffer_length) & modulo;
333 
334  if (read + n_samples < buffer_length) {
335  memmove(temp_src, bptr + read, n_samples * sizeof(*temp_src));
336  } else {
337  int len = FFMIN(n_samples - (read % n_samples), buffer_length - read);
338 
339  memmove(temp_src, bptr + read, len * sizeof(*temp_src));
340  memmove(temp_src + len, bptr, (n_samples - len) * sizeof(*temp_src));
341  }
342 
343  /* multiply signal and IR, and add up the results */
344  dst[0] += s->fdsp->scalarproduct_float(temp_ir, temp_src, n_samples);
345  temp_ir += FFALIGN(n_samples, 32);
346  }
347 
348  /* clippings counter */
349  if (fabs(dst[0]) > 1)
350  *n_clippings += 1;
351 
352  /* move output buffer pointer by +2 to get to next sample of processed channel: */
353  dst += 2;
354  src += in_channels;
355  wr = (wr + 1) & modulo; /* update ringbuffer write position */
356  }
357 
358  *write = wr; /* remember write position in ringbuffer for next call */
359 
360  return 0;
361 }
362 
363 static int sofalizer_fast_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
364 {
365  SOFAlizerContext *s = ctx->priv;
366  ThreadData *td = arg;
367  AVFrame *in = td->in, *out = td->out;
368  int offset = jobnr;
369  int *write = &td->write[jobnr];
370  FFTComplex *hrtf = s->data_hrtf[jobnr]; /* get pointers to current HRTF data */
371  int *n_clippings = &td->n_clippings[jobnr];
372  float *ringbuffer = td->ringbuffer[jobnr];
373  const int n_samples = s->sofa.n_samples; /* length of one IR */
374  const float *src = (const float *)in->data[0]; /* get pointer to audio input buffer */
375  float *dst = (float *)out->data[0]; /* get pointer to audio output buffer */
376  const int in_channels = s->n_conv; /* number of input channels */
377  /* ring buffer length is: longest IR plus max. delay -> next power of 2 */
378  const int buffer_length = s->buffer_length;
379  /* -1 for AND instead of MODULO (applied to powers of 2): */
380  const uint32_t modulo = (uint32_t)buffer_length - 1;
381  FFTComplex *fft_in = s->temp_fft[jobnr]; /* temporary array for FFT input/output data */
382  FFTContext *ifft = s->ifft[jobnr];
383  FFTContext *fft = s->fft[jobnr];
384  const int n_conv = s->n_conv;
385  const int n_fft = s->n_fft;
386  const float fft_scale = 1.0f / s->n_fft;
387  FFTComplex *hrtf_offset;
388  int wr = *write;
389  int n_read;
390  int i, j;
391 
392  dst += offset;
393 
394  /* find minimum between number of samples and output buffer length:
395  * (important, if one IR is longer than the output buffer) */
396  n_read = FFMIN(s->sofa.n_samples, in->nb_samples);
397  for (j = 0; j < n_read; j++) {
398  /* initialize output buf with saved signal from overflow buf */
399  dst[2 * j] = ringbuffer[wr];
400  ringbuffer[wr] = 0.0; /* re-set read samples to zero */
401  /* update ringbuffer read/write position */
402  wr = (wr + 1) & modulo;
403  }
404 
405  /* initialize rest of output buffer with 0 */
406  for (j = n_read; j < in->nb_samples; j++) {
407  dst[2 * j] = 0;
408  }
409 
410  for (i = 0; i < n_conv; i++) {
411  if (i == s->lfe_channel) { /* LFE */
412  for (j = 0; j < in->nb_samples; j++) {
413  /* apply gain to LFE signal and add to output buffer */
414  dst[2 * j] += src[i + j * in_channels] * s->gain_lfe;
415  }
416  continue;
417  }
418 
419  /* outer loop: go through all input channels to be convolved */
420  offset = i * n_fft; /* no. samples already processed */
421  hrtf_offset = hrtf + offset;
422 
423  /* fill FFT input with 0 (we want to zero-pad) */
424  memset(fft_in, 0, sizeof(FFTComplex) * n_fft);
425 
426  for (j = 0; j < in->nb_samples; j++) {
427  /* prepare input for FFT */
428  /* write all samples of current input channel to FFT input array */
429  fft_in[j].re = src[j * in_channels + i];
430  }
431 
432  /* transform input signal of current channel to frequency domain */
433  av_fft_permute(fft, fft_in);
434  av_fft_calc(fft, fft_in);
435  for (j = 0; j < n_fft; j++) {
436  const FFTComplex *hcomplex = hrtf_offset + j;
437  const float re = fft_in[j].re;
438  const float im = fft_in[j].im;
439 
