Merge remote-tracking branch 'qatar/master'

* qatar/master:
  lavfi: add compand audio filter

Conflicts:
	Changelog
	doc/filters.texi
	libavfilter/Makefile
	libavfilter/af_compand.c
	libavfilter/allfilters.c
	libavfilter/version.h

The filter is added as new one so as to ease clean merging of its changes
in debug-able steps
See: 6b68e2a43b3407522080be50a2a19cff2f9715ef
Merged-by: Michael Niedermayer <michaelni@gmx.at>

5 files changed, 627 insertions, 41 deletions

diff --git a/doc/filters.texi b/doc/filters.texi
index f69ce501b7..8e465ce7e0 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -1267,79 +1267,76 @@ side_right.wav
 @end example
 
 @section compand
-
 Compress or expand audio dynamic range.
 
 A description of the accepted options follows.
 
 @table @option
+
 @item attacks
 @item decays
-Set list of times in seconds for each channel over which the instantaneous
-level of the input signal is averaged to determine its volume.
-@option{attacks} refers to increase of volume and @option{decays} refers
-to decrease of volume.
-For most situations, the attack time (response to the audio getting louder)
-should be shorter than the decay time because the human ear is more sensitive
-to sudden loud audio than sudden soft audio.
-Typical value for attack is @code{0.3} seconds and for decay @code{0.8}
-seconds.
+Set list of times in seconds for each channel over which the instantaneous level
+of the input signal is averaged to determine its volume. @var{attacks} refers to
+increase of volume and @var{decays} refers to decrease of volume. For most
+situations, the attack time (response to the audio getting louder) should be
+shorter than the decay time because the human ear is more sensitive to sudden
+loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
+a typical value for decay is 0.8 seconds.
 
 @item points
-Set list of points for transfer function, specified in dB relative to maximum
-possible signal amplitude.
-Each key points list need to be defined using the following syntax:
-@code{x0/y0 x1/y1 x2/y2 ...}.
+Set list of points for the transfer function, specified in dB relative to the
+maximum possible signal amplitude. Each key points list must be defined using
+the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
+@code{x0/y0 x1/y1 x2/y2 ....}
 
-The input values must be in strictly increasing order but the transfer
-function does not have to be monotonically rising.
-The point @code{0/0} is assumed but may be overridden (by @code{0/out-dBn}).
-Typical values for the transfer function are @code{-70/-70 -60/-20}.
+The input values must be in strictly increasing order but the transfer function
+does not have to be monotonically rising. The point @code{0/0} is assumed but
+may be overridden (by @code{0/out-dBn}). Typical values for the transfer
+function are @code{-70/-70|-60/-20}.
 
 @item soft-knee
-Set amount for which the points at where adjacent line segments on the
-transfer function meet will be rounded. Defaults is @code{0.01}.
+Set the curve radius in dB for all joints. Defaults to 0.01.
 
 @item gain
-Set additional gain in dB to be applied at all points on the transfer function
-and allows easy adjustment of the overall gain.
-Default is @code{0}.
+Set additional gain in dB to be applied at all points on the transfer function.
+This allows easy adjustment of the overall gain. Defaults to 0.
 
 @item volume
 Set initial volume in dB to be assumed for each channel when filtering starts.
-This permits the user to supply a nominal level initially, so that,
-for example, a very large gain is not applied to initial signal levels before
-the companding has begun to operate. A typical value for audio which is
-initially quiet is -90 dB. Default is @code{0}.
+This permits the user to supply a nominal level initially, so that, for
+example, a very large gain is not applied to initial signal levels before the
+companding has begun to operate. A typical value for audio which is initially
+quiet is -90 dB. Defaults to 0.
 
