1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
FFmpeg/libavfilter/af_anequalizer.c
Andreas Rheinhardt b4f5201967 avfilter: Replace query_formats callback with union of list and callback
If one looks at the many query_formats callbacks in existence,
one will immediately recognize that there is one type of default
callback for video and a slightly different default callback for
audio: It is "return ff_set_common_formats_from_list(ctx, pix_fmts);"
for video with a filter-specific pix_fmts list. For audio, it is
the same with a filter-specific sample_fmts list together with
ff_set_common_all_samplerates() and ff_set_common_all_channel_counts().

This commit allows to remove the boilerplate query_formats callbacks
by replacing said callback with a union consisting the old callback
and pointers for pixel and sample format arrays. For the not uncommon
case in which these lists only contain a single entry (besides the
sentinel) enum AVPixelFormat and enum AVSampleFormat fields are also
added to the union to store them directly in the AVFilter,
thereby avoiding a relocation.

The state of said union will be contained in a new, dedicated AVFilter
field (the nb_inputs and nb_outputs fields have been shrunk to uint8_t
in order to create a hole for this new field; this is no problem, as
the maximum of all the nb_inputs is four; for nb_outputs it is only
two).

The state's default value coincides with the earlier default of
query_formats being unset, namely that the filter accepts all formats
(and also sample rates and channel counts/layouts for audio)
provided that these properties agree coincide for all inputs and
outputs.

By using different union members for audio and video filters
the type-unsafety of using the same functions for audio and video
lists will furthermore be more confined to formats.c than before.

When the new fields are used, they will also avoid allocations:
Currently something nearly equivalent to ff_default_query_formats()
is called after every successful call to a query_formats callback;
yet in the common case that the newly allocated AVFilterFormats
are not used at all (namely if there are no free links) these newly
allocated AVFilterFormats are freed again without ever being used.
Filters no longer using the callback will not exhibit this any more.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2021-10-05 17:48:25 +02:00

781 lines
24 KiB
C

/*
* Copyright (c) 2001-2010 Krzysztof Foltman, Markus Schmidt, Thor Harald Johansen and others
* Copyright (c) 2015 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/intreadwrite.h"
#include "libavutil/avstring.h"
#include "libavutil/ffmath.h"
#include "libavutil/opt.h"
#include "libavutil/parseutils.h"
#include "avfilter.h"
#include "internal.h"
#include "audio.h"
#define FILTER_ORDER 4
enum FilterType {
BUTTERWORTH,
CHEBYSHEV1,
CHEBYSHEV2,
NB_TYPES
};
typedef struct FoSection {
double a0, a1, a2, a3, a4;
double b0, b1, b2, b3, b4;
double num[4];