440  /* complex multiplication of input signal and HRTFs */
441  /* output channel (real): */
442  fft_in[j].re = re * hcomplex->re - im * hcomplex->im;
443  /* output channel (imag): */
444  fft_in[j].im = re * hcomplex->im + im * hcomplex->re;
445  }
446 
447  /* transform output signal of current channel back to time domain */
448  av_fft_permute(ifft, fft_in);
449  av_fft_calc(ifft, fft_in);
450 
451  for (j = 0; j < in->nb_samples; j++) {
452  /* write output signal of current channel to output buffer */
453  dst[2 * j] += fft_in[j].re * fft_scale;
454  }
455 
456  for (j = 0; j < n_samples - 1; j++) { /* overflow length is IR length - 1 */
457  /* write the rest of output signal to overflow buffer */
458  int write_pos = (wr + j) & modulo;
459 
460  *(ringbuffer + write_pos) += fft_in[in->nb_samples + j].re * fft_scale;
461  }
462  }
463 
464  /* go through all samples of current output buffer: count clippings */
465  for (i = 0; i < out->nb_samples; i++) {
466  /* clippings counter */
467  if (fabs(*dst) > 1) { /* if current output sample > 1 */
468  n_clippings[0]++;
469  }
470 
471  /* move output buffer pointer by +2 to get to next sample of processed channel: */
472  dst += 2;
473  }
474 
475  /* remember read/write position in ringbuffer for next call */
476  *write = wr;
477 
478  return 0;
479 }
480 
481 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
482 {
483  AVFilterContext *ctx = inlink->dst;
484  SOFAlizerContext *s = ctx->priv;
485  AVFilterLink *outlink = ctx->outputs[0];
486  int n_clippings[2] = { 0 };
487  ThreadData td;
488  AVFrame *out;
489 
490  out = ff_get_audio_buffer(outlink, in->nb_samples);
491  if (!out) {
492  av_frame_free(&in);
493  return AVERROR(ENOMEM);
494  }
495  av_frame_copy_props(out, in);
496 
497  td.in = in; td.out = out; td.write = s->write;
498  td.delay = s->delay; td.ir = s->data_ir; td.n_clippings = n_clippings;
499  td.ringbuffer = s->ringbuffer; td.temp_src = s->temp_src;
500  td.temp_fft = s->temp_fft;
501 
502  if (s->type == TIME_DOMAIN) {
503  ctx->internal->execute(ctx, sofalizer_convolute, &td, NULL, 2);
504  } else {
505  ctx->internal->execute(ctx, sofalizer_fast_convolute, &td, NULL, 2);
506  }
507  emms_c();
508 
509  /* display error message if clipping occurred */
510  if (n_clippings[0] + n_clippings[1] > 0) {
511  av_log(ctx, AV_LOG_WARNING, "%d of %d samples clipped. Please reduce gain.\n",
512  n_clippings[0] + n_clippings[1], out->nb_samples * 2);
513  }
514 
515  av_frame_free(&in);
516  return ff_filter_frame(outlink, out);
517 }
518 
520 {
521  struct SOFAlizerContext *s = ctx->priv;
524  int ret, sample_rates[] = { 48000, -1 };
525 
526  ret = ff_add_format(&formats, AV_SAMPLE_FMT_FLT);
527  if (ret)
528  return ret;
529  ret = ff_set_common_formats(ctx, formats);
530  if (ret)
531  return ret;
532 
533  layouts = ff_all_channel_layouts();
534  if (!layouts)
535  return AVERROR(ENOMEM);
536 
537  ret = ff_channel_layouts_ref(layouts, &ctx->inputs[0]->out_channel_layouts);
538  if (ret)
539  return ret;
540 
541  layouts = NULL;
543  if (ret)
544  return ret;
545 
546  ret = ff_channel_layouts_ref(layouts, &ctx->outputs[0]->in_channel_layouts);
547  if (ret)
548  return ret;
549 
550  sample_rates[0] = s->sample_rate;
551  formats = ff_make_format_list(sample_rates);
552  if (!formats)
553  return AVERROR(ENOMEM);
554  return ff_set_common_samplerates(ctx, formats);
555 }
556 
557 static int load_data(AVFilterContext *ctx, int azim, int elev, float radius, int sample_rate)
558 {
559  struct SOFAlizerContext *s = ctx->priv;
560  int n_samples;
561  int n_conv = s->n_conv; /* no. channels to convolve */
562  int n_fft;
563  float delay_l; /* broadband delay for each IR */
564  float delay_r;
565  int nb_input_channels = ctx->inputs[0]->channels; /* no. input channels */
566  float gain_lin = expf((s->gain - 3 * nb_input_channels) / 20 * M_LN10); /* gain - 3dB/channel */
567  FFTComplex *data_hrtf_l = NULL;
568  FFTComplex *data_hrtf_r = NULL;
569  FFTComplex *fft_in_l = NULL;
570  FFTComplex *fft_in_r = NULL;
571  float *data_ir_l = NULL;