 @item delay
-Set delay in seconds. Default is @code{0}. The input audio
-is analysed immediately, but audio is delayed before being fed to the
-volume adjuster. Specifying a delay approximately equal to the attack/decay
-times allows the filter to effectively operate in predictive rather than
-reactive mode.
+Set delay in seconds. The input audio is analyzed immediately, but audio is
+delayed before being fed to the volume adjuster. Specifying a delay
+approximately equal to the attack/decay times allows the filter to effectively
+operate in predictive rather than reactive mode. Defaults to 0.
+
 @end table
 
 @subsection Examples
+
 @itemize
 @item
-Make music with both quiet and loud passages suitable for listening
-in a noisy environment:
+Make music with both quiet and loud passages suitable for listening in a noisy
+environment:
 @example
-compand=.3 .3:1 1:-90/-60 -60/-40 -40/-30 -20/-20:6:0:-90:0.2
+compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
 @end example
 
 @item
-Noise-gate for when the noise is at a lower level than the signal:
+Noise gate for when the noise is at a lower level than the signal:
 @example
-compand=.1 .1:.2 .2:-900/-900 -50.1/-900 -50/-50:.01:0:-90:.1
+compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
 @end example
 
 @item
-Here is another noise-gate, this time for when the noise is at a higher level
+Here is another noise gate, this time for when the noise is at a higher level
 than the signal (making it, in some ways, similar to squelch):
 @example
-compand=.1 .1:.1 .1:-45.1/-45.1 -45/-900 0/-900:.01:45:-90:.1
+compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
 @end example
 @end itemize
 
