2 * Copyright (c) 2018 Paul B Mahol
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "libavutil/avassert.h"
24 #include "libavutil/avstring.h"
25 #include "libavutil/intreadwrite.h"
26 #include "libavutil/opt.h"
27 #include "libavutil/xga_font_data.h"
32 typedef struct ThreadData {
40 typedef struct BiquadContext {
46 typedef struct IIRChannel {
52 BiquadContext *biquads;
56 typedef struct AudioIIRContext {
58 char *a_str, *b_str, *g_str;
59 double dry_gain, wet_gain;
74 enum AVSampleFormat sample_format;
76 int (*iir_channel)(AVFilterContext *ctx, void *arg, int ch, int nb_jobs);
79 static int query_formats(AVFilterContext *ctx)
81 AudioIIRContext *s = ctx->priv;
82 AVFilterFormats *formats;
83 AVFilterChannelLayouts *layouts;
84 enum AVSampleFormat sample_fmts[] = {
88 static const enum AVPixelFormat pix_fmts[] = {
95 AVFilterLink *videolink = ctx->outputs[1];
97 formats = ff_make_format_list(pix_fmts);
98 if ((ret = ff_formats_ref(formats, &videolink->incfg.formats)) < 0)
102 layouts = ff_all_channel_counts();
104 return AVERROR(ENOMEM);
105 ret = ff_set_common_channel_layouts(ctx, layouts);
109 sample_fmts[0] = s->sample_format;
110 formats = ff_make_format_list(sample_fmts);
112 return AVERROR(ENOMEM);
113 ret = ff_set_common_formats(ctx, formats);
117 formats = ff_all_samplerates();
119 return AVERROR(ENOMEM);
120 return ff_set_common_samplerates(ctx, formats);
123 #define IIR_CH(name, type, min, max, need_clipping) \
124 static int iir_ch_## name(AVFilterContext *ctx, void *arg, int ch, int nb_jobs) \
126 AudioIIRContext *s = ctx->priv; \
127 const double ig = s->dry_gain; \
128 const double og = s->wet_gain; \
129 const double mix = s->mix; \
130 ThreadData *td = arg; \
131 AVFrame *in = td->in, *out = td->out; \
132 const type *src = (const type *)in->extended_data[ch]; \
133 double *oc = (double *)s->iir[ch].cache[0]; \
134 double *ic = (double *)s->iir[ch].cache[1]; \
135 const int nb_a = s->iir[ch].nb_ab[0]; \
136 const int nb_b = s->iir[ch].nb_ab[1]; \
137 const double *a = s->iir[ch].ab[0]; \
138 const double *b = s->iir[ch].ab[1]; \
139 const double g = s->iir[ch].g; \
140 int *clippings = &s->iir[ch].clippings; \
141 type *dst = (type *)out->extended_data[ch]; \
144 for (n = 0; n < in->nb_samples; n++) { \
145 double sample = 0.; \
148 memmove(&ic[1], &ic[0], (nb_b - 1) * sizeof(*ic)); \
149 memmove(&oc[1], &oc[0], (nb_a - 1) * sizeof(*oc)); \
150 ic[0] = src[n] * ig; \
151 for (x = 0; x < nb_b; x++) \
152 sample += b[x] * ic[x]; \
154 for (x = 1; x < nb_a; x++) \
155 sample -= a[x] * oc[x]; \
159 sample = sample * mix + ic[0] * (1. - mix); \
160 if (need_clipping && sample < min) { \
163 } else if (need_clipping && sample > max) { \
174 IIR_CH(s16p, int16_t, INT16_MIN, INT16_MAX, 1)
175 IIR_CH(s32p, int32_t, INT32_MIN, INT32_MAX, 1)
176 IIR_CH(fltp, float, -1., 1., 0)
177 IIR_CH(dblp, double, -1., 1., 0)
179 #define SERIAL_IIR_CH(name, type, min, max, need_clipping) \
180 static int iir_ch_serial_## name(AVFilterContext *ctx, void *arg, \
181 int ch, int nb_jobs) \
183 AudioIIRContext *s = ctx->priv; \
184 const double ig = s->dry_gain; \
185 const double og = s->wet_gain; \
186 const double mix = s->mix; \
187 const double imix = 1. - mix; \
188 ThreadData *td = arg; \
189 AVFrame *in = td->in, *out = td->out; \
190 const type *src = (const type *)in->extended_data[ch]; \
191 type *dst = (type *)out->extended_data[ch]; \
192 IIRChannel *iir = &s->iir[ch]; \
193 const double g = iir->g; \
194 int *clippings = &iir->clippings; \
195 int nb_biquads = (FFMAX(iir->nb_ab[0], iir->nb_ab[1]) + 1) / 2; \
198 for (i = nb_biquads - 1; i >= 0; i--) { \
199 const double a1 = -iir->biquads[i].a[1]; \
200 const double a2 = -iir->biquads[i].a[2]; \
201 const double b0 = iir->biquads[i].b[0]; \
202 const double b1 = iir->biquads[i].b[1]; \
203 const double b2 = iir->biquads[i].b[2]; \
204 double w1 = iir->biquads[i].w1; \
205 double w2 = iir->biquads[i].w2; \
207 for (n = 0; n < in->nb_samples; n++) { \
208 double i0 = ig * (i ? dst[n] : src[n]); \
209 double o0 = i0 * b0 + w1; \
211 w1 = b1 * i0 + w2 + a1 * o0; \
212 w2 = b2 * i0 + a2 * o0; \
215 o0 = o0 * mix + imix * i0; \
216 if (need_clipping && o0 < min) { \
219 } else if (need_clipping && o0 > max) { \
226 iir->biquads[i].w1 = w1; \
227 iir->biquads[i].w2 = w2; \
233 SERIAL_IIR_CH(s16p, int16_t, INT16_MIN, INT16_MAX, 1)
234 SERIAL_IIR_CH(s32p, int32_t, INT32_MIN, INT32_MAX, 1)
235 SERIAL_IIR_CH(fltp, float, -1., 1., 0)
236 SERIAL_IIR_CH(dblp, double, -1., 1., 0)
238 #define PARALLEL_IIR_CH(name, type, min, max, need_clipping) \
239 static int iir_ch_parallel_## name(AVFilterContext *ctx, void *arg, \
240 int ch, int nb_jobs) \
242 AudioIIRContext *s = ctx->priv; \
243 const double ig = s->dry_gain; \