572  float *data_ir_r = NULL;
573  int offset = 0; /* used for faster pointer arithmetics in for-loop */
574  int i, j, azim_orig = azim, elev_orig = elev;
575  int filter_length, ret = 0;
576  int n_current;
577  int n_max = 0;
578 
579  s->sofa.easy = mysofa_open(s->filename, sample_rate, &filter_length, &ret);
580  if (!s->sofa.easy || ret) { /* if an invalid SOFA file has been selected */
581  av_log(ctx, AV_LOG_ERROR, "Selected SOFA file is invalid. Please select valid SOFA file.\n");
582  return AVERROR_INVALIDDATA;
583  }
584 
585  n_samples = s->sofa.n_samples;
586 
587  s->data_ir[0] = av_calloc(FFALIGN(n_samples, 32), sizeof(float) * s->n_conv);
588  s->data_ir[1] = av_calloc(FFALIGN(n_samples, 32), sizeof(float) * s->n_conv);
589  s->delay[0] = av_calloc(s->n_conv, sizeof(int));
590  s->delay[1] = av_calloc(s->n_conv, sizeof(int));
591 
592  if (!s->data_ir[0] || !s->data_ir[1] || !s->delay[0] || !s->delay[1]) {
593  ret = AVERROR(ENOMEM);
594  goto fail;
595  }
596 
597  /* get temporary IR for L and R channel */
598  data_ir_l = av_calloc(n_conv * FFALIGN(n_samples, 32), sizeof(*data_ir_l));
599  data_ir_r = av_calloc(n_conv * FFALIGN(n_samples, 32), sizeof(*data_ir_r));
600  if (!data_ir_r || !data_ir_l) {
601  ret = AVERROR(ENOMEM);
602  goto fail;
603  }
604 
605  if (s->type == TIME_DOMAIN) {
606  s->temp_src[0] = av_calloc(FFALIGN(n_samples, 32), sizeof(float));
607  s->temp_src[1] = av_calloc(FFALIGN(n_samples, 32), sizeof(float));
608  if (!s->temp_src[0] || !s->temp_src[1]) {
609  ret = AVERROR(ENOMEM);
610  goto fail;
611  }
612  }
613 
614  s->speaker_azim = av_calloc(s->n_conv, sizeof(*s->speaker_azim));
615  s->speaker_elev = av_calloc(s->n_conv, sizeof(*s->speaker_elev));
616  if (!s->speaker_azim || !s->speaker_elev) {
617  ret = AVERROR(ENOMEM);
618  goto fail;
619  }
620 
621  /* get speaker positions */
622  if ((ret = get_speaker_pos(ctx, s->speaker_azim, s->speaker_elev)) < 0) {
623  av_log(ctx, AV_LOG_ERROR, "Couldn't get speaker positions. Input channel configuration not supported.\n");
624  goto fail;
625  }
626 
627  for (i = 0; i < s->n_conv; i++) {
628  float coordinates[3];
629 
630  /* load and store IRs and corresponding delays */
631  azim = (int)(s->speaker_azim[i] + azim_orig) % 360;
632  elev = (int)(s->speaker_elev[i] + elev_orig) % 90;
633 
634  coordinates[0] = azim;
635  coordinates[1] = elev;
636  coordinates[2] = radius;
637 
638  mysofa_s2c(coordinates);
639 
640  /* get id of IR closest to desired position */
641  mysofa_getfilter_float(s->sofa.easy, coordinates[0], coordinates[1], coordinates[2],
642  data_ir_l + FFALIGN(n_samples, 32) * i,
643  data_ir_r + FFALIGN(n_samples, 32) * i,
644  &delay_l, &delay_r);
645 
646  s->delay[0][i] = delay_l * sample_rate;
647  s->delay[1][i] = delay_r * sample_rate;
648 
649  s->sofa.max_delay = FFMAX3(s->sofa.max_delay, s->delay[0][i], s->delay[1][i]);
650  }
651 
652  /* get size of ringbuffer (longest IR plus max. delay) */
653  /* then choose next power of 2 for performance optimization */
654  n_current = s->sofa.n_samples + s->sofa.max_delay;
655  /* length of longest IR plus max. delay */
656  n_max = FFMAX(n_max, n_current);
657 
658  /* buffer length is longest IR plus max. delay -> next power of 2
659  (32 - count leading zeros gives required exponent) */
660  s->buffer_length = 1 << (32 - ff_clz(n_max));
661  s->n_fft = n_fft = 1 << (32 - ff_clz(n_max + sample_rate));
662 
663  if (s->type == FREQUENCY_DOMAIN) {
664  av_fft_end(s->fft[0]);
665  av_fft_end(s->fft[1]);
666  s->fft[0] = av_fft_init(log2(s->n_fft), 0);
667  s->fft[1] = av_fft_init(log2(s->n_fft), 0);
668  av_fft_end(s->ifft[0]);
669  av_fft_end(s->ifft[1]);
670  s->ifft[0] = av_fft_init(log2(s->n_fft), 1);
671  s->ifft[1] = av_fft_init(log2(s->n_fft), 1);
672 
673  if (!s->fft[0] || !s->fft[1] || !s->ifft[0] || !s->ifft[1]) {
674  av_log(ctx, AV_LOG_ERROR, "Unable to create FFT contexts of size %d.\n", s->n_fft);
675  ret = AVERROR(ENOMEM);
676  goto fail;
677  }
678  }
679 
680  if (s->type == TIME_DOMAIN) {
681  s->ringbuffer[0] = av_calloc(s->buffer_length, sizeof(float) * nb_input_channels);