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index c36b67ec5a..b3bc27c0d0 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -88,6 +88,7 @@ OBJS-$(CONFIG_BIQUAD_FILTER)                 += af_biquads.o
 OBJS-$(CONFIG_CHANNELMAP_FILTER)             += af_channelmap.o
 OBJS-$(CONFIG_CHANNELSPLIT_FILTER)           += af_channelsplit.o
 OBJS-$(CONFIG_COMPAND_FILTER)                += af_compand.o
+OBJS-$(CONFIG_COMPAND_FORK_FILTER)           += af_compand_fork.o
 OBJS-$(CONFIG_EARWAX_FILTER)                 += af_earwax.o
 OBJS-$(CONFIG_EBUR128_FILTER)                += f_ebur128.o
 OBJS-$(CONFIG_EQUALIZER_FILTER)              += af_biquads.o
diff --git a/libavfilter/af_compand_fork.c b/libavfilter/af_compand_fork.c
new file mode 100644
index 0000000000..dceebff2fe
--- /dev/null
+++ b/libavfilter/af_compand_fork.c
@@ -0,0 +1,587 @@
+/*
+ * Copyright (c) 1999 Chris Bagwell
+ * Copyright (c) 1999 Nick Bailey
+ * Copyright (c) 2007 Rob Sykes <robs@users.sourceforge.net>
+ * Copyright (c) 2013 Paul B Mahol
+ * Copyright (c) 2014 Andrew Kelley
+ *
+ * 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
+ */
+
+/**
+ * @file
+ * audio compand filter
+ */
+
+#include <string.h>
+
+#include "libavutil/channel_layout.h"
+#include "libavutil/common.h"
+#include "libavutil/mathematics.h"
+#include "libavutil/mem.h"
+#include "libavutil/opt.h"
+#include "audio.h"
+#include "avfilter.h"
+#include "formats.h"
+#include "internal.h"
+
+typedef struct ChanParam {
+    float attack;
+    float decay;
+    float volume;
+} ChanParam;
+
+typedef struct CompandSegment {
+    float x, y;
+    float a, b;
+} CompandSegment;
+
+typedef struct CompandContext {
+    const AVClass *class;
+    int nb_channels;
+    int nb_segments;
+    char *attacks, *decays, *points;
+    CompandSegment *segments;
+    ChanParam *channels;
+    float in_min_lin;
+    float out_min_lin;
+    double curve_dB;
+    double gain_dB;
+    double initial_volume;
+    double delay;
+    AVFrame *delay_frame;
+    int delay_samples;
+    int delay_count;
+    int delay_index;
+    int64_t pts;
+
+    int (*compand)(AVFilterContext *ctx, AVFrame *frame);
+} CompandContext;
+
+#define OFFSET(x) offsetof(CompandContext, x)
+#define A AV_OPT_FLAG_AUDIO_PARAM
+
+static const AVOption compand_options[] = {
+    { "attacks", "set time over which increase of volume is determined", OFFSET(attacks), AV_OPT_TYPE_STRING, { .str = "0.3" }, 0, 0, A },
+    { "decays", "set time over which decrease of volume is determined", OFFSET(decays), AV_OPT_TYPE_STRING, { .str = "0.8" }, 0, 0, A },
+    { "points", "set points of transfer function", OFFSET(points), AV_OPT_TYPE_STRING, { .str = "-70/-70|-60/-20" }, 0, 0, A },
+    { "soft-knee", "set soft-knee", OFFSET(curve_dB), AV_OPT_TYPE_DOUBLE, { .dbl = 0.01 }, 0.01, 900, A },
+    { "gain", "set output gain", OFFSET(gain_dB), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -900, 900, A },
+    { "volume", "set initial volume", OFFSET(initial_volume), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -900, 0, A },
+    { "delay", "set delay for samples before sending them to volume adjuster", OFFSET(delay), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, 0, 20, A },
+    { NULL }
+};
+
+static const AVClass compand_class = {
+    .class_name = "compand filter",
+    .item_name  = av_default_item_name,
+    .option     = compand_options,
+    .version    = LIBAVUTIL_VERSION_INT,
+};
+
+static av_cold int init(AVFilterContext *ctx)
+{
+    CompandContext *s = ctx->priv;
+    s->pts            = AV_NOPTS_VALUE;
+    return 0;
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+    CompandContext *s = ctx->priv;
+
+    av_freep(&s->channels);
+    av_freep(&s->segments);
+    av_frame_free(&s->delay_frame);
+}
+
+static int query_formats(AVFilterContext *ctx)
+{
