aboutsummaryrefslogtreecommitdiffstats
path: root/libavfilter/af_apsyclip.c
blob: 5afc3930c133050b1525a72ce9b7e344d1704170 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
/*
 * Copyright (c) 2014 - 2021 Jason Jang
 * Copyright (c) 2021 Paul B Mahol
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public License
 * as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public License
 * along with FFmpeg; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "libavutil/mem.h"
#include "libavutil/opt.h"
#include "libavutil/tx.h"
#include "audio.h"
#include "avfilter.h"
#include "filters.h"

typedef struct AudioPsyClipContext {
    const AVClass *class;

    double level_in;
    double level_out;
    double clip_level;
    double adaptive;
    int auto_level;
    int diff_only;
    int iterations;
    char *protections_str;
    double *protections;

    int num_psy_bins;
    int fft_size;
    int overlap;
    int channels;

    int spread_table_rows;
    int *spread_table_index;
    int (*spread_table_range)[2];
    float *window, *inv_window, *spread_table, *margin_curve;

    AVFrame *in;
    AVFrame *in_buffer;
    AVFrame *in_frame;
    AVFrame *out_dist_frame;
    AVFrame *windowed_frame;
    AVFrame *clipping_delta;
    AVFrame *spectrum_buf;
    AVFrame *mask_curve;

    AVTXContext **tx_ctx;
    av_tx_fn tx_fn;
    AVTXContext **itx_ctx;
    av_tx_fn itx_fn;
} AudioPsyClipContext;

#define OFFSET(x) offsetof(AudioPsyClipContext, x)
#define FLAGS AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_RUNTIME_PARAM

static const AVOption apsyclip_options[] = {
    { "level_in",   "set input level",         OFFSET(level_in),   AV_OPT_TYPE_DOUBLE, {.dbl=1},.015625,   64, FLAGS },
    { "level_out",  "set output level",        OFFSET(level_out),  AV_OPT_TYPE_DOUBLE, {.dbl=1},.015625,   64, FLAGS },
    { "clip",       "set clip level",          OFFSET(clip_level), AV_OPT_TYPE_DOUBLE, {.dbl=1},.015625,    1, FLAGS },
    { "diff",       "enable difference",       OFFSET(diff_only),  AV_OPT_TYPE_BOOL,   {.i64=0},      0,    1, FLAGS },
    { "adaptive",   "set adaptive distortion", OFFSET(adaptive),   AV_OPT_TYPE_DOUBLE, {.dbl=0.5},    0,    1, FLAGS },
    { "iterations", "set iterations",          OFFSET(iterations), AV_OPT_TYPE_INT,    {.i64=10},     1,   20, FLAGS },
    { "level",      "set auto level",          OFFSET(auto_level), AV_OPT_TYPE_BOOL,   {.i64=0},      0,    1, FLAGS },
    {NULL}
};

AVFILTER_DEFINE_CLASS(apsyclip);

static void generate_hann_window(float *window, float *inv_window, int size)
{
    for (int i = 0; i < size; i++) {
        float value = 0.5f * (1.f - cosf(2.f * M_PI * i / size));

        window[i] = value;
        // 1/window to calculate unwindowed peak.
        inv_window[i] = value > 0.1f ? 1.f / value : 0.f;
    }
}

static void set_margin_curve(AudioPsyClipContext *s,
                             const int (*points)[2], int num_points, int sample_rate)
{
    int j = 0;

    s->margin_curve[0] = points[0][1];

    for (int i = 0; i < num_points - 1; i++) {
        while (j < s->fft_size / 2 + 1 && j * sample_rate / s->fft_size < points[i + 1][0]) {
            // linearly interpolate between points
            int binHz = j * sample_rate / s->fft_size;
            s->margin_curve[j] = points[i][1] + (binHz - points[i][0]) * (points[i + 1][1] - points[i][1]) / (points[i + 1][0] - points[i][0]);
            j++;
        }
    }
    // handle bins after the last point
    while (j < s->fft_size / 2 + 1) {
        s->margin_curve[j] = points[num_points - 1][1];
        j++;
    }

    // convert margin curve to linear amplitude scale
    for (j = 0; j < s->fft_size / 2 + 1; j++)
        s->margin_curve[j] = powf(10.f, s->margin_curve[j] / 20.f);
}

static void generate_spread_table(AudioPsyClipContext *s)
{
    // Calculate tent-shape function in log-log scale.