244 const double og = s->wet_gain; \
245 const double mix = s->mix; \
246 const double imix = 1. - mix; \
247 ThreadData *td = arg; \
248 AVFrame *in = td->in, *out = td->out; \
249 const type *src = (const type *)in->extended_data[ch]; \
250 type *dst = (type *)out->extended_data[ch]; \
251 IIRChannel *iir = &s->iir[ch]; \
252 const double g = iir->g; \
253 const double fir = iir->fir; \
254 int *clippings = &iir->clippings; \
255 int nb_biquads = (FFMAX(iir->nb_ab[0], iir->nb_ab[1]) + 1) / 2; \
258 for (i = 0; i < nb_biquads; i++) { \
259 const double a1 = -iir->biquads[i].a[1]; \
260 const double a2 = -iir->biquads[i].a[2]; \
261 const double b1 = iir->biquads[i].b[1]; \
262 const double b2 = iir->biquads[i].b[2]; \
263 double w1 = iir->biquads[i].w1; \
264 double w2 = iir->biquads[i].w2; \
266 for (n = 0; n < in->nb_samples; n++) { \
267 double i0 = ig * src[n]; \
270 w1 = b1 * i0 + w2 + a1 * o0; \
271 w2 = b2 * i0 + a2 * o0; \
275 if (need_clipping && o0 < min) { \
278 } else if (need_clipping && o0 > max) { \
285 iir->biquads[i].w1 = w1; \
286 iir->biquads[i].w2 = w2; \
289 for (n = 0; n < in->nb_samples; n++) { \
290 dst[n] += fir * src[n]; \
291 dst[n] = dst[n] * mix + imix * src[n]; \
297 PARALLEL_IIR_CH(s16p, int16_t, INT16_MIN, INT16_MAX, 1)
298 PARALLEL_IIR_CH(s32p, int32_t, INT32_MIN, INT32_MAX, 1)
299 PARALLEL_IIR_CH(fltp, float, -1., 1., 0)
300 PARALLEL_IIR_CH(dblp, double, -1., 1., 0)
302 #define LATTICE_IIR_CH(name, type, min, max, need_clipping) \
303 static int iir_ch_lattice_## name(AVFilterContext *ctx, void *arg, \
304 int ch, int nb_jobs) \
306 AudioIIRContext *s = ctx->priv; \
307 const double ig = s->dry_gain; \
308 const double og = s->wet_gain; \
309 const double mix = s->mix; \
310 ThreadData *td = arg; \
311 AVFrame *in = td->in, *out = td->out; \
312 const type *src = (const type *)in->extended_data[ch]; \
313 double n0, n1, p0, *x = (double *)s->iir[ch].cache[0]; \
314 const int nb_stages = s->iir[ch].nb_ab[1]; \
315 const double *v = s->iir[ch].ab[0]; \
316 const double *k = s->iir[ch].ab[1]; \
317 const double g = s->iir[ch].g; \
318 int *clippings = &s->iir[ch].clippings; \
319 type *dst = (type *)out->extended_data[ch]; \
322 for (n = 0; n < in->nb_samples; n++) { \
323 const double in = src[n] * ig; \
327 for (int i = nb_stages - 1; i >= 0; i--) { \
328 n0 = n1 - k[i] * x[i]; \
329 p0 = n0 * k[i] + x[i]; \
330 out += p0 * v[i+1]; \
336 memmove(&x[1], &x[0], nb_stages * sizeof(*x)); \
339 out = out * mix + in * (1. - mix); \
340 if (need_clipping && out < min) { \
343 } else if (need_clipping && out > max) { \
354 LATTICE_IIR_CH(s16p, int16_t, INT16_MIN, INT16_MAX, 1)
355 LATTICE_IIR_CH(s32p, int32_t, INT32_MIN, INT32_MAX, 1)
356 LATTICE_IIR_CH(fltp, float, -1., 1., 0)
357 LATTICE_IIR_CH(dblp, double, -1., 1., 0)
359 static void count_coefficients(char *item_str, int *nb_items)
367 for (p = item_str; *p && *p != '|'; p++) {
373 static int read_gains(AVFilterContext *ctx, char *item_str, int nb_items)
375 AudioIIRContext *s = ctx->priv;
376 char *p, *arg, *old_str, *prev_arg = NULL, *saveptr = NULL;
379 p = old_str = av_strdup(item_str);
381 return AVERROR(ENOMEM);
382 for (i = 0; i < nb_items; i++) {
383 if (!(arg = av_strtok(p, "|", &saveptr)))
388 return AVERROR(EINVAL);
392 if (av_sscanf(arg, "%lf", &s->iir[i].g) != 1) {
393 av_log(ctx, AV_LOG_ERROR, "Invalid gains supplied: %s\n", arg);
395 return AVERROR(EINVAL);
406 static int read_tf_coefficients(AVFilterContext *ctx, char *item_str, int nb_items, double *dst)
408 char *p, *arg, *old_str, *saveptr = NULL;
411 p = old_str = av_strdup(item_str);
413 return AVERROR(ENOMEM);
414 for (i = 0; i < nb_items; i++) {
415 if (!(arg = av_strtok(p, " ", &saveptr)))
419 if (av_sscanf(arg, "%lf", &dst[i]) != 1) {
420 av_log(ctx, AV_LOG_ERROR, "Invalid coefficients supplied: %s\n", arg);
422 return AVERROR(EINVAL);
431 static int read_zp_coefficients(AVFilterContext *ctx, char *item_str, int nb_items, double *dst, const char *format)
433 char *p, *arg, *old_str, *saveptr = NULL;
436 p = old_str = av_strdup(item_str);
438 return AVERROR(ENOMEM);
439 for (i = 0; i < nb_items; i++) {
440 if (!(arg = av_strtok(p, " ", &saveptr)))
444 if (av_sscanf(arg, format, &dst[i*2], &dst[i*2+1]) != 2) {
445 av_log(ctx, AV_LOG_ERROR, "Invalid coefficients supplied: %s\n", arg);
447 return AVERROR(EINVAL);
456 static const char *const format[] = { "%lf", "%lf %lfi", "%lf %lfr", "%lf %lfd", "%lf %lfi" };
458 static int read_channels(AVFilterContext *ctx, int channels, uint8_t *item_str, int ab)
460 AudioIIRContext *s = ctx->priv;
461 char *p, *arg, *old_str, *prev_arg = NULL, *saveptr = NULL;
464 p = old_str = av_strdup(item_str);
466 return AVERROR(ENOMEM);
467 for (i = 0; i < channels; i++) {