+    AVFilterChannelLayouts *layouts;
+    AVFilterFormats *formats;
+    static const enum AVSampleFormat sample_fmts[] = {
+        AV_SAMPLE_FMT_FLTP,
+        AV_SAMPLE_FMT_NONE
+    };
+
+    layouts = ff_all_channel_layouts();
+    if (!layouts)
+        return AVERROR(ENOMEM);
+    ff_set_common_channel_layouts(ctx, layouts);
+
+    formats = ff_make_format_list(sample_fmts);
+    if (!formats)
+        return AVERROR(ENOMEM);
+    ff_set_common_formats(ctx, formats);
+
+    formats = ff_all_samplerates();
+    if (!formats)
+        return AVERROR(ENOMEM);
+    ff_set_common_samplerates(ctx, formats);
+
+    return 0;
+}
+
+static void count_items(char *item_str, int *nb_items)
+{
+    char *p;
+
+    *nb_items = 1;
+    for (p = item_str; *p; p++) {
+        if (*p == '|')
+            (*nb_items)++;
+    }
+}
+
+static void update_volume(ChanParam *cp, float in)
+{
+    float delta = in - cp->volume;
+
+    if (delta > 0.0)
+        cp->volume += delta * cp->attack;
+    else
+        cp->volume += delta * cp->decay;
+}
+
+static float get_volume(CompandContext *s, float in_lin)
+{
+    CompandSegment *cs;
+    float in_log, out_log;
+    int i;
+
+    if (in_lin < s->in_min_lin)
+        return s->out_min_lin;
+
+    in_log = logf(in_lin);
+
+    for (i = 1; i < s->nb_segments; i++)
+        if (in_log <= s->segments[i].x)
+            break;
+    cs = &s->segments[i - 1];
+    in_log -= cs->x;
+    out_log = cs->y + in_log * (cs->a * in_log + cs->b);
+
+    return expf(out_log);
+}
+
+static int compand_nodelay(AVFilterContext *ctx, AVFrame *frame)
+{
+    CompandContext *s    = ctx->priv;
+    AVFilterLink *inlink = ctx->inputs[0];
+    const int channels   = s->nb_channels;
+    const int nb_samples = frame->nb_samples;
+    AVFrame *out_frame;
+    int chan, i;
+    int err;
+
+    if (av_frame_is_writable(frame)) {
+        out_frame = frame;
+    } else {
+        out_frame = ff_get_audio_buffer(inlink, nb_samples);
+        if (!out_frame) {
+            av_frame_free(&frame);
+            return AVERROR(ENOMEM);
+        }
+        err = av_frame_copy_props(out_frame, frame);
+        if (err < 0) {
+            av_frame_free(&out_frame);
+            av_frame_free(&frame);
+            return err;
+        }
+    }
+
+    for (chan = 0; chan < channels; chan++) {
+        const float *src = (float *)frame->extended_data[chan];
+        float *dst = (float *)out_frame->extended_data[chan];
+        ChanParam *cp = &s->channels[chan];
+
+        for (i = 0; i < nb_samples; i++) {
+            update_volume(cp, fabs(src[i]));
+
+            dst[i] = av_clipf(src[i] * get_volume(s, cp->volume), -1.0f, 1.0f);
+        }
+    }
+
+    if (frame != out_frame)
+        av_frame_free(&frame);
+
+    return ff_filter_frame(ctx->outputs[0], out_frame);
+}
+
+#define MOD(a, b) (((a) >= (b)) ? (a) - (b) : (a))
+
+static int compand_delay(AVFilterContext *ctx, AVFrame *frame)
+{
+    CompandContext *s    = ctx->priv;
+    AVFilterLink *inlink = ctx->inputs[0];
+    const int channels   = s->nb_channels;
+    const int nb_samples = frame->nb_samples;
+    int chan, i, dindex  = 0, oindex, count = 0;
+    AVFrame *out_frame   = NULL;
+    int err;
+
+    if (s->pts == AV_NOPTS_VALUE) {
+        s->pts = (frame->pts == AV_NOPTS_VALUE) ? 0 : frame->pts;
+    }
+
+    for (chan = 0; chan < channels; chan++) {
+        AVFrame *delay_frame = s->delay_frame;
+        const float *src     = (float *)frame->extended_data[chan];
+        float *dbuf          = (float *)delay_frame->extended_data[chan];
+        ChanParam *cp        = &s->channels[chan];
+        float *dst;
+
+        count  = s->delay_count;
+        dindex = s->delay_index;
+        for (i = 0, oindex = 0; i < nb_samples; i++) {
+            const float in = src[i];
+            update_volume(cp, fabs(in));