    // As an optimization, only consider bins close to "bin"
    // This reduces the number of multiplications needed in calculate_mask_curve
    // The masking contribution at faraway bins is negligeable

    // Another optimization to save memory and speed up the calculation of the
    // spread table is to calculate and store only 2 spread functions per
    // octave, and reuse the same spread function for multiple bins.
    int table_index = 0;
    int bin = 0;
    int increment = 1;

    while (bin < s->num_psy_bins) {
        float sum = 0;
        int base_idx = table_index * s->num_psy_bins;
        int start_bin = bin * 3 / 4;
        int end_bin = FFMIN(s->num_psy_bins, ((bin + 1) * 4 + 2) / 3);
        int next_bin;

        for (int j = start_bin; j < end_bin; j++) {
            // add 0.5 so i=0 doesn't get log(0)
            float rel_idx_log = FFABS(logf((j + 0.5f) / (bin + 0.5f)));
            float value;
            if (j >= bin) {
                // mask up
                value = expf(-rel_idx_log * 40.f);
            } else {
                // mask down
                value = expf(-rel_idx_log * 80.f);
            }
            // the spreading function is centred in the row
            sum += value;
            s->spread_table[base_idx + s->num_psy_bins / 2 + j - bin] = value;
        }
        // now normalize it
        for (int j = start_bin; j < end_bin; j++) {
            s->spread_table[base_idx + s->num_psy_bins / 2 + j - bin] /= sum;
        }

        s->spread_table_range[table_index][0] = start_bin - bin;
        s->spread_table_range[table_index][1] = end_bin - bin;

        if (bin <= 1) {
            next_bin = bin + 1;
        } else {
            if ((bin & (bin - 1)) == 0) {
                // power of 2
                increment = bin / 2;
            }

            next_bin = bin + increment;
        }

        // set bins between "bin" and "next_bin" to use this table_index
        for (int i = bin; i < next_bin; i++)
            s->spread_table_index[i] = table_index;

        bin = next_bin;
        table_index++;
    }
}

static int config_input(AVFilterLink *inlink)
{
    AVFilterContext *ctx = inlink->dst;
    AudioPsyClipContext *s = ctx->priv;
    static const int points[][2] = { {0,14}, {125,14}, {250,16}, {500,18}, {1000,20}, {2000,20}, {4000,20}, {8000,17}, {16000,14}, {20000,-10} };
    static const int num_points = 10;
    float scale = 1.f;
    int ret;

    s->fft_size = inlink->sample_rate > 100000 ? 1024 : inlink->sample_rate > 50000 ? 512 : 256;
    s->overlap = s->fft_size / 4;

    // The psy masking calculation is O(n^2),
    // so skip it for frequencies not covered by base sampling rantes (i.e. 44k)
    if (inlink->sample_rate <= 50000) {
        s->num_psy_bins = s->fft_size / 2;
    } else if (inlink->sample_rate <= 100000) {
        s->num_psy_bins = s->fft_size / 4;
    } else {
        s->num_psy_bins = s->fft_size / 8;
    }

    s->window = av_calloc(s->fft_size, sizeof(*s->window));
    s->inv_window = av_calloc(s->fft_size, sizeof(*s->inv_window));
    if (!s->window || !s->inv_window)
        return AVERROR(ENOMEM);

    s->in_buffer      = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->in_frame       = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->out_dist_frame = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->windowed_frame = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->clipping_delta = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->spectrum_buf   = ff_get_audio_buffer(inlink, s->fft_size * 2);
    s->mask_curve     = ff_get_audio_buffer(inlink, s->fft_size / 2 + 1);
    if (!s->in_buffer || !s->in_frame ||
        !s->out_dist_frame || !s->windowed_frame ||
        !s->clipping_delta || !s->spectrum_buf || !s->mask_curve)
        return AVERROR(ENOMEM);

    generate_hann_window(s->window, s->inv_window, s->fft_size);

    s->margin_curve = av_calloc(s->fft_size / 2 + 1, sizeof(*s->margin_curve));
    if (!s->margin_curve)
        return AVERROR(ENOMEM);