468 IIRChannel *iir = &s->iir[i];
470 if (!(arg = av_strtok(p, "|", &saveptr)))
475 return AVERROR(EINVAL);
478 count_coefficients(arg, &iir->nb_ab[ab]);
481 iir->cache[ab] = av_calloc(iir->nb_ab[ab] + 1, sizeof(double));
482 iir->ab[ab] = av_calloc(iir->nb_ab[ab] * (!!s->format + 1), sizeof(double));
483 if (!iir->ab[ab] || !iir->cache[ab]) {
485 return AVERROR(ENOMEM);
489 ret = read_zp_coefficients(ctx, arg, iir->nb_ab[ab], iir->ab[ab], format[s->format]);
491 ret = read_tf_coefficients(ctx, arg, iir->nb_ab[ab], iir->ab[ab]);
505 static void cmul(double re, double im, double re2, double im2, double *RE, double *IM)
507 *RE = re * re2 - im * im2;
508 *IM = re * im2 + re2 * im;
511 static int expand(AVFilterContext *ctx, double *pz, int n, double *coefs)
515 for (int i = 1; i <= n; i++) {
516 for (int j = n - i; j < n; j++) {
519 cmul(coefs[2 * (j + 1)], coefs[2 * (j + 1) + 1],
520 pz[2 * (i - 1)], pz[2 * (i - 1) + 1], &re, &im);
523 coefs[2 * j + 1] -= im;
527 for (int i = 0; i < n + 1; i++) {
528 if (fabs(coefs[2 * i + 1]) > FLT_EPSILON) {
529 av_log(ctx, AV_LOG_ERROR, "coefs: %f of z^%d is not real; poles/zeros are not complex conjugates.\n",
530 coefs[2 * i + 1], i);
531 return AVERROR(EINVAL);
538 static void normalize_coeffs(AVFilterContext *ctx, int ch)
540 AudioIIRContext *s = ctx->priv;
541 IIRChannel *iir = &s->iir[ch];
547 for (int i = 0; i < iir->nb_ab[1]; i++) {
548 sum_den += iir->ab[1][i];
551 if (sum_den > 1e-6) {
552 double factor, sum_num = 0.;
554 for (int i = 0; i < iir->nb_ab[0]; i++) {
555 sum_num += iir->ab[0][i];
558 factor = sum_num / sum_den;
560 for (int i = 0; i < iir->nb_ab[1]; i++) {
561 iir->ab[1][i] *= factor;
566 static int convert_zp2tf(AVFilterContext *ctx, int channels)
568 AudioIIRContext *s = ctx->priv;
569 int ch, i, j, ret = 0;
571 for (ch = 0; ch < channels; ch++) {
572 IIRChannel *iir = &s->iir[ch];
575 topc = av_calloc((iir->nb_ab[1] + 1) * 2, sizeof(*topc));
576 botc = av_calloc((iir->nb_ab[0] + 1) * 2, sizeof(*botc));
577 if (!topc || !botc) {
578 ret = AVERROR(ENOMEM);
582 ret = expand(ctx, iir->ab[0], iir->nb_ab[0], botc);
587 ret = expand(ctx, iir->ab[1], iir->nb_ab[1], topc);
592 for (j = 0, i = iir->nb_ab[1]; i >= 0; j++, i--) {
593 iir->ab[1][j] = topc[2 * i];
597 for (j = 0, i = iir->nb_ab[0]; i >= 0; j++, i--) {
598 iir->ab[0][j] = botc[2 * i];
602 normalize_coeffs(ctx, ch);
614 static int decompose_zp2biquads(AVFilterContext *ctx, int channels)
616 AudioIIRContext *s = ctx->priv;
619 for (ch = 0; ch < channels; ch++) {
620 IIRChannel *iir = &s->iir[ch];
621 int nb_biquads = (FFMAX(iir->nb_ab[0], iir->nb_ab[1]) + 1) / 2;
622 int current_biquad = 0;
624 iir->biquads = av_calloc(nb_biquads, sizeof(BiquadContext));
626 return AVERROR(ENOMEM);
628 while (nb_biquads--) {
629 Pair outmost_pole = { -1, -1 };
630 Pair nearest_zero = { -1, -1 };
631 double zeros[4] = { 0 };
632 double poles[4] = { 0 };
635 double min_distance = DBL_MAX;
640 for (i = 0; i < iir->nb_ab[0]; i++) {
643 if (isnan(iir->ab[0][2 * i]) || isnan(iir->ab[0][2 * i + 1]))
645 mag = hypot(iir->ab[0][2 * i], iir->ab[0][2 * i + 1]);
653 for (i = 0; i < iir->nb_ab[0]; i++) {
654 if (isnan(iir->ab[0][2 * i]) || isnan(iir->ab[0][2 * i + 1]))
657 if (iir->ab[0][2 * i ] == iir->ab[0][2 * outmost_pole.a ] &&
658 iir->ab[0][2 * i + 1] == -iir->ab[0][2 * outmost_pole.a + 1]) {
664 av_log(ctx, AV_LOG_VERBOSE, "outmost_pole is %d.%d\n", outmost_pole.a, outmost_pole.b);
666 if (outmost_pole.a < 0 || outmost_pole.b < 0)
667 return AVERROR(EINVAL);
669 for (i = 0; i < iir->nb_ab[1]; i++) {
672 if (isnan(iir->ab[1][2 * i]) || isnan(iir->ab[1][2 * i + 1]))
674 distance = hypot(iir->ab[0][2 * outmost_pole.a ] - iir->ab[1][2 * i ],
675 iir->ab[0][2 * outmost_pole.a + 1] - iir->ab[1][2 * i + 1]);
677 if (distance < min_distance) {
678 min_distance = distance;
683 for (i = 0; i < iir->nb_ab[1]; i++) {
684 if (isnan(iir->ab[1][2 * i]) || isnan(iir->ab[1][2 * i + 1]))
687 if (iir->ab[1][2 * i ] == iir->ab[1][2 * nearest_zero.a ] &&
688 iir->ab[1][2 * i + 1] == -iir->ab[1][2 * nearest_zero.a + 1]) {
694 av_log(ctx, AV_LOG_VERBOSE, "nearest_zero is %d.%d\n", nearest_zero.a, nearest_zero.b);
696 if (nearest_zero.a < 0 || nearest_zero.b < 0)
697 return AVERROR(EINVAL);
699 poles[0] = iir->ab[0][2 * outmost_pole.a ];
700 poles[1] = iir->ab[0][2 * outmost_pole.a + 1];
702 zeros[0] = iir->ab[1][2 * nearest_zero.a ];
703 zeros[1] = iir->ab[1][2 * nearest_zero.a + 1];
705 if (nearest_zero.a == nearest_zero.b && outmost_pole.a == outmost_pole.b) {
712 poles[2] = iir->ab[0][2 * outmost_pole.b ];
713 poles[3] = iir->ab[0][2 * outmost_pole.b + 1];
715 zeros[2] = iir->ab[1][2 * nearest_zero.b ];
716 zeros[3] = iir->ab[1][2 * nearest_zero.b + 1];
719 ret = expand(ctx, zeros, 2, b);