double denum[4];
} FoSection;
typedef struct EqualizatorFilter {
int ignore;
int channel;
int type;
double freq;
double gain;
double width;
FoSection section[2];
} EqualizatorFilter;
typedef struct AudioNEqualizerContext {
const AVClass *class;
char *args;
char *colors;
int draw_curves;
int w, h;
double mag;
int fscale;
int nb_filters;
int nb_allocated;
EqualizatorFilter *filters;
AVFrame *video;
} AudioNEqualizerContext;
#define OFFSET(x) offsetof(AudioNEqualizerContext, x)
#define A AV_OPT_FLAG_AUDIO_PARAM
#define V AV_OPT_FLAG_VIDEO_PARAM
#define F AV_OPT_FLAG_FILTERING_PARAM
static const AVOption anequalizer_options[] = {
{ "params", NULL, OFFSET(args), AV_OPT_TYPE_STRING, {.str=""}, 0, 0, A|F },
{ "curves", "draw frequency response curves", OFFSET(draw_curves), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, V|F },
{ "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "hd720"}, 0, 0, V|F },
{ "mgain", "set max gain", OFFSET(mag), AV_OPT_TYPE_DOUBLE, {.dbl=60}, -900, 900, V|F },
{ "fscale", "set frequency scale", OFFSET(fscale), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, V|F, "fscale" },
{ "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, V|F, "fscale" },
{ "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, V|F, "fscale" },
{ "colors", "set channels curves colors", OFFSET(colors), AV_OPT_TYPE_STRING, {.str = "red|green|blue|yellow|orange|lime|pink|magenta|brown" }, 0, 0, V|F },
{ NULL }
};
AVFILTER_DEFINE_CLASS(anequalizer);
static void draw_curves(AVFilterContext *ctx, AVFilterLink *inlink, AVFrame *out)
{
AudioNEqualizerContext *s = ctx->priv;
char *colors, *color, *saveptr = NULL;
int ch, i, n;
colors = av_strdup(s->colors);
if (!colors)
return;
memset(out->data[0], 0, s->h * out->linesize[0]);
for (ch = 0; ch < inlink->channels; ch++) {
uint8_t fg[4] = { 0xff, 0xff, 0xff, 0xff };
int prev_v = -1;
double f;
color = av_strtok(ch == 0 ? colors : NULL, " |", &saveptr);
if (color)
av_parse_color(fg, color, -1, ctx);
for (f = 0; f < s->w; f++) {
double zr, zi, zr2, zi2;
double Hr, Hi;
double Hmag = 1;
double w;
int v, y, x;
w = M_PI * (s->fscale ? pow(s->w - 1, f / s->w) : f) / (s->w - 1);
zr = cos(w);
zr2 = zr * zr;
zi = -sin(w);
zi2 = zi * zi;
for (n = 0; n < s->nb_filters; n++) {
if (s->filters[n].channel != ch ||
s->filters[n].ignore)
continue;
for (i = 0; i < FILTER_ORDER / 2; i++) {
FoSection *S = &s->filters[n].section[i];
/* H *= (((((S->b4 * z + S->b3) * z + S->b2) * z + S->b1) * z + S->b0) /
((((S->a4 * z + S->a3) * z + S->a2) * z + S->a1) * z + S->a0)); */
Hr = S->b4*(1-8*zr2*zi2) + S->b2*(zr2-zi2) + zr*(S->b1+S->b3*(zr2-3*zi2))+ S->b0;
Hi = zi*(S->b3*(3*zr2-zi2) + S->b1 + 2*zr*(2*S->b4*(zr2-zi2) + S->b2));
Hmag *= hypot(Hr, Hi);
Hr = S->a4*(1-8*zr2*zi2) + S->a2*(zr2-zi2) + zr*(S->a1+S->a3*(zr2-3*zi2))+ S->a0;
Hi = zi*(S->a3*(3*zr2-zi2) + S->a1 + 2*zr*(2*S->a4*(zr2-zi2) + S->a2));
Hmag /= hypot(Hr, Hi);
}
}
v = av_clip((1. + -20 * log10(Hmag) / s->mag) * s->h / 2, 0, s->h - 1);