+
+            if (count >= s->delay_samples) {
+                if (!out_frame) {
+                    out_frame = ff_get_audio_buffer(inlink, nb_samples - i);
+                    if (!out_frame) {
+                        av_frame_free(&frame);
+                        return AVERROR(ENOMEM);
+                    }
+                    err = av_frame_copy_props(out_frame, frame);
+                    if (err < 0) {
+                        av_frame_free(&out_frame);
+                        av_frame_free(&frame);
+                        return err;
+                    }
+                    out_frame->pts = s->pts;
+                    s->pts += av_rescale_q(nb_samples - i,
+                        (AVRational){ 1, inlink->sample_rate },
+                        inlink->time_base);
+                }
+
+                dst = (float *)out_frame->extended_data[chan];
+                dst[oindex++] = av_clipf(dbuf[dindex] *
+                        get_volume(s, cp->volume), -1.0f, 1.0f);
+            } else {
+                count++;
+            }
+
+            dbuf[dindex] = in;
+            dindex = MOD(dindex + 1, s->delay_samples);
+        }
+    }
+
+    s->delay_count = count;
+    s->delay_index = dindex;
+
+    av_frame_free(&frame);
+    return out_frame ? ff_filter_frame(ctx->outputs[0], out_frame) : 0;
+}
+
+static int compand_drain(AVFilterLink *outlink)
+{
+    AVFilterContext *ctx = outlink->src;
+    CompandContext *s    = ctx->priv;
+    const int channels   = s->nb_channels;
+    AVFrame *frame       = NULL;
+    int chan, i, dindex;
+
+    /* 2048 is to limit output frame size during drain */
+    frame = ff_get_audio_buffer(outlink, FFMIN(2048, s->delay_count));
+    if (!frame)
+        return AVERROR(ENOMEM);
+    frame->pts = s->pts;
+    s->pts += av_rescale_q(frame->nb_samples,
+            (AVRational){ 1, outlink->sample_rate }, outlink->time_base);
+
+    for (chan = 0; chan < channels; chan++) {
+        AVFrame *delay_frame = s->delay_frame;
+        float *dbuf = (float *)delay_frame->extended_data[chan];
+        float *dst = (float *)frame->extended_data[chan];
+        ChanParam *cp = &s->channels[chan];
+
+        dindex = s->delay_index;
+        for (i = 0; i < frame->nb_samples; i++) {
+            dst[i] = av_clipf(dbuf[dindex] * get_volume(s, cp->volume),
+                    -1.0f, 1.0f);
+            dindex = MOD(dindex + 1, s->delay_samples);
+        }
+    }
+    s->delay_count -= frame->nb_samples;
+    s->delay_index = dindex;
+
+    return ff_filter_frame(outlink, frame);
+}
+
+static int config_output(AVFilterLink *outlink)
+{
+    AVFilterContext *ctx  = outlink->src;
+    CompandContext *s     = ctx->priv;
+    const int sample_rate = outlink->sample_rate;
+    double radius         = s->curve_dB * M_LN10 / 20.0;
+    char *p, *saveptr     = NULL;
+    const int channels    =
+        av_get_channel_layout_nb_channels(outlink->channel_layout);
+    int nb_attacks, nb_decays, nb_points;
+    int new_nb_items, num;
+    int i;
+    int err;
+
+
+    count_items(s->attacks, &nb_attacks);
+    count_items(s->decays, &nb_decays);
+    count_items(s->points, &nb_points);
+
+    if (channels <= 0) {
+        av_log(ctx, AV_LOG_ERROR, "Invalid number of channels: %d\n", channels);
+        return AVERROR(EINVAL);
+    }
+
+    if (nb_attacks > channels || nb_decays > channels) {
+        av_log(ctx, AV_LOG_ERROR,
+                "Number of attacks/decays bigger than number of channels.\n");
+        return AVERROR(EINVAL);
+    }
+
+    uninit(ctx);
+
+    s->nb_channels = channels;
+    s->channels = av_mallocz_array(channels, sizeof(*s->channels));
+    s->nb_segments = (nb_points + 4) * 2;
+    s->segments = av_mallocz_array(s->nb_segments, sizeof(*s->segments));
+
+    if (!s->channels || !s->segments) {
