[FFmpeg-devel] [PATCH] avfilter: add stereoupmix
Lingjiang Fang
vacingfang at foxmail.com
Mon Dec 14 06:13:19 EET 2020
On Wed, 9 Dec 2020 18:20:07 +0100
Paul B Mahol <onemda at gmail.com> wrote:
>Signed-off-by: Paul B Mahol <onemda at gmail.com>
>---
> doc/filters.texi | 34 ++++
> libavfilter/Makefile | 1 +
> libavfilter/af_stereoupmix.c | 352 +++++++++++++++++++++++++++++++++++
> libavfilter/allfilters.c | 1 +
> 4 files changed, 388 insertions(+)
> create mode 100644 libavfilter/af_stereoupmix.c
>
>diff --git a/doc/filters.texi b/doc/filters.texi
>index 9dfe95f40d..325753c8f4 100644
>--- a/doc/filters.texi
>+++ b/doc/filters.texi
>@@ -5817,6 +5817,40 @@ Convert M/S signal to L/R:
> @end example
> @end itemize
>
>+ at section stereoupmix
>+Upmix stereo audio.
>+
>+This filter upmixes stereo audio using adaptive panning method.
As far as I know, we have a filter surround has similar function,
can you describe the difference between these two filter.
sorry if asked stupid question
>+
>+The filter accepts the following options:
>+
>+ at table @option
>+ at item upmix
>+Set the upmix mode. Can be one of the following:
>+ at table @samp
>+ at item 2.1
>+ at item 3.0
>+ at item 3.1
>+ at item 4.0
>+ at item 4.1
>+ at item 5.0
>+ at item 5.1
>+ at end table
>+Default value is @var{5.1}.
>+
>+ at item center
>+Set the center audio strength. Allowed range is from 0.0 to 1.0.
>+Default value is 0.5.
>+
>+ at item ambience
>+Set the ambience audio strength. Allowed range is from 0.0 to 1.0.
>+Default value is 0.5.
>+ at end table
>+
>+ at subsection Commands
>+
>+This filter supports the all above options except @code{upmix} as
>@ref{commands}. +
> @section stereowiden
>
> This filter enhance the stereo effect by suppressing signal common to
> both
>diff --git a/libavfilter/Makefile b/libavfilter/Makefile
>index 1af85a71a0..a9d76a1eaf 100644
>--- a/libavfilter/Makefile
>+++ b/libavfilter/Makefile
>@@ -145,6 +145,7 @@ OBJS-$(CONFIG_SILENCEREMOVE_FILTER) +=
>af_silenceremove.o
> OBJS-$(CONFIG_SOFALIZER_FILTER) += af_sofalizer.o
> OBJS-$(CONFIG_SPEECHNORM_FILTER) += af_speechnorm.o
> OBJS-$(CONFIG_STEREOTOOLS_FILTER) += af_stereotools.o
>+OBJS-$(CONFIG_STEREOUPMIX_FILTER) += af_stereoupmix.o
> OBJS-$(CONFIG_STEREOWIDEN_FILTER) += af_stereowiden.o
> OBJS-$(CONFIG_SUPEREQUALIZER_FILTER) += af_superequalizer.o
> OBJS-$(CONFIG_SURROUND_FILTER) += af_surround.o
>diff --git a/libavfilter/af_stereoupmix.c
>b/libavfilter/af_stereoupmix.c new file mode 100644
>index 0000000000..813f21b088
>--- /dev/null
>+++ b/libavfilter/af_stereoupmix.c
>@@ -0,0 +1,352 @@
>+/*
>+ * Copyright (c) 2020 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/avassert.h"
>+#include "libavutil/channel_layout.h"
>+#include "libavutil/opt.h"
>+#include "avfilter.h"
>+#include "audio.h"
>+#include "formats.h"
>+
>+enum UpmixMode {
>+ UPMIX_2_1,