x = lrint(f);
if (prev_v == -1)
prev_v = v;
if (v <= prev_v) {
for (y = v; y <= prev_v; y++)
AV_WL32(out->data[0] + y * out->linesize[0] + x * 4, AV_RL32(fg));
} else {
for (y = prev_v; y <= v; y++)
AV_WL32(out->data[0] + y * out->linesize[0] + x * 4, AV_RL32(fg));
}
prev_v = v;
}
}
av_free(colors);
}
static int config_video(AVFilterLink *outlink)
{
AVFilterContext *ctx = outlink->src;
AudioNEqualizerContext *s = ctx->priv;
AVFilterLink *inlink = ctx->inputs[0];
AVFrame *out;
outlink->w = s->w;
outlink->h = s->h;
av_frame_free(&s->video);
s->video = out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out)
return AVERROR(ENOMEM);
outlink->sample_aspect_ratio = (AVRational){1,1};
draw_curves(ctx, inlink, out);
return 0;
}
static av_cold int init(AVFilterContext *ctx)
{
AudioNEqualizerContext *s = ctx->priv;
AVFilterPad pad, vpad;
int ret;
pad = (AVFilterPad){
.name = "out0",
.type = AVMEDIA_TYPE_AUDIO,
};
ret = ff_append_outpad(ctx, &pad);
if (ret < 0)
return ret;
if (s->draw_curves) {
vpad = (AVFilterPad){
.name = "out1",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_video,
};
ret = ff_append_outpad(ctx, &vpad);
if (ret < 0)
return ret;
}
return 0;
}
static int query_formats(AVFilterContext *ctx)
{
AVFilterLink *inlink = ctx->inputs[0];
AVFilterLink *outlink = ctx->outputs[0];
AudioNEqualizerContext *s = ctx->priv;
AVFilterFormats *formats;
AVFilterChannelLayouts *layouts;
static const enum AVPixelFormat pix_fmts[] = { AV_PIX_FMT_RGBA, AV_PIX_FMT_NONE };
static const enum AVSampleFormat sample_fmts[] = {
AV_SAMPLE_FMT_DBLP,
AV_SAMPLE_FMT_NONE
};
int ret;
if (s->draw_curves) {
AVFilterLink *videolink = ctx->outputs[1];
formats = ff_make_format_list(pix_fmts);
if ((ret = ff_formats_ref(formats, &videolink->incfg.formats)) < 0)
return ret;
}
formats = ff_make_format_list(sample_fmts);
if ((ret = ff_formats_ref(formats, &inlink->outcfg.formats)) < 0 ||
(ret = ff_formats_ref(formats, &outlink->incfg.formats)) < 0)
return ret;
layouts = ff_all_channel_counts();
if ((ret = ff_channel_layouts_ref(layouts, &inlink->outcfg.channel_layouts)) < 0 ||
(ret = ff_channel_layouts_ref(layouts, &outlink->incfg.channel_layouts)) < 0)
return ret;
formats = ff_all_samplerates();
if ((ret = ff_formats_ref(formats, &inlink->outcfg.samplerates)) < 0 ||
(ret = ff_formats_ref(formats, &outlink->incfg.samplerates)) < 0)
return ret;
return 0;
}
static av_cold void uninit(AVFilterContext *ctx)
{
AudioNEqualizerContext *s = ctx->priv;
av_frame_free(&s->video);
av_freep(&s->filters);
s->nb_filters = 0;
s->nb_allocated = 0;
}
static void butterworth_fo_section(FoSection *S, double beta,
double si, double g, double g0,
double D, double c0)
{
if (c0 == 1 || c0 == -1) {
S->b0 = (g*g*beta*beta + 2*g*g0*si*beta + g0*g0)/D;
S->b1 = 2*c0*(g*g*beta*beta - g0*g0)/D;
S->b2 = (g*g*beta*beta - 2*g0*g*beta*si + g0*g0)/D;
S->b3 = 0;
S->b4 = 0;
S->a0 = 1;
S->a1 = 2*c0*(beta*beta - 1)/D;
S->a2 = (beta*beta - 2*beta*si + 1)/D;
S->a3 = 0;
S->a4 = 0;