+        uninit(ctx);
+        return AVERROR(ENOMEM);
+    }
+
+    p = s->attacks;
+    for (i = 0, new_nb_items = 0; i < nb_attacks; i++) {
+        char *tstr = strtok_r(p, "|", &saveptr);
+        p = NULL;
+        new_nb_items += sscanf(tstr, "%f", &s->channels[i].attack) == 1;
+        if (s->channels[i].attack < 0) {
+            uninit(ctx);
+            return AVERROR(EINVAL);
+        }
+    }
+    nb_attacks = new_nb_items;
+
+    p = s->decays;
+    for (i = 0, new_nb_items = 0; i < nb_decays; i++) {
+        char *tstr = strtok_r(p, "|", &saveptr);
+        p = NULL;
+        new_nb_items += sscanf(tstr, "%f", &s->channels[i].decay) == 1;
+        if (s->channels[i].decay < 0) {
+            uninit(ctx);
+            return AVERROR(EINVAL);
+        }
+    }
+    nb_decays = new_nb_items;
+
+    if (nb_attacks != nb_decays) {
+        av_log(ctx, AV_LOG_ERROR,
+                "Number of attacks %d differs from number of decays %d.\n",
+                nb_attacks, nb_decays);
+        uninit(ctx);
+        return AVERROR(EINVAL);
+    }
+
+#define S(x) s->segments[2 * ((x) + 1)]
+    p = s->points;
+    for (i = 0, new_nb_items = 0; i < nb_points; i++) {
+        char *tstr = strtok_r(p, "|", &saveptr);
+        p = NULL;
+        if (sscanf(tstr, "%f/%f", &S(i).x, &S(i).y) != 2) {
+            av_log(ctx, AV_LOG_ERROR,
+                    "Invalid and/or missing input/output value.\n");
+            uninit(ctx);
+            return AVERROR(EINVAL);
+        }
+        if (i && S(i - 1).x > S(i).x) {
+            av_log(ctx, AV_LOG_ERROR,
+                    "Transfer function input values must be increasing.\n");
+            uninit(ctx);
+            return AVERROR(EINVAL);
+        }
+        S(i).y -= S(i).x;
+        av_log(ctx, AV_LOG_DEBUG, "%d: x=%f y=%f\n", i, S(i).x, S(i).y);
+        new_nb_items++;
+    }
+    num = new_nb_items;
+
+    /* Add 0,0 if necessary */
+    if (num == 0 || S(num - 1).x)
+        num++;
+
+#undef S
+#define S(x) s->segments[2 * (x)]
+    /* Add a tail off segment at the start */
+    S(0).x = S(1).x - 2 * s->curve_dB;
+    S(0).y = S(1).y;
+    num++;
+
+    /* Join adjacent colinear segments */
+    for (i = 2; i < num; i++) {
+        double g1 = (S(i - 1).y - S(i - 2).y) * (S(i - 0).x - S(i - 1).x);
+        double g2 = (S(i - 0).y - S(i - 1).y) * (S(i - 1).x - S(i - 2).x);
+        int j;
+
+        /* here we purposefully lose precision so that we can compare floats */
+        if (fabs(g1 - g2))
+            continue;
+        num--;
+        for (j = --i; j < num; j++)
+            S(j) = S(j + 1);
+    }
+
+    for (i = 0; !i || s->segments[i - 2].x; i += 2) {
+        s->segments[i].y += s->gain_dB;
+        s->segments[i].x *= M_LN10 / 20;
+        s->segments[i].y *= M_LN10 / 20;
+    }
+
+#define L(x) s->segments[i - (x)]
+    for (i = 4; s->segments[i - 2].x; i += 2) {
+        double x, y, cx, cy, in1, in2, out1, out2, theta, len, r;
+
+        L(4).a = 0;
+        L(4).b = (L(2).y - L(4).y) / (L(2).x - L(4).x);
+
+        L(2).a = 0;
+        L(2).b = (L(0).y - L(2).y) / (L(0).x - L(2).x);
+
+        theta = atan2(L(2).y - L(4).y, L(2).x - L(4).x);
+        len = sqrt(pow(L(2).x - L(4).x, 2.) + pow(L(2).y - L(4).y, 2.));
+        r = FFMIN(radius, len);
+        L(3).x = L(2).x - r * cos(theta);
+        L(3).y = L(2).y - r * sin(theta);
+
+        theta = atan2(L(0).y - L(2).y, L(0).x - L(2).x);
+        len = sqrt(pow(L(0).x - L(2).x, 2.) + pow(L(0).y - L(2).y, 2.));
+        r = FFMIN(radius, len / 2);
+        x = L(2).x + r * cos(theta);
+        y = L(2).y + r * sin(theta);
+
+        cx = (L(3).x + L(2).x + x) / 3;
+        cy = (L(3).y + L(2).y + y) / 3;
+
+        L(2).x = x;
+        L(2).y = y;
+
+        in1  = cx - L(3).x;
+        out1 = cy - L(3).y;
+        in2  = L(2).x - L(3).x;