723 ret = expand(ctx, poles, 2, a);
727 iir->ab[0][2 * outmost_pole.a] = iir->ab[0][2 * outmost_pole.a + 1] = NAN;
728 iir->ab[0][2 * outmost_pole.b] = iir->ab[0][2 * outmost_pole.b + 1] = NAN;
729 iir->ab[1][2 * nearest_zero.a] = iir->ab[1][2 * nearest_zero.a + 1] = NAN;
730 iir->ab[1][2 * nearest_zero.b] = iir->ab[1][2 * nearest_zero.b + 1] = NAN;
732 iir->biquads[current_biquad].a[0] = 1.;
733 iir->biquads[current_biquad].a[1] = a[2] / a[4];
734 iir->biquads[current_biquad].a[2] = a[0] / a[4];
735 iir->biquads[current_biquad].b[0] = b[4] / a[4];
736 iir->biquads[current_biquad].b[1] = b[2] / a[4];
737 iir->biquads[current_biquad].b[2] = b[0] / a[4];
740 fabs(iir->biquads[current_biquad].b[0] +
741 iir->biquads[current_biquad].b[1] +
742 iir->biquads[current_biquad].b[2]) > 1e-6) {
743 factor = (iir->biquads[current_biquad].a[0] +
744 iir->biquads[current_biquad].a[1] +
745 iir->biquads[current_biquad].a[2]) /
746 (iir->biquads[current_biquad].b[0] +
747 iir->biquads[current_biquad].b[1] +
748 iir->biquads[current_biquad].b[2]);
750 av_log(ctx, AV_LOG_VERBOSE, "factor=%f\n", factor);
752 iir->biquads[current_biquad].b[0] *= factor;
753 iir->biquads[current_biquad].b[1] *= factor;
754 iir->biquads[current_biquad].b[2] *= factor;
757 iir->biquads[current_biquad].b[0] *= (current_biquad ? 1.0 : iir->g);
758 iir->biquads[current_biquad].b[1] *= (current_biquad ? 1.0 : iir->g);
759 iir->biquads[current_biquad].b[2] *= (current_biquad ? 1.0 : iir->g);
761 av_log(ctx, AV_LOG_VERBOSE, "a=%f %f %f:b=%f %f %f\n",
762 iir->biquads[current_biquad].a[0],
763 iir->biquads[current_biquad].a[1],
764 iir->biquads[current_biquad].a[2],
765 iir->biquads[current_biquad].b[0],
766 iir->biquads[current_biquad].b[1],
767 iir->biquads[current_biquad].b[2]);
776 static void biquad_process(double *x, double *y, int length,
777 double b0, double b1, double b2,
778 double a1, double a2)
780 double w1 = 0., w2 = 0.;
785 for (int n = 0; n < length; n++) {
786 double out, in = x[n];
788 y[n] = out = in * b0 + w1;
789 w1 = b1 * in + w2 + a1 * out;
790 w2 = b2 * in + a2 * out;
794 static void solve(double *matrix, double *vector, int n, double *y, double *x, double *lu)
798 for (int i = 0; i < n; i++) {
799 for (int j = i; j < n; j++) {
801 for (int k = 0; k < i; k++)
802 sum += lu[i * n + k] * lu[k * n + j];
803 lu[i * n + j] = matrix[j * n + i] - sum;
805 for (int j = i + 1; j < n; j++) {
807 for (int k = 0; k < i; k++)
808 sum += lu[j * n + k] * lu[k * n + i];
809 lu[j * n + i] = (1. / lu[i * n + i]) * (matrix[i * n + j] - sum);
813 for (int i = 0; i < n; i++) {
815 for (int k = 0; k < i; k++)
816 sum += lu[i * n + k] * y[k];
817 y[i] = vector[i] - sum;
820 for (int i = n - 1; i >= 0; i--) {
822 for (int k = i + 1; k < n; k++)
823 sum += lu[i * n + k] * x[k];
824 x[i] = (1 / lu[i * n + i]) * (y[i] - sum);
828 static int convert_serial2parallel(AVFilterContext *ctx, int channels)
830 AudioIIRContext *s = ctx->priv;
833 for (int ch = 0; ch < channels; ch++) {
834 IIRChannel *iir = &s->iir[ch];
835 int nb_biquads = (FFMAX(iir->nb_ab[0], iir->nb_ab[1]) + 1) / 2;
836 int length = nb_biquads * 2 + 1;
837 double *impulse = av_calloc(length, sizeof(*impulse));
838 double *y = av_calloc(length, sizeof(*y));
839 double *resp = av_calloc(length, sizeof(*resp));
840 double *M = av_calloc((length - 1) * 2 * nb_biquads, sizeof(*M));
841 double *W = av_calloc((length - 1) * 2 * nb_biquads, sizeof(*W));
843 if (!impulse || !y || !resp || !M) {
849 return AVERROR(ENOMEM);
854 for (int n = 0; n < nb_biquads; n++) {
855 BiquadContext *biquad = &iir->biquads[n];
857 biquad_process(n ? y : impulse, y, length,
858 biquad->b[0], biquad->b[1], biquad->b[2],
859 biquad->a[1], biquad->a[2]);
862 for (int n = 0; n < nb_biquads; n++) {
863 BiquadContext *biquad = &iir->biquads[n];
865 biquad_process(impulse, resp, length - 1,
866 1., 0., 0., biquad->a[1], biquad->a[2]);
868 memcpy(M + n * 2 * (length - 1), resp, sizeof(*resp) * (length - 1));
869 memcpy(M + n * 2 * (length - 1) + length, resp, sizeof(*resp) * (length - 2));
870 memset(resp, 0, length * sizeof(*resp));
873 solve(M, &y[1], length - 1, &impulse[1], resp, W);
877 for (int n = 0; n < nb_biquads; n++) {
878 BiquadContext *biquad = &iir->biquads[n];
881 biquad->b[1] = resp[n * 2 + 0];
882 biquad->b[2] = resp[n * 2 + 1];
898 static void convert_pr2zp(AVFilterContext *ctx, int channels)
900 AudioIIRContext *s = ctx->priv;
903 for (ch = 0; ch < channels; ch++) {
904 IIRChannel *iir = &s->iir[ch];
907 for (n = 0; n < iir->nb_ab[0]; n++) {
908 double r = iir->ab[0][2*n];
909 double angle = iir->ab[0][2*n+1];
911 iir->ab[0][2*n] = r * cos(angle);
912 iir->ab[0][2*n+1] = r * sin(angle);
915 for (n = 0; n < iir->nb_ab[1]; n++) {
916 double r = iir->ab[1][2*n];