>+ UPMIX_3_0,
>+ UPMIX_3_1,
>+ UPMIX_4_0,
>+ UPMIX_4_1,
>+ UPMIX_5_0,
>+ UPMIX_5_1,
>+ NB_UPMIX
>+};
>+
>+typedef struct StereoUpmixContext {
>+ const AVClass *class;
>+
>+ int upmix;
>+ float center;
>+ float ambience;
>+
>+ uint64_t out_layout;
>+
>+ float fl, fr;
>+ float y;
>+ float pk;
>+ float wl, wr;
>+
>+ float a[2];
>+ float b[3];
>+ float z[2];
>+} StereoUpmixContext;
>+
>+#define OFFSET(x) offsetof(StereoUpmixContext, x)
>+#define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
>+#define TFLAGS
>AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
>+ +static const AVOption stereoupmix_options[] = {
>+ { "upmix", "set upmix mode", OFFSET(upmix),
>AV_OPT_TYPE_INT, {.i64=UPMIX_5_1}, 0, NB_UPMIX-1, FLAGS, "upmix" },
>+ { "2.1", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_2_1}, 0, 0, FLAGS, "upmix" },
>+ { "3.0", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_3_0}, 0, 0, FLAGS, "upmix" },
>+ { "3.1", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_3_1}, 0, 0, FLAGS, "upmix" },
>+ { "4.0", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_4_0}, 0, 0, FLAGS, "upmix" },
>+ { "4.1", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_4_1}, 0, 0, FLAGS, "upmix" },
>+ { "5.0", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_5_0}, 0, 0, FLAGS, "upmix" },
>+ { "5.1", NULL, 0,
>AV_OPT_TYPE_CONST, {.i64=UPMIX_5_1}, 0, 0, FLAGS, "upmix" },
>+ { "center", "set center strength", OFFSET(center),
>AV_OPT_TYPE_FLOAT, {.dbl=0.5}, 0, 1, TFLAGS },
>+ { "ambience", "set ambience strength", OFFSET(ambience),
>AV_OPT_TYPE_FLOAT, {.dbl=0.5}, 0, 1, TFLAGS },
>+ { NULL }
>+};
>+
>+AVFILTER_DEFINE_CLASS(stereoupmix);
>+
>+static av_cold int init(AVFilterContext *ctx)
>+{
>+ StereoUpmixContext *s = ctx->priv;
>+
>+ s->wl = s->wr = M_SQRT1_2;
>+
>+ switch (s->upmix) {
>+ case UPMIX_2_1:
>+ s->out_layout = AV_CH_LAYOUT_2POINT1;
>+ break;
>+ case UPMIX_3_0:
>+ s->out_layout = AV_CH_LAYOUT_SURROUND;
>+ break;
>+ case UPMIX_3_1:
>+ s->out_layout = AV_CH_LAYOUT_3POINT1;
>+ break;
>+ case UPMIX_4_0:
>+ s->out_layout = AV_CH_LAYOUT_4POINT0;
>+ break;
>+ case UPMIX_4_1:
>+ s->out_layout = AV_CH_LAYOUT_4POINT1;
>+ break;
>+ case UPMIX_5_0:
>+ s->out_layout = AV_CH_LAYOUT_5POINT0_BACK;
>+ break;
>+ case UPMIX_5_1:
>+ s->out_layout = AV_CH_LAYOUT_5POINT1_BACK;
>+ break;
>+ default:
>+ av_assert0(0);
>+ }
>+
>+ return 0;
>+}
>+
>+static int query_formats(AVFilterContext *ctx)
>+{
>+ StereoUpmixContext *s = ctx->priv;
>+ AVFilterFormats *formats = NULL;
>+ AVFilterChannelLayouts *layouts = NULL;
>+ int ret;
>+
>+ ret = ff_add_format(&formats, AV_SAMPLE_FMT_FLT);
>+ if (ret)
>+ return ret;
>+ ret = ff_set_common_formats(ctx, formats);
>+ if (ret)
>+ return ret;
>+
>+ layouts = NULL;
>+ ret = ff_add_channel_layout(&layouts, s->out_layout);
>+ if (ret)
>+ return ret;
>+
>+ ret = ff_channel_layouts_ref(layouts,