} else {
S->b0 = (g*g*beta*beta + 2*g*g0*si*beta + g0*g0)/D;
S->b1 = -4*c0*(g0*g0 + g*g0*si*beta)/D;
S->b2 = 2*(g0*g0*(1 + 2*c0*c0) - g*g*beta*beta)/D;
S->b3 = -4*c0*(g0*g0 - g*g0*si*beta)/D;
S->b4 = (g*g*beta*beta - 2*g*g0*si*beta + g0*g0)/D;
S->a0 = 1;
S->a1 = -4*c0*(1 + si*beta)/D;
S->a2 = 2*(1 + 2*c0*c0 - beta*beta)/D;
S->a3 = -4*c0*(1 - si*beta)/D;
S->a4 = (beta*beta - 2*si*beta + 1)/D;
}
}
static void butterworth_bp_filter(EqualizatorFilter *f,
int N, double w0, double wb,
double G, double Gb, double G0)
{
double g, c0, g0, beta;
double epsilon;
int r = N % 2;
int L = (N - r) / 2;
int i;
if (G == 0 && G0 == 0) {
f->section[0].a0 = 1;
f->section[0].b0 = 1;
f->section[1].a0 = 1;
f->section[1].b0 = 1;
return;
}
G = ff_exp10(G/20);
Gb = ff_exp10(Gb/20);
G0 = ff_exp10(G0/20);
epsilon = sqrt((G * G - Gb * Gb) / (Gb * Gb - G0 * G0));
g = pow(G, 1.0 / N);
g0 = pow(G0, 1.0 / N);
beta = pow(epsilon, -1.0 / N) * tan(wb/2);
c0 = cos(w0);
for (i = 1; i <= L; i++) {
double ui = (2.0 * i - 1) / N;
double si = sin(M_PI * ui / 2.0);
double Di = beta * beta + 2 * si * beta + 1;
butterworth_fo_section(&f->section[i - 1], beta, si, g, g0, Di, c0);
}
}
static void chebyshev1_fo_section(FoSection *S, double a,
double c, double tetta_b,
double g0, double si, double b,
double D, double c0)
{
if (c0 == 1 || c0 == -1) {
S->b0 = (tetta_b*tetta_b*(b*b+g0*g0*c*c) + 2*g0*b*si*tetta_b*tetta_b + g0*g0)/D;
S->b1 = 2*c0*(tetta_b*tetta_b*(b*b+g0*g0*c*c) - g0*g0)/D;
S->b2 = (tetta_b*tetta_b*(b*b+g0*g0*c*c) - 2*g0*b*si*tetta_b + g0*g0)/D;
S->b3 = 0;
S->b4 = 0;
S->a0 = 1;
S->a1 = 2*c0*(tetta_b*tetta_b*(a*a+c*c) - 1)/D;
S->a2 = (tetta_b*tetta_b*(a*a+c*c) - 2*a*si*tetta_b + 1)/D;
S->a3 = 0;
S->a4 = 0;
} else {
S->b0 = ((b*b + g0*g0*c*c)*tetta_b*tetta_b + 2*g0*b*si*tetta_b + g0*g0)/D;
S->b1 = -4*c0*(g0*g0 + g0*b*si*tetta_b)/D;
S->b2 = 2*(g0*g0*(1 + 2*c0*c0) - (b*b + g0*g0*c*c)*tetta_b*tetta_b)/D;
S->b3 = -4*c0*(g0*g0 - g0*b*si*tetta_b)/D;
S->b4 = ((b*b + g0*g0*c*c)*tetta_b*tetta_b - 2*g0*b*si*tetta_b + g0*g0)/D;
S->a0 = 1;
S->a1 = -4*c0*(1 + a*si*tetta_b)/D;
S->a2 = 2*(1 + 2*c0*c0 - (a*a + c*c)*tetta_b*tetta_b)/D;
S->a3 = -4*c0*(1 - a*si*tetta_b)/D;
S->a4 = ((a*a + c*c)*tetta_b*tetta_b - 2*a*si*tetta_b + 1)/D;
}
}
static void chebyshev1_bp_filter(EqualizatorFilter *f,
int N, double w0, double wb,
double G, double Gb, double G0)
{
double a, b, c0, g0, alfa, beta, tetta_b;
double epsilon;
int r = N % 2;
int L = (N - r) / 2;
int i;
if (G == 0 && G0 == 0) {
f->section[0].a0 = 1;
f->section[0].b0 = 1;
f->section[1].a0 = 1;
f->section[1].b0 = 1;
return;
}
G = ff_exp10(G/20);
Gb = ff_exp10(Gb/20);
G0 = ff_exp10(G0/20);
epsilon = sqrt((G*G - Gb*Gb) / (Gb*Gb - G0*G0));
g0 = pow(G0,1.0/N);
alfa = pow(1.0/epsilon + sqrt(1 + 1/(epsilon*epsilon)), 1.0/N);
beta = pow(G/epsilon + Gb * sqrt(1 + 1/(epsilon*epsilon)), 1.0/N);
a = 0.5 * (alfa - 1.0/alfa);
b = 0.5 * (beta - g0*g0*(1/beta));
tetta_b = tan(wb/2);
c0 = cos(w0);
for (i = 1; i <= L; i++) {
double ui = (2.0*i-1.0)/N;