917 double angle = iir->ab[1][2*n+1];
919 iir->ab[1][2*n] = r * cos(angle);
920 iir->ab[1][2*n+1] = r * sin(angle);
925 static void convert_sp2zp(AVFilterContext *ctx, int channels)
927 AudioIIRContext *s = ctx->priv;
930 for (ch = 0; ch < channels; ch++) {
931 IIRChannel *iir = &s->iir[ch];
934 for (n = 0; n < iir->nb_ab[0]; n++) {
935 double sr = iir->ab[0][2*n];
936 double si = iir->ab[0][2*n+1];
938 iir->ab[0][2*n] = exp(sr) * cos(si);
939 iir->ab[0][2*n+1] = exp(sr) * sin(si);
942 for (n = 0; n < iir->nb_ab[1]; n++) {
943 double sr = iir->ab[1][2*n];
944 double si = iir->ab[1][2*n+1];
946 iir->ab[1][2*n] = exp(sr) * cos(si);
947 iir->ab[1][2*n+1] = exp(sr) * sin(si);
952 static double fact(double i)
956 return i * fact(i - 1.);
959 static double coef_sf2zf(double *a, int N, int n)
963 for (int i = 0; i <= N; i++) {
966 for (int k = FFMAX(n - N + i, 0); k <= FFMIN(i, n); k++) {
967 acc += ((fact(i) * fact(N - i)) /
968 (fact(k) * fact(i - k) * fact(n - k) * fact(N - i - n + k))) *
969 ((k & 1) ? -1. : 1.);
972 z += a[i] * pow(2., i) * acc;
978 static void convert_sf2tf(AVFilterContext *ctx, int channels)
980 AudioIIRContext *s = ctx->priv;
983 for (ch = 0; ch < channels; ch++) {
984 IIRChannel *iir = &s->iir[ch];
985 double *temp0 = av_calloc(iir->nb_ab[0], sizeof(*temp0));
986 double *temp1 = av_calloc(iir->nb_ab[1], sizeof(*temp1));
988 if (!temp0 || !temp1)
991 memcpy(temp0, iir->ab[0], iir->nb_ab[0] * sizeof(*temp0));
992 memcpy(temp1, iir->ab[1], iir->nb_ab[1] * sizeof(*temp1));
994 for (int n = 0; n < iir->nb_ab[0]; n++)
995 iir->ab[0][n] = coef_sf2zf(temp0, iir->nb_ab[0] - 1, n);
997 for (int n = 0; n < iir->nb_ab[1]; n++)
998 iir->ab[1][n] = coef_sf2zf(temp1, iir->nb_ab[1] - 1, n);
1006 static void convert_pd2zp(AVFilterContext *ctx, int channels)
1008 AudioIIRContext *s = ctx->priv;
1011 for (ch = 0; ch < channels; ch++) {
1012 IIRChannel *iir = &s->iir[ch];
1015 for (n = 0; n < iir->nb_ab[0]; n++) {
1016 double r = iir->ab[0][2*n];
1017 double angle = M_PI*iir->ab[0][2*n+1]/180.;
1019 iir->ab[0][2*n] = r * cos(angle);
1020 iir->ab[0][2*n+1] = r * sin(angle);
1023 for (n = 0; n < iir->nb_ab[1]; n++) {
1024 double r = iir->ab[1][2*n];
1025 double angle = M_PI*iir->ab[1][2*n+1]/180.;
1027 iir->ab[1][2*n] = r * cos(angle);
1028 iir->ab[1][2*n+1] = r * sin(angle);
1033 static void check_stability(AVFilterContext *ctx, int channels)
1035 AudioIIRContext *s = ctx->priv;
1038 for (ch = 0; ch < channels; ch++) {
1039 IIRChannel *iir = &s->iir[ch];
1041 for (int n = 0; n < iir->nb_ab[0]; n++) {
1042 double pr = hypot(iir->ab[0][2*n], iir->ab[0][2*n+1]);
1045 av_log(ctx, AV_LOG_WARNING, "pole %d at channel %d is unstable\n", n, ch);
1052 static void drawtext(AVFrame *pic, int x, int y, const char *txt, uint32_t color)
1054 const uint8_t *font;
1058 font = avpriv_cga_font, font_height = 8;
1060 for (i = 0; txt[i]; i++) {
1063 uint8_t *p = pic->data[0] + y * pic->linesize[0] + (x + i * 8) * 4;
1064 for (char_y = 0; char_y < font_height; char_y++) {
1065 for (mask = 0x80; mask; mask >>= 1) {
1066 if (font[txt[i] * font_height + char_y] & mask)
1070 p += pic->linesize[0] - 8 * 4;
1075 static void draw_line(AVFrame *out, int x0, int y0, int x1, int y1, uint32_t color)
1077 int dx = FFABS(x1-x0);
1078 int dy = FFABS(y1-y0), sy = y0 < y1 ? 1 : -1;
1079 int err = (dx>dy ? dx : -dy) / 2, e2;
1082 AV_WL32(out->data[0] + y0 * out->linesize[0] + x0 * 4, color);
1084 if (x0 == x1 && y0 == y1)
1101 static double distance(double x0, double x1, double y0, double y1)
1103 return hypot(x0 - x1, y0 - y1);
1106 static void get_response(int channel, int format, double w,
1107 const double *b, const double *a,
1108 int nb_b, int nb_a, double *magnitude, double *phase)
1110 double realz, realp;
1111 double imagz, imagp;
1116 realz = 0., realp = 0.;
1117 imagz = 0., imagp = 0.;
1118 for (int x = 0; x < nb_a; x++) {
1119 realz += cos(-x * w) * a[x];
1120 imagz += sin(-x * w) * a[x];
1123 for (int x = 0; x < nb_b; x++) {
1124 realp += cos(-x * w) * b[x];
1125 imagp += sin(-x * w) * b[x];
1128 div = realp * realp + imagp * imagp;
1129 real = (realz * realp + imagz * imagp) / div;
1130 imag = (imagz * realp - imagp * realz) / div;
1132 *magnitude = hypot(real, imag);
1133 *phase = atan2(imag, real);
1135 double p = 1., z = 1.;
1138 for (int x = 0; x < nb_a; x++) {
1139 z *= distance(cos(w), a[2 * x], sin(w), a[2 * x + 1]);
1140 acc += atan2(sin(w) - a[2 * x + 1], cos(w) - a[2 * x]);
1143 for (int x = 0; x < nb_b; x++) {
1144 p *= distance(cos(w), b[2 * x], sin(w), b[2 * x + 1]);
1145 acc -= atan2(sin(w) - b[2 * x + 1], cos(w) - b[2 * x]);
1153 static void draw_response(AVFilterContext *ctx, AVFrame *out, int sample_rate)
1155 AudioIIRContext *s = ctx->priv;
1156 double *mag, *phase, *temp, *delay, min = DBL_MAX, max = -DBL_MAX;