    s->spread_table_rows = av_log2(s->num_psy_bins) * 2;
    s->spread_table = av_calloc(s->spread_table_rows * s->num_psy_bins, sizeof(*s->spread_table));
    if (!s->spread_table)
        return AVERROR(ENOMEM);

    s->spread_table_range = av_calloc(s->spread_table_rows * 2, sizeof(*s->spread_table_range));
    if (!s->spread_table_range)
        return AVERROR(ENOMEM);

    s->spread_table_index = av_calloc(s->num_psy_bins, sizeof(*s->spread_table_index));
    if (!s->spread_table_index)
        return AVERROR(ENOMEM);

    set_margin_curve(s, points, num_points, inlink->sample_rate);

    generate_spread_table(s);

    s->channels = inlink->ch_layout.nb_channels;

    s->tx_ctx = av_calloc(s->channels, sizeof(*s->tx_ctx));
    s->itx_ctx = av_calloc(s->channels, sizeof(*s->itx_ctx));
    if (!s->tx_ctx || !s->itx_ctx)
        return AVERROR(ENOMEM);

    for (int ch = 0; ch < s->channels; ch++) {
        ret = av_tx_init(&s->tx_ctx[ch], &s->tx_fn, AV_TX_FLOAT_FFT, 0, s->fft_size, &scale, 0);
        if (ret < 0)
            return ret;

        ret = av_tx_init(&s->itx_ctx[ch], &s->itx_fn, AV_TX_FLOAT_FFT, 1, s->fft_size, &scale, 0);
        if (ret < 0)
            return ret;
    }

    return 0;
}

static void apply_window(AudioPsyClipContext *s,
                         const float *in_frame, float *out_frame, const int add_to_out_frame)
{
    const float *window = s->window;

    for (int i = 0; i < s->fft_size; i++) {
        if (add_to_out_frame) {
            out_frame[i] += in_frame[i] * window[i];
        } else {
            out_frame[i] = in_frame[i] * window[i];
        }
    }
}

static void calculate_mask_curve(AudioPsyClipContext *s,
                                 const float *spectrum, float *mask_curve)
{
    for (int i = 0; i < s->fft_size / 2 + 1; i++)
        mask_curve[i] = 0;

    for (int i = 0; i < s->num_psy_bins; i++) {
        int base_idx, start_bin, end_bin, table_idx;
        float magnitude;
        int range[2];

        if (i == 0) {
            magnitude = FFABS(spectrum[0]);
        } else if (i == s->fft_size / 2) {
            magnitude = FFABS(spectrum[s->fft_size]);
        } else {
            // Because the input signal is real, the + and - frequencies are redundant.
            // Multiply the magnitude by 2 to simulate adding up the + and - frequencies.
            magnitude = hypotf(spectrum[2 * i], spectrum[2 * i + 1]) * 2;
        }

        table_idx = s->spread_table_index[i];
        range[0] = s->spread_table_range[table_idx][0];
        range[1] = s->spread_table_range[table_idx][1];
        base_idx = table_idx * s->num_psy_bins;
        start_bin = FFMAX(0, i + range[0]);
        end_bin = FFMIN(s->num_psy_bins, i + range[1]);

        for (int j = start_bin; j < end_bin; j++)
            mask_curve[j] += s->spread_table[base_idx + s->num_psy_bins / 2 + j - i] * magnitude;
    }

    // for ultrasonic frequencies, skip the O(n^2) spread calculation and just copy the magnitude
    for (int i = s->num_psy_bins; i < s->fft_size / 2 + 1; i++) {
        float magnitude;
        if (i == s->fft_size / 2) {
            magnitude = FFABS(spectrum[s->fft_size]);
        } else {
            // Because the input signal is real, the + and - frequencies are redundant.
            // Multiply the magnitude by 2 to simulate adding up the + and - frequencies.
            magnitude = hypotf(spectrum[2 * i], spectrum[2 * i + 1]) * 2;
        }

        mask_curve[i] = magnitude;
    }

    for (int i = 0; i < s->fft_size / 2 + 1; i++)
        mask_curve[i] = mask_curve[i] / s->margin_curve[i];
}