>&ctx->outputs[0]->incfg.channel_layouts);
>+ if (ret)
>+ return ret;
>+
>+ layouts = NULL;
>+ ret = ff_add_channel_layout(&layouts, AV_CH_LAYOUT_STEREO);
>+ if (ret)
>+ return ret;
>+
>+ ret = ff_channel_layouts_ref(layouts,
>&ctx->inputs[0]->outcfg.channel_layouts);
>+ if (ret)
>+ return ret;
>+
>+ formats = ff_all_samplerates();
>+ if (!formats)
>+ return AVERROR(ENOMEM);
>+ return ff_set_common_samplerates(ctx, formats);
>+}
>+
>+static void upmix(AVFilterContext *ctx, AVFrame *out, AVFrame *in)
>+{
>+ StereoUpmixContext *s = ctx->priv;
>+ const float center = s->center;
>+ const float ambience = s->ambience;
>+ const float *src = (const float *)in->data[0];
>+ const float b0 = s->b[0];
>+ const float b1 = s->b[1];
>+ const float b2 = s->b[2];
>+ const float a1 = s->a[0];
>+ const float a2 = s->a[1];
>+ float *dst = (float *)out->data[0];
>+ float wl = s->wl, wr = s->wr;
>+ float fl = s->fl, fr = s->fr;
>+ float pk = s->pk;
>+ float y = s->y;
>+ float z0 = s->z[0];
>+ float z1 = s->z[1];
>+
>+ for (int n = 0; n < in->nb_samples; n++) {
>+ float nwl, nwr;
>+ float clr, cl, cr;
>+ float slr, sl, sr;
>+ float beta, gamma, q;
>+ float ro = 0.001f;
>+ float sing, cosb, sinb;
>+ float FL, FR, FC, LFE, SL, SR, SC;
>+ float mid, side;
>+ float g, ig2, gg;
>+
>+ nwl = wl + ro * y * (fl - wl * y);
>+ nwr = wr + ro * y * (fr - wr * y);
>+ fl = src[n * 2 ];
>+ fr = src[n * 2 + 1];
>+ y = nwl * fl + nwr * fr;
>+ q = nwr * fl - nwl * fr;
>+ pk = pk + ro * (2.f * fl * fr - (fl * fl + fr * fr) * pk);
>+ av_assert0(pk >= -1.f && pk <= 1.f);
>+ beta = asinf(1.f - fmaxf(pk, 0.f));
>+ gamma = acosf(1.f + fminf(pk, 0.f));
>+ cosb = cosf(beta);
>+ sinb = sinf(beta);
>+ sing = sinf(gamma);
>+ g = cosb * cosb;
>+ gg = sinb * sinb;
>+ ig2 = (1.f - g) * (1.f - g);
>+
>+ clr = (nwr * nwr - nwl * nwl) * cosb;
>+ slr = (nwr * nwr - nwl * nwl) * sing;
>+ cl = clr < 0 ? -clr : 0.f;
>+ cr = clr >= 0 ? clr : 0.f;
>+ sl = slr < 0 ? -slr : 0.f;
>+ sr = slr >= 0 ? slr : 0.f;
>+ mid = 0.5f * (fl + fr);
>+ side = 0.5f * (fr - fl);
>+ FC = cosb * nwl * nwr * y + gg * mid * center;
>+ FL = cl * y + g * nwl * q + ig2 * fl;
>+ FR = cr * y + g * nwr * q + ig2 * fr;
>+ SC = sinb * q;
>+ SL = SC + sl * side * ambience;
>+ SR = SC + sr * side * ambience;
>+ LFE = FC * b0 + z0;
>+ z0 = b1 * FC + z1 + a1 * LFE;
>+ z1 = b2 * FC + a2 * LFE;
>+
>+ switch (s->upmix) {
>+ case UPMIX_2_1:
>+ dst[n * 3 + 0] = fl;
>+ dst[n * 3 + 1] = fr;
>+ dst[n * 3 + 2] = LFE;
>+ break;
>+ case UPMIX_3_0:
>+ dst[n * 3 + 0] = fl;
>+ dst[n * 3 + 1] = fr;
>+ dst[n * 3 + 2] = FC;
>+ break;
>+ case UPMIX_3_1:
>+ dst[n * 4 + 0] = fl;
>+ dst[n * 4 + 1] = fr;
>+ dst[n * 4 + 2] = FC;
>+ dst[n * 4 + 3] = LFE;
>+ break;
>+ case UPMIX_4_0:
>+ dst[n * 4 + 0] = FL;
>+ dst[n * 4 + 1] = FR;
>+ dst[n * 4 + 2] = FC;
>+ dst[n * 4 + 3] = SC;
>+ break;