1157 double min_delay = DBL_MAX, max_delay = -DBL_MAX, min_phase, max_phase;
1158 int prev_ymag = -1, prev_yphase = -1, prev_ydelay = -1;
1162 memset(out->data[0], 0, s->h * out->linesize[0]);
1164 phase = av_malloc_array(s->w, sizeof(*phase));
1165 temp = av_malloc_array(s->w, sizeof(*temp));
1166 mag = av_malloc_array(s->w, sizeof(*mag));
1167 delay = av_malloc_array(s->w, sizeof(*delay));
1168 if (!mag || !phase || !delay || !temp)
1171 ch = av_clip(s->ir_channel, 0, s->channels - 1);
1172 for (i = 0; i < s->w; i++) {
1173 const double *b = s->iir[ch].ab[0];
1174 const double *a = s->iir[ch].ab[1];
1175 const int nb_b = s->iir[ch].nb_ab[0];
1176 const int nb_a = s->iir[ch].nb_ab[1];
1177 double w = i * M_PI / (s->w - 1);
1180 get_response(ch, s->format, w, b, a, nb_b, nb_a, &m, &p);
1182 mag[i] = s->iir[ch].g * m;
1184 min = fmin(min, mag[i]);
1185 max = fmax(max, mag[i]);
1189 for (i = 0; i < s->w - 1; i++) {
1190 double d = phase[i] - phase[i + 1];
1191 temp[i + 1] = ceil(fabs(d) / (2. * M_PI)) * 2. * M_PI * ((d > M_PI) - (d < -M_PI));
1194 min_phase = phase[0];
1195 max_phase = phase[0];
1196 for (i = 1; i < s->w; i++) {
1197 temp[i] += temp[i - 1];
1198 phase[i] += temp[i];
1199 min_phase = fmin(min_phase, phase[i]);
1200 max_phase = fmax(max_phase, phase[i]);
1203 for (i = 0; i < s->w - 1; i++) {
1204 double div = s->w / (double)sample_rate;
1206 delay[i + 1] = -(phase[i] - phase[i + 1]) / div;
1207 min_delay = fmin(min_delay, delay[i + 1]);
1208 max_delay = fmax(max_delay, delay[i + 1]);
1210 delay[0] = delay[1];
1212 for (i = 0; i < s->w; i++) {
1213 int ymag = mag[i] / max * (s->h - 1);
1214 int ydelay = (delay[i] - min_delay) / (max_delay - min_delay) * (s->h - 1);
1215 int yphase = (phase[i] - min_phase) / (max_phase - min_phase) * (s->h - 1);
1217 ymag = s->h - 1 - av_clip(ymag, 0, s->h - 1);
1218 yphase = s->h - 1 - av_clip(yphase, 0, s->h - 1);
1219 ydelay = s->h - 1 - av_clip(ydelay, 0, s->h - 1);
1223 if (prev_yphase < 0)
1224 prev_yphase = yphase;
1225 if (prev_ydelay < 0)
1226 prev_ydelay = ydelay;
1228 draw_line(out, i, ymag, FFMAX(i - 1, 0), prev_ymag, 0xFFFF00FF);
1229 draw_line(out, i, yphase, FFMAX(i - 1, 0), prev_yphase, 0xFF00FF00);
1230 draw_line(out, i, ydelay, FFMAX(i - 1, 0), prev_ydelay, 0xFF00FFFF);
1233 prev_yphase = yphase;
1234 prev_ydelay = ydelay;
1237 if (s->w > 400 && s->h > 100) {
1238 drawtext(out, 2, 2, "Max Magnitude:", 0xDDDDDDDD);
1239 snprintf(text, sizeof(text), "%.2f", max);
1240 drawtext(out, 15 * 8 + 2, 2, text, 0xDDDDDDDD);
1242 drawtext(out, 2, 12, "Min Magnitude:", 0xDDDDDDDD);
1243 snprintf(text, sizeof(text), "%.2f", min);
1244 drawtext(out, 15 * 8 + 2, 12, text, 0xDDDDDDDD);
1246 drawtext(out, 2, 22, "Max Phase:", 0xDDDDDDDD);
1247 snprintf(text, sizeof(text), "%.2f", max_phase);
1248 drawtext(out, 15 * 8 + 2, 22, text, 0xDDDDDDDD);
1250 drawtext(out, 2, 32, "Min Phase:", 0xDDDDDDDD);
1251 snprintf(text, sizeof(text), "%.2f", min_phase);
1252 drawtext(out, 15 * 8 + 2, 32, text, 0xDDDDDDDD);
1254 drawtext(out, 2, 42, "Max Delay:", 0xDDDDDDDD);
1255 snprintf(text, sizeof(text), "%.2f", max_delay);
1256 drawtext(out, 11 * 8 + 2, 42, text, 0xDDDDDDDD);
1258 drawtext(out, 2, 52, "Min Delay:", 0xDDDDDDDD);
1259 snprintf(text, sizeof(text), "%.2f", min_delay);
1260 drawtext(out, 11 * 8 + 2, 52, text, 0xDDDDDDDD);
1270 static int config_output(AVFilterLink *outlink)
1272 AVFilterContext *ctx = outlink->src;
1273 AudioIIRContext *s = ctx->priv;
1274 AVFilterLink *inlink = ctx->inputs[0];
1277 s->channels = inlink->channels;
1278 s->iir = av_calloc(s->channels, sizeof(*s->iir));
1280 return AVERROR(ENOMEM);
1282 ret = read_gains(ctx, s->g_str, inlink->channels);
1286 ret = read_channels(ctx, inlink->channels, s->a_str, 0);
1290 ret = read_channels(ctx, inlink->channels, s->b_str, 1);
1294 if (s->format == -1) {
1295 convert_sf2tf(ctx, inlink->channels);
1297 } else if (s->format == 2) {
1298 convert_pr2zp(ctx, inlink->channels);
1299 } else if (s->format == 3) {
1300 convert_pd2zp(ctx, inlink->channels);
1301 } else if (s->format == 4) {
1302 convert_sp2zp(ctx, inlink->channels);
1304 if (s->format > 0) {
1305 check_stability(ctx, inlink->channels);
1308 av_frame_free(&s->video);
1310 s->video = ff_get_video_buffer(ctx->outputs[1], s->w, s->h);
1312 return AVERROR(ENOMEM);
1314 draw_response(ctx, s->video, inlink->sample_rate);
1318 av_log(ctx, AV_LOG_WARNING, "transfer function coefficients format is not recommended for too high number of zeros/poles.\n");
1320 if (s->format > 0 && s->process == 0) {
1321 av_log(ctx, AV_LOG_WARNING, "Direct processsing is not recommended for zp coefficients format.\n");
1323 ret = convert_zp2tf(ctx, inlink->channels);
1326 } else if (s->format == -2 && s->process > 0) {