static void clip_to_window(AudioPsyClipContext *s,
                           const float *windowed_frame, float *clipping_delta, float delta_boost)
{
    const float *window = s->window;

    for (int i = 0; i < s->fft_size; i++) {
        const float limit = s->clip_level * window[i];
        const float effective_value = windowed_frame[i] + clipping_delta[i];

        if (effective_value > limit) {
            clipping_delta[i] += (limit - effective_value) * delta_boost;
        } else if (effective_value < -limit) {
            clipping_delta[i] += (-limit - effective_value) * delta_boost;
        }
    }
}

static void limit_clip_spectrum(AudioPsyClipContext *s,
                                float *clip_spectrum, const float *mask_curve)
{
    // bin 0
    float relative_distortion_level = FFABS(clip_spectrum[0]) / mask_curve[0];

    if (relative_distortion_level > 1.f)
        clip_spectrum[0] /= relative_distortion_level;

    // bin 1..N/2-1
    for (int i = 1; i < s->fft_size / 2; i++) {
        float real = clip_spectrum[i * 2];
        float imag = clip_spectrum[i * 2 + 1];
        // Because the input signal is real, the + and - frequencies are redundant.
        // Multiply the magnitude by 2 to simulate adding up the + and - frequencies.
        relative_distortion_level = hypotf(real, imag) * 2 / mask_curve[i];
        if (relative_distortion_level > 1.0) {
            clip_spectrum[i * 2] /= relative_distortion_level;
            clip_spectrum[i * 2 + 1] /= relative_distortion_level;
            clip_spectrum[s->fft_size * 2 - i * 2] /= relative_distortion_level;
            clip_spectrum[s->fft_size * 2 - i * 2 + 1] /= relative_distortion_level;
        }
    }
    // bin N/2
    relative_distortion_level = FFABS(clip_spectrum[s->fft_size]) / mask_curve[s->fft_size / 2];
    if (relative_distortion_level > 1.f)
        clip_spectrum[s->fft_size] /= relative_distortion_level;
}

static void r2c(float *buffer, int size)
{
    for (int i = size - 1; i >= 0; i--)
        buffer[2 * i] = buffer[i];

    for (int i = size - 1; i >= 0; i--)
        buffer[2 * i + 1] = 0.f;
}

static void c2r(float *buffer, int size)
{
    for (int i = 0; i < size; i++)
        buffer[i] = buffer[2 * i];

    for (int i = 0; i < size; i++)
        buffer[i + size] = 0.f;
}

static void feed(AVFilterContext *ctx, int ch,
                 const float *in_samples, float *out_samples, int diff_only,
                 float *in_frame, float *out_dist_frame,
                 float *windowed_frame, float *clipping_delta,
                 float *spectrum_buf, float *mask_curve)
{
    AudioPsyClipContext *s = ctx->priv;
    const float clip_level_inv = 1.f / s->clip_level;
    const float level_out = s->level_out;
    float orig_peak = 0;
    float peak;

    // shift in/out buffers
    for (int i = 0; i < s->fft_size - s->overlap; i++) {
        in_frame[i] = in_frame[i + s->overlap];
        out_dist_frame[i] = out_dist_frame[i + s->overlap];
    }

    for (int i = 0; i < s->overlap; i++) {
        in_frame[i + s->fft_size - s->overlap] = in_samples[i];
        out_dist_frame[i + s->fft_size - s->overlap] = 0.f;
    }

    apply_window(s, in_frame, windowed_frame, 0);
    r2c(windowed_frame, s->fft_size);
    s->tx_fn(s->tx_ctx[ch], spectrum_buf, windowed_frame, sizeof(AVComplexFloat));
    c2r(windowed_frame, s->fft_size);
    calculate_mask_curve(s, spectrum_buf, mask_curve);

    // It would be easier to calculate the peak from the unwindowed input.
    // This is just for consistency with the clipped peak calculateion
    // because the inv_window zeros out samples on the edge of the window.
    for (int i = 0; i < s->fft_size; i++)
        orig_peak = FFMAX(orig_peak, FFABS(windowed_frame[i] * s->inv_window[i]));
    orig_peak *= clip_level_inv;
    peak = orig_peak;