+        out2 = L(2).y - L(3).y;
+        L(3).a = (out2 / in2 - out1 / in1) / (in2 - in1);
+        L(3).b = out1 / in1 - L(3).a * in1;
+    }
+    L(3).x = 0;
+    L(3).y = L(2).y;
+
+    s->in_min_lin  = exp(s->segments[1].x);
+    s->out_min_lin = exp(s->segments[1].y);
+
+    for (i = 0; i < channels; i++) {
+        ChanParam *cp = &s->channels[i];
+
+        if (cp->attack > 1.0 / sample_rate)
+            cp->attack = 1.0 - exp(-1.0 / (sample_rate * cp->attack));
+        else
+            cp->attack = 1.0;
+        if (cp->decay > 1.0 / sample_rate)
+            cp->decay = 1.0 - exp(-1.0 / (sample_rate * cp->decay));
+        else
+            cp->decay = 1.0;
+        cp->volume = pow(10.0, s->initial_volume / 20);
+    }
+
+    s->delay_samples = s->delay * sample_rate;
+    if (s->delay_samples <= 0) {
+        s->compand = compand_nodelay;
+        return 0;
+    }
+
+    s->delay_frame = av_frame_alloc();
+    if (!s->delay_frame) {
+        uninit(ctx);
+        return AVERROR(ENOMEM);
+    }
+
+    s->delay_frame->format         = outlink->format;
+    s->delay_frame->nb_samples     = s->delay_samples;
+    s->delay_frame->channel_layout = outlink->channel_layout;
+
+    err = av_frame_get_buffer(s->delay_frame, 32);
+    if (err)
+        return err;
+
+    s->compand = compand_delay;
+    return 0;
+}
+
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
+{
+    AVFilterContext *ctx = inlink->dst;
+    CompandContext *s    = ctx->priv;
+
+    return s->compand(ctx, frame);
+}
+
+static int request_frame(AVFilterLink *outlink)
+{
+    AVFilterContext *ctx = outlink->src;
+    CompandContext *s    = ctx->priv;
+    int ret;
+
+    ret = ff_request_frame(ctx->inputs[0]);
+
+    if (ret == AVERROR_EOF && s->delay_count)
+        ret = compand_drain(outlink);
+
+    return ret;
+}
+
+static const AVFilterPad compand_inputs[] = {
+    {
+        .name         = "default",
+        .type         = AVMEDIA_TYPE_AUDIO,
+        .filter_frame = filter_frame,
+    },
+    { NULL }
+};
+
+static const AVFilterPad compand_outputs[] = {
+    {
+        .name          = "default",
+        .request_frame = request_frame,
+        .config_props  = config_output,
+        .type          = AVMEDIA_TYPE_AUDIO,
+    },
+    { NULL }
+};
+
+
+AVFilter ff_af_compand_fork = {
+    .name           = "compand_fork",
+    .description    = NULL_IF_CONFIG_SMALL(
+            "Compress or expand audio dynamic range."),
+    .query_formats  = query_formats,
+    .priv_size      = sizeof(CompandContext),
+    .priv_class     = &compand_class,
+    .init           = init,
+    .uninit         = uninit,
+    .inputs         = compand_inputs,
+    .outputs        = compand_outputs,
+};
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index d042b64ce3..d9a9975323 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -83,6 +83,7 @@ void avfilter_register_all(void)
     REGISTER_FILTER(CHANNELMAP,     channelmap,     af);
     REGISTER_FILTER(CHANNELSPLIT,   channelsplit,   af);
     REGISTER_FILTER(COMPAND,        compand,        af);
+    REGISTER_FILTER(COMPAND_FORK,   compand_fork,   af);
     REGISTER_FILTER(EARWAX,         earwax,         af);
     REGISTER_FILTER(EBUR128,        ebur128,        af);
     REGISTER_FILTER(EQUALIZER,      equalizer,      af);
diff --git a/libavfilter/version.h b/libavfilter/version.h
index ad47466ad0..a33ab490d0 100644
--- a/libavfilter/version.h
+++ b/libavfilter/version.h
@@ -30,8 +30,8 @@
 #include "libavutil/version.h"
 
 #define LIBAVFILTER_VERSION_MAJOR   4
-#define LIBAVFILTER_VERSION_MINOR   1
-#define LIBAVFILTER_VERSION_MICRO 103
+#define LIBAVFILTER_VERSION_MINOR   2
+#define LIBAVFILTER_VERSION_MICRO 100
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \
                                                LIBAVFILTER_VERSION_MINOR, \