1327 av_log(ctx, AV_LOG_ERROR, "Only direct processing is implemented for lattice-ladder function.\n");
1328 return AVERROR_PATCHWELCOME;
1329 } else if (s->format <= 0 && s->process == 1) {
1330 av_log(ctx, AV_LOG_ERROR, "Serial processing is not implemented for transfer function.\n");
1331 return AVERROR_PATCHWELCOME;
1332 } else if (s->format <= 0 && s->process == 2) {
1333 av_log(ctx, AV_LOG_ERROR, "Parallel processing is not implemented for transfer function.\n");
1334 return AVERROR_PATCHWELCOME;
1335 } else if (s->format > 0 && s->process == 1) {
1336 ret = decompose_zp2biquads(ctx, inlink->channels);
1339 } else if (s->format > 0 && s->process == 2) {
1340 if (s->precision > 1)
1341 av_log(ctx, AV_LOG_WARNING, "Parallel processing is not recommended for fixed-point precisions.\n");
1342 ret = decompose_zp2biquads(ctx, inlink->channels);
1345 ret = convert_serial2parallel(ctx, inlink->channels);
1350 for (ch = 0; s->format == -2 && ch < inlink->channels; ch++) {
1351 IIRChannel *iir = &s->iir[ch];
1353 if (iir->nb_ab[0] != iir->nb_ab[1] + 1) {
1354 av_log(ctx, AV_LOG_ERROR, "Number of ladder coefficients must be one more than number of reflection coefficients.\n");
1355 return AVERROR(EINVAL);
1359 for (ch = 0; s->format == 0 && ch < inlink->channels; ch++) {
1360 IIRChannel *iir = &s->iir[ch];
1362 for (i = 1; i < iir->nb_ab[0]; i++) {
1363 iir->ab[0][i] /= iir->ab[0][0];
1366 iir->ab[0][0] = 1.0;
1367 for (i = 0; i < iir->nb_ab[1]; i++) {
1368 iir->ab[1][i] *= iir->g;
1371 normalize_coeffs(ctx, ch);
1374 switch (inlink->format) {
1375 case AV_SAMPLE_FMT_DBLP: s->iir_channel = s->process == 2 ? iir_ch_parallel_dblp : s->process == 1 ? iir_ch_serial_dblp : iir_ch_dblp; break;
1376 case AV_SAMPLE_FMT_FLTP: s->iir_channel = s->process == 2 ? iir_ch_parallel_fltp : s->process == 1 ? iir_ch_serial_fltp : iir_ch_fltp; break;
1377 case AV_SAMPLE_FMT_S32P: s->iir_channel = s->process == 2 ? iir_ch_parallel_s32p : s->process == 1 ? iir_ch_serial_s32p : iir_ch_s32p; break;
1378 case AV_SAMPLE_FMT_S16P: s->iir_channel = s->process == 2 ? iir_ch_parallel_s16p : s->process == 1 ? iir_ch_serial_s16p : iir_ch_s16p; break;
1381 if (s->format == -2) {
1382 switch (inlink->format) {
1383 case AV_SAMPLE_FMT_DBLP: s->iir_channel = iir_ch_lattice_dblp; break;
1384 case AV_SAMPLE_FMT_FLTP: s->iir_channel = iir_ch_lattice_fltp; break;
1385 case AV_SAMPLE_FMT_S32P: s->iir_channel = iir_ch_lattice_s32p; break;
1386 case AV_SAMPLE_FMT_S16P: s->iir_channel = iir_ch_lattice_s16p; break;
1393 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
1395 AVFilterContext *ctx = inlink->dst;
1396 AudioIIRContext *s = ctx->priv;
1397 AVFilterLink *outlink = ctx->outputs[0];
1402 if (av_frame_is_writable(in) && s->process != 2) {
1405 out = ff_get_audio_buffer(outlink, in->nb_samples);
1408 return AVERROR(ENOMEM);
1410 av_frame_copy_props(out, in);
1415 ctx->internal->execute(ctx, s->iir_channel, &td, NULL, outlink->channels);
1417 for (ch = 0; ch < outlink->channels; ch++) {
1418 if (s->iir[ch].clippings > 0)
1419 av_log(ctx, AV_LOG_WARNING, "Channel %d clipping %d times. Please reduce gain.\n",
1420 ch, s->iir[ch].clippings);
1421 s->iir[ch].clippings = 0;
1428 AVFilterLink *outlink = ctx->outputs[1];
1429 int64_t old_pts = s->video->pts;
1430 int64_t new_pts = av_rescale_q(out->pts, ctx->inputs[0]->time_base, outlink->time_base);
1432 if (new_pts > old_pts) {
1435 s->video->pts = new_pts;
1436 clone = av_frame_clone(s->video);
1438 return AVERROR(ENOMEM);
1439 ret = ff_filter_frame(outlink, clone);
1445 return ff_filter_frame(outlink, out);
1448 static int config_video(AVFilterLink *outlink)
1450 AVFilterContext *ctx = outlink->src;
1451 AudioIIRContext *s = ctx->priv;
1453 outlink->sample_aspect_ratio = (AVRational){1,1};
1456 outlink->frame_rate = s->rate;
1457 outlink->time_base = av_inv_q(outlink->frame_rate);
1462 static av_cold int init(AVFilterContext *ctx)
1464 AudioIIRContext *s = ctx->priv;
1465 AVFilterPad pad, vpad;
1468 if (!s->a_str || !s->b_str || !s->g_str) {
1469 av_log(ctx, AV_LOG_ERROR, "Valid coefficients are mandatory.\n");
1470 return AVERROR(EINVAL);
1473 switch (s->precision) {
1474 case 0: s->sample_format = AV_SAMPLE_FMT_DBLP; break;
1475 case 1: s->sample_format = AV_SAMPLE_FMT_FLTP; break;
1476 case 2: s->sample_format = AV_SAMPLE_FMT_S32P; break;
1477 case 3: s->sample_format = AV_SAMPLE_FMT_S16P; break;
1478 default: return AVERROR_BUG;
1481 pad = (AVFilterPad){
1483 .type = AVMEDIA_TYPE_AUDIO,
1484 .config_props = config_output,
1487 ret = ff_insert_outpad(ctx, 0, &pad);
1492 vpad = (AVFilterPad){
1493 .name = "filter_response",
1494 .type = AVMEDIA_TYPE_VIDEO,
1495 .config_props = config_video,
1498 ret = ff_insert_outpad(ctx, 1, &vpad);