682  s->ringbuffer[1] = av_calloc(s->buffer_length, sizeof(float) * nb_input_channels);
683  } else {
684  /* get temporary HRTF memory for L and R channel */
685  data_hrtf_l = av_malloc_array(n_fft, sizeof(*data_hrtf_l) * n_conv);
686  data_hrtf_r = av_malloc_array(n_fft, sizeof(*data_hrtf_r) * n_conv);
687  if (!data_hrtf_r || !data_hrtf_l) {
688  ret = AVERROR(ENOMEM);
689  goto fail;
690  }
691 
692  s->ringbuffer[0] = av_calloc(s->buffer_length, sizeof(float));
693  s->ringbuffer[1] = av_calloc(s->buffer_length, sizeof(float));
694  s->temp_fft[0] = av_malloc_array(s->n_fft, sizeof(FFTComplex));
695  s->temp_fft[1] = av_malloc_array(s->n_fft, sizeof(FFTComplex));
696  if (!s->temp_fft[0] || !s->temp_fft[1]) {
697  ret = AVERROR(ENOMEM);
698  goto fail;
699  }
700  }
701 
702  if (!s->ringbuffer[0] || !s->ringbuffer[1]) {
703  ret = AVERROR(ENOMEM);
704  goto fail;
705  }
706 
707  if (s->type == FREQUENCY_DOMAIN) {
708  fft_in_l = av_calloc(n_fft, sizeof(*fft_in_l));
709  fft_in_r = av_calloc(n_fft, sizeof(*fft_in_r));
710  if (!fft_in_l || !fft_in_r) {
711  ret = AVERROR(ENOMEM);
712  goto fail;
713  }
714  }
715 
716  for (i = 0; i < s->n_conv; i++) {
717  float *lir, *rir;
718 
719  offset = i * FFALIGN(n_samples, 32); /* no. samples already written */
720 
721  lir = data_ir_l + offset;
722  rir = data_ir_r + offset;
723 
724  if (s->type == TIME_DOMAIN) {
725  for (j = 0; j < n_samples; j++) {
726  /* load reversed IRs of the specified source position
727  * sample-by-sample for left and right ear; and apply gain */
728  s->data_ir[0][offset + j] = lir[n_samples - 1 - j] * gain_lin;
729  s->data_ir[1][offset + j] = rir[n_samples - 1 - j] * gain_lin;
730  }
731  } else {
732  memset(fft_in_l, 0, n_fft * sizeof(*fft_in_l));
733  memset(fft_in_r, 0, n_fft * sizeof(*fft_in_r));
734 
735  offset = i * n_fft; /* no. samples already written */
736  for (j = 0; j < n_samples; j++) {
737  /* load non-reversed IRs of the specified source position
738  * sample-by-sample and apply gain,
739  * L channel is loaded to real part, R channel to imag part,
740  * IRs ared shifted by L and R delay */
741  fft_in_l[s->delay[0][i] + j].re = lir[j] * gain_lin;
742  fft_in_r[s->delay[1][i] + j].re = rir[j] * gain_lin;
743  }
744 
745  /* actually transform to frequency domain (IRs -> HRTFs) */
746  av_fft_permute(s->fft[0], fft_in_l);
747  av_fft_calc(s->fft[0], fft_in_l);
748  memcpy(data_hrtf_l + offset, fft_in_l, n_fft * sizeof(*fft_in_l));
749  av_fft_permute(s->fft[0], fft_in_r);
750  av_fft_calc(s->fft[0], fft_in_r);
751  memcpy(data_hrtf_r + offset, fft_in_r, n_fft * sizeof(*fft_in_r));
752  }
753  }
754 
755  if (s->type == FREQUENCY_DOMAIN) {
756  s->data_hrtf[0] = av_malloc_array(n_fft * s->n_conv, sizeof(FFTComplex));
757  s->data_hrtf[1] = av_malloc_array(n_fft * s->n_conv, sizeof(FFTComplex));
758  if (!s->data_hrtf[0] || !s->data_hrtf[1]) {
759  ret = AVERROR(ENOMEM);
760  goto fail;
761  }
762 
763  memcpy(s->data_hrtf[0], data_hrtf_l, /* copy HRTF data to */
764  sizeof(FFTComplex) * n_conv * n_fft); /* filter struct */
765  memcpy(s->data_hrtf[1], data_hrtf_r,
766  sizeof(FFTComplex) * n_conv * n_fft);
767  }
768 
769 fail:
770  av_freep(&data_hrtf_l); /* free temporary HRTF memory */
771  av_freep(&data_hrtf_r);
772 
773  av_freep(&data_ir_l); /* free temprary IR memory */
774  av_freep(&data_ir_r);
775 
776  av_freep(&fft_in_l); /* free temporary FFT memory */
777  av_freep(&fft_in_r);
778 
779  return ret;
780 }
781 
783 {
784  SOFAlizerContext *s = ctx->priv;
785  int ret;
786 
787  if (!s->filename) {
788  av_log(ctx, AV_LOG_ERROR, "Valid SOFA filename must be set.\n");
789  return AVERROR(EINVAL);
790  }
791 
792  /* preload SOFA file, */
793  ret = preload_sofa(ctx, s->filename, &s->sample_rate);
794  if (ret) {
795  /* file loading error */
796  av_log(ctx, AV_LOG_ERROR, "Error while loading SOFA file: '%s'\n", s->filename);
797  } else { /* no file loading error, resampling not required */
798  av_log(ctx, AV_LOG_DEBUG, "File '%s' loaded.\n", s->filename);