double ci = cos(M_PI*ui/2.0);
double si = sin(M_PI*ui/2.0);
double Di = (a*a + ci*ci)*tetta_b*tetta_b + 2.0*a*si*tetta_b + 1;
chebyshev1_fo_section(&f->section[i - 1], a, ci, tetta_b, g0, si, b, Di, c0);
}
}
static void chebyshev2_fo_section(FoSection *S, double a,
double c, double tetta_b,
double g, double si, double b,
double D, double c0)
{
if (c0 == 1 || c0 == -1) {
S->b0 = (g*g*tetta_b*tetta_b + 2*tetta_b*g*b*si + b*b + g*g*c*c)/D;
S->b1 = 2*c0*(g*g*tetta_b*tetta_b - b*b - g*g*c*c)/D;
S->b2 = (g*g*tetta_b*tetta_b - 2*tetta_b*g*b*si + b*b + g*g*c*c)/D;
S->b3 = 0;
S->b4 = 0;
S->a0 = 1;
S->a1 = 2*c0*(tetta_b*tetta_b - a*a - c*c)/D;
S->a2 = (tetta_b*tetta_b - 2*tetta_b*a*si + a*a + c*c)/D;
S->a3 = 0;
S->a4 = 0;
} else {
S->b0 = (g*g*tetta_b*tetta_b + 2*g*b*si*tetta_b + b*b + g*g*c*c)/D;
S->b1 = -4*c0*(b*b + g*g*c*c + g*b*si*tetta_b)/D;
S->b2 = 2*((b*b + g*g*c*c)*(1 + 2*c0*c0) - g*g*tetta_b*tetta_b)/D;
S->b3 = -4*c0*(b*b + g*g*c*c - g*b*si*tetta_b)/D;
S->b4 = (g*g*tetta_b*tetta_b - 2*g*b*si*tetta_b + b*b + g*g*c*c)/D;
S->a0 = 1;
S->a1 = -4*c0*(a*a + c*c + a*si*tetta_b)/D;
S->a2 = 2*((a*a + c*c)*(1 + 2*c0*c0) - tetta_b*tetta_b)/D;
S->a3 = -4*c0*(a*a + c*c - a*si*tetta_b)/D;
S->a4 = (tetta_b*tetta_b - 2*a*si*tetta_b + a*a + c*c)/D;
}
}
static void chebyshev2_bp_filter(EqualizatorFilter *f,
int N, double w0, double wb,
double G, double Gb, double G0)
{
double a, b, c0, tetta_b;
double epsilon, g, eu, ew;
int r = N % 2;
int L = (N - r) / 2;
int i;
if (G == 0 && G0 == 0) {
f->section[0].a0 = 1;
f->section[0].b0 = 1;
f->section[1].a0 = 1;
f->section[1].b0 = 1;
return;
}
G = ff_exp10(G/20);
Gb = ff_exp10(Gb/20);
G0 = ff_exp10(G0/20);
epsilon = sqrt((G*G - Gb*Gb) / (Gb*Gb - G0*G0));
g = pow(G, 1.0 / N);
eu = pow(epsilon + sqrt(1 + epsilon*epsilon), 1.0/N);
ew = pow(G0*epsilon + Gb*sqrt(1 + epsilon*epsilon), 1.0/N);
a = (eu - 1.0/eu)/2.0;
b = (ew - g*g/ew)/2.0;
tetta_b = tan(wb/2);
c0 = cos(w0);
for (i = 1; i <= L; i++) {
double ui = (2.0 * i - 1.0)/N;
double ci = cos(M_PI * ui / 2.0);
double si = sin(M_PI * ui / 2.0);
double Di = tetta_b*tetta_b + 2*a*si*tetta_b + a*a + ci*ci;
chebyshev2_fo_section(&f->section[i - 1], a, ci, tetta_b, g, si, b, Di, c0);
}
}
static double butterworth_compute_bw_gain_db(double gain)
{
double bw_gain = 0;
if (gain <= -6)
bw_gain = gain + 3;
else if(gain > -6 && gain < 6)
bw_gain = gain * 0.5;
else if(gain >= 6)
bw_gain = gain - 3;
return bw_gain;
}
static double chebyshev1_compute_bw_gain_db(double gain)
{
double bw_gain = 0;
if (gain <= -6)
bw_gain = gain + 1;
else if(gain > -6 && gain < 6)
bw_gain = gain * 0.9;
else if(gain >= 6)
bw_gain = gain - 1;
return bw_gain;
}
static double chebyshev2_compute_bw_gain_db(double gain)
{
double bw_gain = 0;
if (gain <= -6)
bw_gain = -3;
else if(gain > -6 && gain < 6)
bw_gain = gain * 0.3;
else if(gain >= 6)
bw_gain = 3;
return bw_gain;
}
static inline double hz_2_rad(double x, double fs)
{
return 2 * M_PI * x / fs;
}
static void equalizer(EqualizatorFilter *f, double sample_rate)