    // clear clipping_delta
    for (int i = 0; i < s->fft_size * 2; i++)
        clipping_delta[i] = 0.f;

    // repeat clipping-filtering process a few times to control both the peaks and the spectrum
    for (int i = 0; i < s->iterations; i++) {
        float mask_curve_shift = 1.122f; // 1.122 is 1dB
        // The last 1/3 of rounds have boosted delta to help reach the peak target faster
        float delta_boost = 1.f;
        if (i >= s->iterations - s->iterations / 3) {
            // boosting the delta when largs peaks are still present is dangerous
            if (peak < 2.f)
                delta_boost = 2.f;
        }

        clip_to_window(s, windowed_frame, clipping_delta, delta_boost);

        r2c(clipping_delta, s->fft_size);
        s->tx_fn(s->tx_ctx[ch], spectrum_buf, clipping_delta, sizeof(AVComplexFloat));

        limit_clip_spectrum(s, spectrum_buf, mask_curve);

        s->itx_fn(s->itx_ctx[ch], clipping_delta, spectrum_buf, sizeof(AVComplexFloat));
        c2r(clipping_delta, s->fft_size);

        for (int i = 0; i < s->fft_size; i++)
            clipping_delta[i] /= s->fft_size;

        peak = 0;
        for (int i = 0; i < s->fft_size; i++)
            peak = FFMAX(peak, FFABS((windowed_frame[i] + clipping_delta[i]) * s->inv_window[i]));
        peak *= clip_level_inv;

        // Automatically adjust mask_curve as necessary to reach peak target
        if (orig_peak > 1.f && peak > 1.f) {
            float diff_achieved = orig_peak - peak;
            if (i + 1 < s->iterations - s->iterations / 3 && diff_achieved > 0) {
                float diff_needed = orig_peak - 1.f;
                float diff_ratio = diff_needed / diff_achieved;
                // If a good amount of peak reduction was already achieved,
                // don't shift the mask_curve by the full peak value
                // On the other hand, if only a little peak reduction was achieved,
                // don't shift the mask_curve by the enormous diff_ratio.
                diff_ratio = FFMIN(diff_ratio, peak);
                mask_curve_shift = FFMAX(mask_curve_shift, diff_ratio);
            } else {
                // If the peak got higher than the input or we are in the last 1/3 rounds,
                // go back to the heavy-handed peak heuristic.
                mask_curve_shift = FFMAX(mask_curve_shift, peak);
            }
        }

        mask_curve_shift = 1.f + (mask_curve_shift - 1.f) * s->adaptive;

        // Be less strict in the next iteration.
        // This helps with peak control.
        for (int i = 0; i < s->fft_size / 2 + 1; i++)
            mask_curve[i] *= mask_curve_shift;
    }

    // do overlap & add
    apply_window(s, clipping_delta, out_dist_frame, 1);

    for (int i = 0; i < s->overlap; i++) {
        // 4 times overlap with squared hanning window results in 1.5 time increase in amplitude
        if (!ctx->is_disabled) {
            out_samples[i] = out_dist_frame[i] / 1.5f;
            if (!diff_only)
                out_samples[i] += in_frame[i];
            if (s->auto_level)
                out_samples[i] *= clip_level_inv;
            out_samples[i] *= level_out;
        } else {
            out_samples[i] = in_frame[i];
        }
    }
}

static int psy_channel(AVFilterContext *ctx, AVFrame *in, AVFrame *out, int ch)
{
    AudioPsyClipContext *s = ctx->priv;
    const float *src = (const float *)in->extended_data[ch];
    float *in_buffer = (float *)s->in_buffer->extended_data[ch];
    float *dst = (float *)out->extended_data[ch];

    for (int n = 0; n < s->overlap; n++)
        in_buffer[n] = src[n] * s->level_in;

    feed(ctx, ch, in_buffer, dst, s->diff_only,
         (float *)(s->in_frame->extended_data[ch]),
         (float *)(s->out_dist_frame->extended_data[ch]),
         (float *)(s->windowed_frame->extended_data[ch]),
         (float *)(s->clipping_delta->extended_data[ch]),
         (float *)(s->spectrum_buf->extended_data[ch]),
         (float *)(s->mask_curve->extended_data[ch]));