799  }
800 
801  if (ret) {
802  av_log(ctx, AV_LOG_ERROR, "No valid SOFA file could be loaded. Please specify valid SOFA file.\n");
803  return ret;
804  }
805 
807  if (!s->fdsp)
808  return AVERROR(ENOMEM);
809 
810  return 0;
811 }
812 
813 static int config_input(AVFilterLink *inlink)
814 {
815  AVFilterContext *ctx = inlink->dst;
816  SOFAlizerContext *s = ctx->priv;
817  int ret;
818 
819  if (s->type == FREQUENCY_DOMAIN) {
820  inlink->partial_buf_size =
821  inlink->min_samples =
822  inlink->max_samples = inlink->sample_rate;
823  }
824 
825  /* gain -3 dB per channel, -6 dB to get LFE on a similar level */
826  s->gain_lfe = expf((s->gain - 3 * inlink->channels - 6 + s->lfe_gain) / 20 * M_LN10);
827 
828  s->n_conv = inlink->channels;
829 
830  /* load IRs to data_ir[0] and data_ir[1] for required directions */
831  if ((ret = load_data(ctx, s->rotation, s->elevation, s->radius, inlink->sample_rate)) < 0)
832  return ret;
833 
834  av_log(ctx, AV_LOG_DEBUG, "Samplerate: %d Channels to convolute: %d, Length of ringbuffer: %d x %d\n",
835  inlink->sample_rate, s->n_conv, inlink->channels, s->buffer_length);
836 
837  return 0;
838 }
839 
841 {
842  SOFAlizerContext *s = ctx->priv;
843 
844  close_sofa(&s->sofa);
845  av_fft_end(s->ifft[0]);
846  av_fft_end(s->ifft[1]);
847  av_fft_end(s->fft[0]);
848  av_fft_end(s->fft[1]);
849  av_freep(&s->delay[0]);
850  av_freep(&s->delay[1]);
851  av_freep(&s->data_ir[0]);
852  av_freep(&s->data_ir[1]);
853  av_freep(&s->ringbuffer[0]);
854  av_freep(&s->ringbuffer[1]);
855  av_freep(&s->speaker_azim);
856  av_freep(&s->speaker_elev);
857  av_freep(&s->temp_src[0]);
858  av_freep(&s->temp_src[1]);
859  av_freep(&s->temp_fft[0]);
860  av_freep(&s->temp_fft[1]);
861  av_freep(&s->data_hrtf[0]);
862  av_freep(&s->data_hrtf[1]);
863  av_freep(&s->fdsp);
864 }
865 
866 #define OFFSET(x) offsetof(SOFAlizerContext, x)
867 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
868 
869 static const AVOption sofalizer_options[] = {
870  { "sofa", "sofa filename", OFFSET(filename), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
871  { "gain", "set gain in dB", OFFSET(gain), AV_OPT_TYPE_FLOAT, {.dbl=0}, -20, 40, .flags = FLAGS },
872  { "rotation", "set rotation" , OFFSET(rotation), AV_OPT_TYPE_FLOAT, {.dbl=0}, -360, 360, .flags = FLAGS },
873  { "elevation", "set elevation", OFFSET(elevation), AV_OPT_TYPE_FLOAT, {.dbl=0}, -90, 90, .flags = FLAGS },
874  { "radius", "set radius", OFFSET(radius), AV_OPT_TYPE_FLOAT, {.dbl=1}, 0, 3, .flags = FLAGS },
875  { "type", "set processing", OFFSET(type), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, .flags = FLAGS, "type" },
876  { "time", "time domain", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, .flags = FLAGS, "type" },
877  { "freq", "frequency domain", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, .flags = FLAGS, "type" },
878  { "speakers", "set speaker custom positions", OFFSET(speakers_pos), AV_OPT_TYPE_STRING, {.str=0}, 0, 0, .flags = FLAGS },
879  { "lfegain", "set lfe gain", OFFSET(lfe_gain), AV_OPT_TYPE_FLOAT, {.dbl=0}, -9, 9, .flags = FLAGS },
880  { NULL }
881 };
882 
883 AVFILTER_DEFINE_CLASS(sofalizer);
884 
885 static const AVFilterPad inputs[] = {
886  {
887  .name = "default",
888  .type = AVMEDIA_TYPE_AUDIO,
889  .config_props = config_input,
890  .filter_frame = filter_frame,
891  },
892  { NULL }
893 };
894 
895 static const AVFilterPad outputs[] = {
896  {
897  .name = "default",
898  .type = AVMEDIA_TYPE_AUDIO,
899  },
900  { NULL }
901 };
902 
904  .name = "sofalizer",
905  .description = NULL_IF_CONFIG_SMALL("SOFAlizer (Spatially Oriented Format for Acoustics)."),
906  .priv_size = sizeof(SOFAlizerContext),
907  .priv_class = &sofalizer_class,
908  .init = init,
909  .uninit = uninit,
911  .inputs = inputs,
912  .outputs = outputs,
914 };
static int sofalizer_fast_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
Definition: af_sofalizer.c:363
float(* scalarproduct_float)(const float *v1, const float *v2, int len)
Calculate the scalar product of two vectors of floats.