{
double w0 = hz_2_rad(f->freq, sample_rate);
double wb = hz_2_rad(f->width, sample_rate);
double bw_gain;
switch (f->type) {
case BUTTERWORTH:
bw_gain = butterworth_compute_bw_gain_db(f->gain);
butterworth_bp_filter(f, FILTER_ORDER, w0, wb, f->gain, bw_gain, 0);
break;
case CHEBYSHEV1:
bw_gain = chebyshev1_compute_bw_gain_db(f->gain);
chebyshev1_bp_filter(f, FILTER_ORDER, w0, wb, f->gain, bw_gain, 0);
break;
case CHEBYSHEV2:
bw_gain = chebyshev2_compute_bw_gain_db(f->gain);
chebyshev2_bp_filter(f, FILTER_ORDER, w0, wb, f->gain, bw_gain, 0);
break;
}
}
static int add_filter(AudioNEqualizerContext *s, AVFilterLink *inlink)
{
equalizer(&s->filters[s->nb_filters], inlink->sample_rate);
if (s->nb_filters >= s->nb_allocated - 1) {
EqualizatorFilter *filters;
filters = av_calloc(s->nb_allocated, 2 * sizeof(*s->filters));
if (!filters)
return AVERROR(ENOMEM);
memcpy(filters, s->filters, sizeof(*s->filters) * s->nb_allocated);
av_free(s->filters);
s->filters = filters;
s->nb_allocated *= 2;
}
s->nb_filters++;
return 0;
}
static int config_input(AVFilterLink *inlink)
{
AVFilterContext *ctx = inlink->dst;
AudioNEqualizerContext *s = ctx->priv;
char *args = av_strdup(s->args);
char *saveptr = NULL;
int ret = 0;
if (!args)
return AVERROR(ENOMEM);
s->nb_allocated = 32 * inlink->channels;
s->filters = av_calloc(inlink->channels, 32 * sizeof(*s->filters));
if (!s->filters) {
s->nb_allocated = 0;
av_free(args);
return AVERROR(ENOMEM);
}
while (1) {
char *arg = av_strtok(s->nb_filters == 0 ? args : NULL, "|", &saveptr);
if (!arg)
break;
s->filters[s->nb_filters].type = 0;
if (sscanf(arg, "c%d f=%lf w=%lf g=%lf t=%d", &s->filters[s->nb_filters].channel,
&s->filters[s->nb_filters].freq,
&s->filters[s->nb_filters].width,
&s->filters[s->nb_filters].gain,
&s->filters[s->nb_filters].type) != 5 &&
sscanf(arg, "c%d f=%lf w=%lf g=%lf", &s->filters[s->nb_filters].channel,
&s->filters[s->nb_filters].freq,
&s->filters[s->nb_filters].width,
&s->filters[s->nb_filters].gain) != 4 ) {
av_free(args);
return AVERROR(EINVAL);
}
if (s->filters[s->nb_filters].freq < 0 ||
s->filters[s->nb_filters].freq > inlink->sample_rate / 2.0)
s->filters[s->nb_filters].ignore = 1;
if (s->filters[s->nb_filters].channel < 0 ||
s->filters[s->nb_filters].channel >= inlink->channels)
s->filters[s->nb_filters].ignore = 1;
s->filters[s->nb_filters].type = av_clip(s->filters[s->nb_filters].type, 0, NB_TYPES - 1);
ret = add_filter(s, inlink);
if (ret < 0)
break;
}
av_free(args);
return ret;
}
static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
char *res, int res_len, int flags)
{
AudioNEqualizerContext *s = ctx->priv;
AVFilterLink *inlink = ctx->inputs[0];
int ret = AVERROR(ENOSYS);
if (!strcmp(cmd, "change")) {
double freq, width, gain;
int filter;
if (sscanf(args, "%d|f=%lf|w=%lf|g=%lf", &filter, &freq, &width, &gain) != 4)
return AVERROR(EINVAL);
if (filter < 0 || filter >= s->nb_filters)
return AVERROR(EINVAL);