    return 0;
}

static int psy_channels(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
    AudioPsyClipContext *s = ctx->priv;
    AVFrame *out = arg;
    const int start = (out->ch_layout.nb_channels * jobnr) / nb_jobs;
    const int end = (out->ch_layout.nb_channels * (jobnr+1)) / nb_jobs;

    for (int ch = start; ch < end; ch++)
        psy_channel(ctx, s->in, out, ch);

    return 0;
}

static int filter_frame(AVFilterLink *inlink, AVFrame *in)
{
    AVFilterContext *ctx = inlink->dst;
    AVFilterLink *outlink = ctx->outputs[0];
    AudioPsyClipContext *s = ctx->priv;
    AVFrame *out;
    int ret;

    out = ff_get_audio_buffer(outlink, s->overlap);
    if (!out) {
        ret = AVERROR(ENOMEM);
        goto fail;
    }

    s->in = in;
    av_frame_copy_props(out, in);
    ff_filter_execute(ctx, psy_channels, out, NULL,
                      FFMIN(outlink->ch_layout.nb_channels, ff_filter_get_nb_threads(ctx)));

    out->pts = in->pts;
    out->nb_samples = in->nb_samples;
    ret = ff_filter_frame(outlink, out);
fail:
    av_frame_free(&in);
    s->in = NULL;
    return ret < 0 ? ret : 0;
}

static int activate(AVFilterContext *ctx)
{
    AVFilterLink *inlink = ctx->inputs[0];
    AVFilterLink *outlink = ctx->outputs[0];
    AudioPsyClipContext *s = ctx->priv;
    AVFrame *in = NULL;
    int ret = 0, status;
    int64_t pts;

    FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);

    ret = ff_inlink_consume_samples(inlink, s->overlap, s->overlap, &in);
    if (ret < 0)
        return ret;

    if (ret > 0) {
        return filter_frame(inlink, in);
    } else if (ff_inlink_acknowledge_status(inlink, &status, &pts)) {
        ff_outlink_set_status(outlink, status, pts);
        return 0;
    } else {
        if (ff_inlink_queued_samples(inlink) >= s->overlap) {
            ff_filter_set_ready(ctx, 10);
        } else if (ff_outlink_frame_wanted(outlink)) {
            ff_inlink_request_frame(inlink);
        }
        return 0;
    }
}

static av_cold void uninit(AVFilterContext *ctx)
{
    AudioPsyClipContext *s = ctx->priv;

    av_freep(&s->window);
    av_freep(&s->inv_window);
    av_freep(&s->spread_table);
    av_freep(&s->spread_table_range);
    av_freep(&s->spread_table_index);
    av_freep(&s->margin_curve);

    av_frame_free(&s->in_buffer);
    av_frame_free(&s->in_frame);
    av_frame_free(&s->out_dist_frame);
    av_frame_free(&s->windowed_frame);
    av_frame_free(&s->clipping_delta);
    av_frame_free(&s->spectrum_buf);
    av_frame_free(&s->mask_curve);

    for (int ch = 0; ch < s->channels; ch++) {
        if (s->tx_ctx)
            av_tx_uninit(&s->tx_ctx[ch]);
        if (s->itx_ctx)
            av_tx_uninit(&s->itx_ctx[ch]);
    }

    av_freep(&s->tx_ctx);
    av_freep(&s->itx_ctx);
}

static const AVFilterPad inputs[] = {
    {
        .name         = "default",
        .type         = AVMEDIA_TYPE_AUDIO,
        .config_props = config_input,
    },
};

const AVFilter ff_af_apsyclip = {
    .name            = "apsyclip",
    .description     = NULL_IF_CONFIG_SMALL("Audio Psychoacoustic Clipper."),
    .priv_size       = sizeof(AudioPsyClipContext),
    .priv_class      = &apsyclip_class,
    .uninit          = uninit,
    FILTER_INPUTS(inputs),
    FILTER_OUTPUTS(ff_audio_default_filterpad),
    FILTER_SINGLE_SAMPLEFMT(AV_SAMPLE_FMT_FLTP),
    .flags           = AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL |
                       AVFILTER_FLAG_SLICE_THREADS,
    .activate        = activate,
    .process_command = ff_filter_process_command,
};