Definition: float_dsp.h:175
#define NULL
Definition: coverity.c:32
FFTComplex * data_hrtf[2]
Definition: af_sofalizer.c:99
const char * s
Definition: avisynth_c.h:768
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:59
AVFrame * out
Definition: af_aiir.c:31
This structure describes decoded (raw) audio or video data.
Definition: frame.h:218
#define AV_CH_TOP_FRONT_RIGHT
AVOption.
Definition: opt.h:246
av_cold void av_fft_end(FFTContext *s)
Definition: avfft.c:48
float re
Definition: fft.c:82
float ** temp_src
Definition: af_headphone.c:160
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:182
Main libavfilter public API header.
AVFILTER_DEFINE_CLASS(sofalizer)
#define AV_CH_TOP_FRONT_LEFT
FFTContext * fft[2]
Definition: af_sofalizer.c:98
#define AV_CH_TOP_FRONT_CENTER
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(const uint8_t *) pi - 0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(const int16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(const int16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(const int32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(const int32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(const int64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0f/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64, *(const int64_t *) pi *(1.0/(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float *) pi *(INT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double *) pi *(INT64_C(1)<< 63))) #define FMT_PAIR_FUNC(out, in) static conv_func_type *const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB *AV_SAMPLE_FMT_NB]={ FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64), };static void cpy1(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, len);} static void cpy2(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 2 *len);} static void cpy4(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 4 *len);} static void cpy8(uint8_t **dst, const uint8_t **src, int len){ memcpy(*dst, *src, 8 *len);} AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, const int *ch_map, int flags) { AudioConvert *ctx;conv_func_type *f=fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt)+AV_SAMPLE_FMT_NB *av_get_packed_sample_fmt(in_fmt)];if(!f) return NULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) return NULL;if(channels==1){ in_fmt=av_get_planar_sample_fmt(in_fmt);out_fmt=av_get_planar_sample_fmt(out_fmt);} ctx->channels=channels;ctx->conv_f=f;ctx->ch_map=ch_map;if(in_fmt==AV_SAMPLE_FMT_U8||in_fmt==AV_SAMPLE_FMT_U8P) memset(ctx->silence, 0x80, sizeof(ctx->silence));if(out_fmt==in_fmt &&!ch_map) { switch(av_get_bytes_per_sample(in_fmt)){ case 1:ctx->simd_f=cpy1;break;case 2:ctx->simd_f=cpy2;break;case 4:ctx->simd_f=cpy4;break;case 8:ctx->simd_f=cpy8;break;} } if(HAVE_X86ASM &&HAVE_MMX) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels);if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels);if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels);return ctx;} void swri_audio_convert_free(AudioConvert **ctx) { av_freep(ctx);} int swri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, int len) { int ch;int off=0;const int os=(out->planar ? 1 :out->ch_count) *out->bps;unsigned misaligned=0;av_assert0(ctx->channels==out->ch_count);if(ctx->in_simd_align_mask) { int planes=in->planar ? in->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) in->ch[ch];misaligned|=m &ctx->in_simd_align_mask;} if(ctx->out_simd_align_mask) { int planes=out->planar ? out->ch_count :1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) out->ch[ch];misaligned|=m &ctx->out_simd_align_mask;} if(ctx->simd_f &&!ctx->ch_map &&!misaligned){ off=len &~15;av_assert1(off >=0);av_assert1(off<=len);av_assert2(ctx->channels==SWR_CH_MAX||!in->ch[ctx->channels]);if(off >0){ if(out->planar==in->planar){ int planes=out->planar ? out->ch_count :1;for(ch=0;ch< planes;ch++){ ctx->simd_f(out-> ch ch
Definition: audioconvert.c:56
#define AV_CH_LOW_FREQUENCY_2
FFTSample re
Definition: avfft.h:38
void av_fft_permute(FFTContext *s, FFTComplex *z)
Do the permutation needed BEFORE calling ff_fft_calc().