if (freq < 0 || freq > inlink->sample_rate / 2.0)
return AVERROR(EINVAL);
s->filters[filter].freq = freq;
s->filters[filter].width = width;
s->filters[filter].gain = gain;
equalizer(&s->filters[filter], inlink->sample_rate);
if (s->draw_curves)
draw_curves(ctx, inlink, s->video);
ret = 0;
}
return ret;
}
static inline double section_process(FoSection *S, double in)
{
double out;
out = S->b0 * in;
out+= S->b1 * S->num[0] - S->denum[0] * S->a1;
out+= S->b2 * S->num[1] - S->denum[1] * S->a2;
out+= S->b3 * S->num[2] - S->denum[2] * S->a3;
out+= S->b4 * S->num[3] - S->denum[3] * S->a4;
S->num[3] = S->num[2];
S->num[2] = S->num[1];
S->num[1] = S->num[0];
S->num[0] = in;
S->denum[3] = S->denum[2];
S->denum[2] = S->denum[1];
S->denum[1] = S->denum[0];
S->denum[0] = out;
return out;
}
static double process_sample(FoSection *s1, double in)
{
double p0 = in, p1;
int i;
for (i = 0; i < FILTER_ORDER / 2; i++) {
p1 = section_process(&s1[i], p0);
p0 = p1;
}
return p1;
}
static int filter_channels(AVFilterContext *ctx, void *arg,
int jobnr, int nb_jobs)
{
AudioNEqualizerContext *s = ctx->priv;
AVFrame *buf = arg;
const int start = (buf->channels * jobnr) / nb_jobs;
const int end = (buf->channels * (jobnr+1)) / nb_jobs;
for (int i = 0; i < s->nb_filters; i++) {
EqualizatorFilter *f = &s->filters[i];
double *bptr;
if (f->gain == 0. || f->ignore)
continue;
if (f->channel < start ||
f->channel >= end)
continue;
bptr = (double *)buf->extended_data[f->channel];
for (int n = 0; n < buf->nb_samples; n++) {
double sample = bptr[n];
sample = process_sample(f->section, sample);
bptr[n] = sample;
}
}
return 0;
}
static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
{
AVFilterContext *ctx = inlink->dst;
AudioNEqualizerContext *s = ctx->priv;
AVFilterLink *outlink = ctx->outputs[0];
if (!ctx->is_disabled)
ff_filter_execute(ctx, filter_channels, buf, NULL,
FFMIN(inlink->channels, ff_filter_get_nb_threads(ctx)));
if (s->draw_curves) {
AVFrame *clone;
const int64_t pts = buf->pts +
av_rescale_q(buf->nb_samples, (AVRational){ 1, inlink->sample_rate },
outlink->time_base);
int ret;
s->video->pts = pts;
clone = av_frame_clone(s->video);
if (!clone)
return AVERROR(ENOMEM);
ret = ff_filter_frame(ctx->outputs[1], clone);
if (ret < 0)
return ret;
}
return ff_filter_frame(outlink, buf);
}
static const AVFilterPad inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.flags = AVFILTERPAD_FLAG_NEEDS_WRITABLE,
.config_props = config_input,
.filter_frame = filter_frame,
},
};
const AVFilter ff_af_anequalizer = {
.name = "anequalizer",
.description = NULL_IF_CONFIG_SMALL("Apply high-order audio parametric multi band equalizer."),
.priv_size = sizeof(AudioNEqualizerContext),
.priv_class = &anequalizer_class,
.init = init,
.uninit = uninit,
FILTER_INPUTS(inputs),
.outputs = NULL,
FILTER_QUERY_FUNC(query_formats),
.process_command = process_command,
.flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS |
AVFILTER_FLAG_SUPPORT_TIMELINE_INTERNAL |
AVFILTER_FLAG_SLICE_THREADS,
};