1506 static av_cold void uninit(AVFilterContext *ctx)
1508 AudioIIRContext *s = ctx->priv;
1512 for (ch = 0; ch < s->channels; ch++) {
1513 IIRChannel *iir = &s->iir[ch];
1514 av_freep(&iir->ab[0]);
1515 av_freep(&iir->ab[1]);
1516 av_freep(&iir->cache[0]);
1517 av_freep(&iir->cache[1]);
1518 av_freep(&iir->biquads);
1523 av_frame_free(&s->video);
1526 static const AVFilterPad inputs[] = {
1529 .type = AVMEDIA_TYPE_AUDIO,
1530 .filter_frame = filter_frame,
1535 #define OFFSET(x) offsetof(AudioIIRContext, x)
1536 #define AF AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
1537 #define VF AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
1539 static const AVOption aiir_options[] = {
1540 { "zeros", "set B/numerator/zeros/reflection coefficients", OFFSET(b_str), AV_OPT_TYPE_STRING, {.str="1+0i 1-0i"}, 0, 0, AF },
1541 { "z", "set B/numerator/zeros/reflection coefficients", OFFSET(b_str), AV_OPT_TYPE_STRING, {.str="1+0i 1-0i"}, 0, 0, AF },
1542 { "poles", "set A/denominator/poles/ladder coefficients", OFFSET(a_str), AV_OPT_TYPE_STRING, {.str="1+0i 1-0i"}, 0, 0, AF },
1543 { "p", "set A/denominator/poles/ladder coefficients", OFFSET(a_str), AV_OPT_TYPE_STRING, {.str="1+0i 1-0i"}, 0, 0, AF },
1544 { "gains", "set channels gains", OFFSET(g_str), AV_OPT_TYPE_STRING, {.str="1|1"}, 0, 0, AF },
1545 { "k", "set channels gains", OFFSET(g_str), AV_OPT_TYPE_STRING, {.str="1|1"}, 0, 0, AF },
1546 { "dry", "set dry gain", OFFSET(dry_gain), AV_OPT_TYPE_DOUBLE, {.dbl=1}, 0, 1, AF },
1547 { "wet", "set wet gain", OFFSET(wet_gain), AV_OPT_TYPE_DOUBLE, {.dbl=1}, 0, 1, AF },
1548 { "format", "set coefficients format", OFFSET(format), AV_OPT_TYPE_INT, {.i64=1}, -2, 4, AF, "format" },
1549 { "f", "set coefficients format", OFFSET(format), AV_OPT_TYPE_INT, {.i64=1}, -2, 4, AF, "format" },
1550 { "ll", "lattice-ladder function", 0, AV_OPT_TYPE_CONST, {.i64=-2}, 0, 0, AF, "format" },
1551 { "sf", "analog transfer function", 0, AV_OPT_TYPE_CONST, {.i64=-1}, 0, 0, AF, "format" },
1552 { "tf", "digital transfer function", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AF, "format" },
1553 { "zp", "Z-plane zeros/poles", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AF, "format" },
1554 { "pr", "Z-plane zeros/poles (polar radians)", 0, AV_OPT_TYPE_CONST, {.i64=2}, 0, 0, AF, "format" },
1555 { "pd", "Z-plane zeros/poles (polar degrees)", 0, AV_OPT_TYPE_CONST, {.i64=3}, 0, 0, AF, "format" },
1556 { "sp", "S-plane zeros/poles", 0, AV_OPT_TYPE_CONST, {.i64=4}, 0, 0, AF, "format" },
1557 { "process", "set kind of processing", OFFSET(process), AV_OPT_TYPE_INT, {.i64=1}, 0, 2, AF, "process" },
1558 { "r", "set kind of processing", OFFSET(process), AV_OPT_TYPE_INT, {.i64=1}, 0, 2, AF, "process" },
1559 { "d", "direct", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AF, "process" },
1560 { "s", "serial", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AF, "process" },
1561 { "p", "parallel", 0, AV_OPT_TYPE_CONST, {.i64=2}, 0, 0, AF, "process" },
1562 { "precision", "set filtering precision", OFFSET(precision),AV_OPT_TYPE_INT, {.i64=0}, 0, 3, AF, "precision" },
1563 { "e", "set precision", OFFSET(precision),AV_OPT_TYPE_INT, {.i64=0}, 0, 3, AF, "precision" },
1564 { "dbl", "double-precision floating-point", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AF, "precision" },
1565 { "flt", "single-precision floating-point", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AF, "precision" },
1566 { "i32", "32-bit integers", 0, AV_OPT_TYPE_CONST, {.i64=2}, 0, 0, AF, "precision" },
1567 { "i16", "16-bit integers", 0, AV_OPT_TYPE_CONST, {.i64=3}, 0, 0, AF, "precision" },
1568 { "normalize", "normalize coefficients", OFFSET(normalize),AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, AF },
1569 { "n", "normalize coefficients", OFFSET(normalize),AV_OPT_TYPE_BOOL, {.i64=1}, 0, 1, AF },
1570 { "mix", "set mix", OFFSET(mix), AV_OPT_TYPE_DOUBLE, {.dbl=1}, 0, 1, AF },
1571 { "response", "show IR frequency response", OFFSET(response), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, VF },
1572 { "channel", "set IR channel to display frequency response", OFFSET(ir_channel), AV_OPT_TYPE_INT, {.i64=0}, 0, 1024, VF },
1573 { "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "hd720"}, 0, 0, VF },
1574 { "rate", "set video rate", OFFSET(rate), AV_OPT_TYPE_VIDEO_RATE, {.str = "25"}, 0, INT32_MAX, VF },
1578 AVFILTER_DEFINE_CLASS(aiir);
1580 AVFilter ff_af_aiir = {
1582 .description = NULL_IF_CONFIG_SMALL("Apply Infinite Impulse Response filter with supplied coefficients."),
1583 .priv_size = sizeof(AudioIIRContext),
1584 .priv_class = &aiir_class,
1587 .query_formats = query_formats,
1589 .flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS |
1590 AVFILTER_FLAG_SLICE_THREADS,