Definition: avfft.c:38
#define AV_CH_SURROUND_DIRECT_RIGHT
#define AV_CH_LAYOUT_STEREO
AVFilter ff_af_sofalizer
Definition: af_sofalizer.c:903
#define src
Definition: vp8dsp.c:254
#define log2(x)
Definition: libm.h:404
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
Definition: mem.c:244
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:283
static int close_sofa(struct MySofa *sofa)
Definition: af_sofalizer.c:104
const char * name
Pad name.
Definition: internal.h:60
uint64_t av_get_channel_layout(const char *name)
Return a channel layout id that matches name, or 0 if no match is found.
AVFilterLink ** inputs
array of pointers to input links
Definition: avfilter.h:346
int ff_channel_layouts_ref(AVFilterChannelLayouts *f, AVFilterChannelLayouts **ref)
Add *ref as a new reference to f.
Definition: formats.c:435
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1080
float ** ringbuffer
Definition: af_headphone.c:159
AVFrame * in
Definition: af_aiir.c:31
static char buffer[20]
Definition: seek.c:32
#define AV_CH_WIDE_LEFT
uint8_t
#define av_cold
Definition: attributes.h:82
AVOptions.
int ** delay
Definition: af_headphone.c:156
#define AV_CH_TOP_BACK_LEFT
#define AV_CH_WIDE_RIGHT
#define AV_CH_TOP_BACK_CENTER
#define emms_c()
Definition: internal.h:55
#define AV_CH_LOW_FREQUENCY
#define FLAGS
Definition: af_sofalizer.c:867
static int flags
Definition: log.c:55
#define AV_CH_BACK_LEFT
float * rir
Definition: af_sofalizer.c:47
#define FFALIGN(x, a)
Definition: macros.h:48
#define av_log(a,...)
A filter pad used for either input or output.
Definition: internal.h:54
#define expf(x)
Definition: libm.h:283
AVFloatDSPContext * fdsp
Definition: af_sofalizer.c:101
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:176
int ff_set_common_formats(AVFilterContext *ctx, AVFilterFormats *formats)
A helper for query_formats() which sets all links to the same list of formats.
Definition: formats.c:568
#define td
Definition: regdef.h:70
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:127
int ff_add_channel_layout(AVFilterChannelLayouts **l, uint64_t channel_layout)
Definition: formats.c:343
AVFrame * ff_get_audio_buffer(AVFilterLink *link, int nb_samples)
Request an audio samples buffer with a specific set of permissions.
Definition: audio.c:86
static const uint16_t mask[17]
Definition: lzw.c:38
static int get_speaker_pos(AVFilterContext *ctx, float *speaker_azim, float *speaker_elev)
Definition: af_sofalizer.c:191
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
Definition: af_sofalizer.c:481
#define AVERROR(e)
Definition: error.h:43
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:202
static int query_formats(AVFilterContext *ctx)
Definition: af_sofalizer.c:519
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:186
void * priv
private data for use by the filter
Definition: avfilter.h:353
#define AVFILTER_FLAG_SLICE_THREADS
The filter supports multithreading by splitting frames into multiple parts and processing them concur...
Definition: avfilter.h:116
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:197
const char * arg
Definition: jacosubdec.c:66
float * data_ir[2]
Definition: af_sofalizer.c:84
int ff_add_format(AVFilterFormats **avff, int64_t fmt)
Add fmt to the list of media formats contained in *avff.
Definition: formats.c:337
FFTContext * av_fft_init(int nbits, int inverse)
Set up a complex FFT.
Definition: avfft.c:28
static const uint8_t offset[127][2]
Definition: vf_spp.c:92
#define FFMAX(a, b)
Definition: common.h:94
#define fail()
Definition: checkasm.h:116
#define AV_CH_STEREO_RIGHT
See AV_CH_STEREO_LEFT.
#define AV_CH_TOP_CENTER
float * ringbuffer[2]
Definition: af_sofalizer.c:74
float * speaker_azim
Definition: af_sofalizer.c:64
Definition: fft.h:88
static int load_data(AVFilterContext *ctx, int azim, int elev, float radius, int sample_rate)
Definition: af_sofalizer.c:557
audio channel layout utility functions
#define FFMIN(a, b)
Definition: common.h:96
#define ff_clz
Definition: intmath.h:142
const char AVS_Value args
Definition: avisynth_c.h:780
AVFormatContext * ctx
Definition: movenc.c:48
#define TIME_DOMAIN
Definition: af_sofalizer.c:41
int * n_clippings
Definition: af_headphone.c:158
#define AV_CH_FRONT_LEFT_OF_CENTER
#define AV_CH_FRONT_CENTER
static const AVFilterPad outputs[]
Definition: af_sofalizer.c:895
AVFilterChannelLayouts * ff_all_channel_layouts(void)
Construct an empty AVFilterChannelLayouts/AVFilterFormats struct – representing any channel layout (...