>+ case UPMIX_4_1:
>+ dst[n * 5 + 0] = FL;
>+ dst[n * 5 + 1] = FR;
>+ dst[n * 5 + 2] = FC;
>+ dst[n * 5 + 3] = LFE;
>+ dst[n * 5 + 4] = SC;
>+ break;
>+ case UPMIX_5_0:
>+ dst[n * 5 + 0] = FL;
>+ dst[n * 5 + 1] = FR;
>+ dst[n * 5 + 2] = FC;
>+ dst[n * 5 + 3] = SL;
>+ dst[n * 5 + 4] = SR;
>+ break;
>+ case UPMIX_5_1:
>+ dst[n * 6 + 0] = FL;
>+ dst[n * 6 + 1] = FR;
>+ dst[n * 6 + 2] = FC;
>+ dst[n * 6 + 3] = LFE;
>+ dst[n * 6 + 4] = SL;
>+ dst[n * 6 + 5] = SR;
>+ break;
>+ default:
>+ av_assert0(0);
>+ }
>+
>+ wl = nwl;
>+ wr = nwr;
>+ }
>+
>+ s->pk = pk;
>+ s->y = y;
>+
>+ s->fl = fl;
>+ s->fr = fr;
>+
>+ s->wl = wl;
>+ s->wr = wr;
>+
>+ s->z[0] = z0;
>+ s->z[1] = z1;
>+}
>+
>+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
>+{
>+ AVFilterContext *ctx = inlink->dst;
>+ AVFilterLink *outlink = ctx->outputs[0];
>+ AVFrame *out;
>+
>+ out = ff_get_audio_buffer(outlink, in->nb_samples);
>+ if (!out) {
>+ av_frame_free(&in);
>+ return AVERROR(ENOMEM);
>+ }
>+ av_frame_copy_props(out, in);
>+
>+ upmix(ctx, out, in);
>+
>+ av_frame_free(&in);
>+ return ff_filter_frame(outlink, out);
>+}
>+
>+static int config_output(AVFilterLink *outlink)
>+{
>+ AVFilterContext *ctx = outlink->src;
>+ StereoUpmixContext *s = ctx->priv;
>+ double w0 = 2. * M_PI * 120. / outlink->sample_rate;
>+ double alpha = sin(w0) / 2. * sqrt(2.);
>+ double a0 = 1. + alpha;
>+
>+ s->a[0] = 2. * cos(w0);
>+ s->a[1] = alpha - 1.;
>+ s->b[0] = (1. - cos(w0)) / 2.;
>+ s->b[1] = 1. - cos(w0);
>+ s->b[2] = (1. - cos(w0)) / 2.;
>+
>+ s->a[0] /= a0;
>+ s->a[1] /= a0;
>+ s->b[0] /= a0;
>+ s->b[1] /= a0;
>+ s->b[2] /= a0;
>+
>+ return 0;
>+}
>+
>+static const AVFilterPad inputs[] = {
>+ {
>+ .name = "default",
>+ .type = AVMEDIA_TYPE_AUDIO,
>+ .filter_frame = filter_frame,
>+ },
>+ { NULL }
>+};
>+
>+static const AVFilterPad outputs[] = {
>+ {
>+ .name = "default",
>+ .type = AVMEDIA_TYPE_AUDIO,
>+ .config_props = config_output,
>+ },
>+ { NULL }
>+};
>+
>+AVFilter ff_af_stereoupmix = {
>+ .name = "stereoupmix",
>+ .description = NULL_IF_CONFIG_SMALL("Upmix stereo audio."),
>+ .query_formats = query_formats,
>+ .priv_size = sizeof(StereoUpmixContext),
>+ .priv_class = &stereoupmix_class,
>+ .init = init,
>+ .inputs = inputs,
>+ .outputs = outputs,
>+ .process_command = ff_filter_process_command,
>+};
>diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
>index 16e1f08a29..bb96ca88ab 100644
>--- a/libavfilter/allfilters.c
>+++ b/libavfilter/allfilters.c
>@@ -139,6 +139,7 @@ extern AVFilter ff_af_silenceremove;
> extern AVFilter ff_af_sofalizer;
> extern AVFilter ff_af_speechnorm;
> extern AVFilter ff_af_stereotools;
>+extern AVFilter ff_af_stereoupmix;
> extern AVFilter ff_af_stereowiden;
> extern AVFilter ff_af_superequalizer;
> extern AVFilter ff_af_surround;
--
Best regards,
Lingjiang Fang
More information about the ffmpeg-devel
mailing list