Definition: formats.c:401
#define AV_CH_FRONT_RIGHT_OF_CENTER
A list of supported channel layouts.
Definition: formats.h:85
static const AVOption sofalizer_options[]
Definition: af_sofalizer.c:869
FFTComplex ** temp_fft
Definition: af_headphone.c:161
sample_rate
static int preload_sofa(AVFilterContext *ctx, char *filename, int *samplingrate)
Definition: af_sofalizer.c:112
char * av_strdup(const char *s)
Duplicate a string.
Definition: mem.c:251
int max_delay
Definition: af_sofalizer.c:48
FFT functions.
static int config_input(AVFilterLink *inlink)
Definition: af_sofalizer.c:813
#define AV_CH_FRONT_LEFT
void * buf
Definition: avisynth_c.h:690
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(const int16_t *) pi >> 8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(const int32_t *) pi >> 24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) #define SET_CONV_FUNC_GROUP(ofmt, ifmt) static void set_generic_function(AudioConvert *ac) { } void ff_audio_convert_free(AudioConvert **ac) { if(! *ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);} AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, int sample_rate, int apply_map) { AudioConvert *ac;int in_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) return NULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method !=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt) > 2) { ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc) { av_free(ac);return NULL;} return ac;} in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar) { ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar ? ac->channels :1;} else if(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;else ac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);return ac;} int ff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in) { int use_generic=1;int len=in->nb_samples;int p;if(ac->dc) { av_log(ac->avr, AV_LOG_TRACE, "%d samples - audio_convert: %s to %s (dithered)\", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));return ff_convert_dither(ac-> in
#define AV_CH_TOP_BACK_RIGHT
Describe the class of an AVClass context structure.
Definition: log.h:67
Filter definition.
Definition: avfilter.h:144
struct MYSOFA_EASY * easy
Definition: af_sofalizer.c:45
float im
Definition: fft.c:82
cl_device_type type
const char * name
Filter name.
Definition: avfilter.h:148
float * speaker_elev
Definition: af_sofalizer.c:65
static const AVFilterPad inputs[]
Definition: af_sofalizer.c:885
AVFilterLink ** outputs
array of pointers to output links
Definition: avfilter.h:350
enum MovChannelLayoutTag * layouts
Definition: mov_chan.c:434
float ** ir
Definition: af_headphone.c:157
AVFilterInternal * internal
An opaque struct for libavfilter internal use.
Definition: avfilter.h:378
#define AV_CH_BACK_CENTER
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:232
#define AV_CH_SIDE_RIGHT
static int sofalizer_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
Definition: af_sofalizer.c:277
char * av_strtok(char *s, const char *delim, char **saveptr)
Split the string into several tokens which can be accessed by successive calls to av_strtok()...
Definition: avstring.c:184
#define M_LN10
Definition: mathematics.h:43
float * lir
Definition: af_sofalizer.c:47
FFTContext * ifft[2]
Definition: af_sofalizer.c:98
int
static int parse_channel_name(char **arg, int *rchannel, char *buf)
Definition: af_sofalizer.c:133
sample_rates
FFTSample im
Definition: avfft.h:38
#define FREQUENCY_DOMAIN
Definition: af_sofalizer.c:42
avfilter_execute_func * execute
Definition: internal.h:155
static av_cold int init(AVFilterContext *ctx)
Definition: af_sofalizer.c:782
#define av_free(p)
int len
static void parse_speaker_pos(AVFilterContext *ctx, int64_t in_channel_layout)
Definition: af_sofalizer.c:158
float * temp_src[2]
Definition: af_sofalizer.c:86
A list of supported formats for one end of a filter link.
Definition: formats.h:64
uint64_t layout
#define AV_CH_SURROUND_DIRECT_LEFT
int n_samples
Definition: af_sofalizer.c:46
An instance of a filter.
Definition: avfilter.h:338
#define AV_CH_FRONT_RIGHT
VirtualSpeaker vspkrpos[64]
Definition: af_sofalizer.c:96
FILE * out
Definition: movenc.c:54
#define av_freep(p)
#define OFFSET(x)
Definition: af_sofalizer.c:866
#define av_malloc_array(a, b)
formats
Definition: signature.h:48
#define AV_CH_SIDE_LEFT
internal API functions
FFTComplex * temp_fft[2]
Definition: af_sofalizer.c:87
static av_cold void uninit(AVFilterContext *ctx)
Definition: af_sofalizer.c:840
void av_fft_calc(FFTContext *s, FFTComplex *z)
Do a complex FFT with the parameters defined in av_fft_init().
Definition: avfft.c:43
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:284
int ff_set_common_samplerates(AVFilterContext *ctx, AVFilterFormats *samplerates)
Definition: formats.c:556
int av_frame_copy_props(AVFrame *dst, const AVFrame *src)
Copy only "metadata" fields from src to dst.
Definition: frame.c:652
#define FFMAX3(a, b, c)
Definition: common.h:95
#define AV_CH_BACK_RIGHT
#define AV_CH_STEREO_LEFT
Stereo downmix.