RepoMirrors/ffmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2024-12-27 01:42:20 +00:00
Code Issues Projects Releases Wiki Activity

avfilter: add showspectrumpic filter

Browse Source 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
This commit is contained in:
Paul B Mahol 2015-12-31 20:45:37 +01:00
parent 7f7a9dd782
commit 72280d1c6c
6 changed files with 370 additions and 46 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
Changelog
View File

@ -50,6 +50,7 @@ version <next>:
- VAAPI VP9 hwaccel
- audio high-order multiband parametric equalizer
- automatic bitstream filtering
- showspectrumpic filter
version 2.8:

106
doc/filters.texi
View File

@ -14704,6 +14704,112 @@ ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
@end example
@end itemize
@section showspectrumpic
Convert input audio to a single video frame, representing the audio frequency
spectrum.
The filter accepts the following options:
@table @option
@item size, s
Specify the video size for the output. For the syntax of this option, check the
@ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
Default value is @code{4096x2048}.
@item mode
Specify display mode.
It accepts the following values:
@table @samp
@item combined
all channels are displayed in the same row
@item separate
all channels are displayed in separate rows
@end table
Default value is @samp{combined}.
@item color
Specify display color mode.
It accepts the following values:
@table @samp
@item channel
each channel is displayed in a separate color
@item intensity
each channel is displayed using the same color scheme
@item rainbow
each channel is displayed using the rainbow color scheme
@item moreland
each channel is displayed using the moreland color scheme
@item nebulae
each channel is displayed using the nebulae color scheme
@item fire
each channel is displayed using the fire color scheme
@end table
Default value is @samp{intensity}.
@item scale
Specify scale used for calculating intensity color values.
It accepts the following values:
@table @samp
@item lin
linear
@item sqrt
square root, default
@item cbrt
cubic root
@item log
logarithmic
@end table
Default value is @samp{log}.
@item saturation
Set saturation modifier for displayed colors. Negative values provide
alternative color scheme. @code{0} is no saturation at all.
Saturation must be in [-10.0, 10.0] range.
Default value is @code{1}.
@item win_func
Set window function.
It accepts the following values:
@table @samp
@item rect
@item bartlett
@item hann
@item hanning
@item hamming
@item blackman
@item welch
@item flattop
@item bharris
@item bnuttall
@item bhann
@item sine
@item nuttall
@item lanczos
@item gauss
@end table
Default value is @code{hann}.
@item orientation
Set orientation of time vs frequency axis. Can be @code{vertical} or
@code{horizontal}. Default is @code{vertical}.
@end table
@subsection Examples
@itemize
@item
Extract an audio spectrogram of a whole audio track
in a 1024x1024 picture using @command{ffmpeg}:
@example
ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
@end example
@end itemize
@section showvolume
Convert input audio volume to a video output.

1
libavfilter/Makefile
View File

@ -286,6 +286,7 @@ OBJS-$(CONFIG_CONCAT_FILTER) += avf_concat.o
OBJS-$(CONFIG_SHOWCQT_FILTER) += avf_showcqt.o lswsutils.o lavfutils.o
OBJS-$(CONFIG_SHOWFREQS_FILTER) += avf_showfreqs.o window_func.o
OBJS-$(CONFIG_SHOWSPECTRUM_FILTER) += avf_showspectrum.o window_func.o
OBJS-$(CONFIG_SHOWSPECTRUMPIC_FILTER) += avf_showspectrum.o window_func.o
OBJS-$(CONFIG_SHOWVOLUME_FILTER) += avf_showvolume.o
OBJS-$(CONFIG_SHOWWAVES_FILTER) += avf_showwaves.o
OBJS-$(CONFIG_SHOWWAVESPIC_FILTER) += avf_showwaves.o

1
libavfilter/allfilters.c
View File

@ -306,6 +306,7 @@ void avfilter_register_all(void)
REGISTER_FILTER(SHOWCQT, showcqt, avf);
REGISTER_FILTER(SHOWFREQS, showfreqs, avf);
REGISTER_FILTER(SHOWSPECTRUM, showspectrum, avf);
REGISTER_FILTER(SHOWSPECTRUMPIC, showspectrumpic, avf);
REGISTER_FILTER(SHOWVOLUME, showvolume, avf);
REGISTER_FILTER(SHOWWAVES, showwaves, avf);
REGISTER_FILTER(SHOWWAVESPIC, showwavespic, avf);

305
libavfilter/avf_showspectrum.c
View File

@ -63,6 +63,7 @@ typedef struct {
int rdft_bits; ///< number of bits (RDFT window size = 1<<rdft_bits)
FFTSample **rdft_data; ///< bins holder for each (displayed) channels
float *window_func_lut; ///< Window function LUT
float **magnitudes;
int win_func;
int win_size;
double win_scale;
@ -71,6 +72,7 @@ typedef struct {
float *combine_buffer; ///< color combining buffer (3 * h items)
AVAudioFifo *fifo;
int64_t pts;
int single_pic;
} ShowSpectrumContext;
#define OFFSET(x) offsetof(ShowSpectrumContext, x)
@ -186,6 +188,9 @@ static av_cold void uninit(AVFilterContext *ctx)
av_freep(&s->rdft_data[i]);
av_freep(&s->rdft_data);
av_freep(&s->window_func_lut);
for (i = 0; i < s->nb_display_channels; i++)
av_freep(&s->magnitudes[i]);
av_freep(&s->magnitudes);
av_frame_free(&s->outpicref);
av_audio_fifo_free(s->fifo);
}
@ -229,6 +234,9 @@ static int config_output(AVFilterLink *outlink)
int i, rdft_bits, h, w;
float overlap;
if (!strcmp(ctx->filter->name, "showspectrumpic"))
s->single_pic = 1;
outlink->w = s->w;
outlink->h = s->h;
@ -267,6 +275,15 @@ static int config_output(AVFilterLink *outlink)
av_freep(&s->rdft_data);
s->nb_display_channels = inlink->channels;
s->magnitudes = av_calloc(s->nb_display_channels, sizeof(*s->magnitudes));
if (!s->magnitudes)
return AVERROR(ENOMEM);
for (i = 0; i < s->nb_display_channels; i++) {
s->magnitudes[i] = av_calloc(s->orientation == VERTICAL ? s->h : s->w, sizeof(**s->magnitudes));
if (!s->magnitudes[i])
return AVERROR(ENOMEM);
}
s->rdft_data = av_calloc(s->nb_display_channels, sizeof(*s->rdft_data));
if (!s->rdft_data)
return AVERROR(ENOMEM);
@ -340,34 +357,61 @@ static int config_output(AVFilterLink *outlink)
return 0;
}
static int request_frame(AVFilterLink *outlink)
static void run_rdft(ShowSpectrumContext *s, AVFrame *fin)
{
ShowSpectrumContext *s = outlink->src->priv;
AVFilterLink *inlink = outlink->src->inputs[0];
unsigned i;
int ret;
int ch, n;
ret = ff_request_frame(inlink);
if (ret == AVERROR_EOF && s->sliding == FULLFRAME && s->xpos > 0 &&
s->outpicref) {
if (s->orientation == VERTICAL) {
for (i = 0; i < outlink->h; i++) {
memset(s->outpicref->data[0] + i * s->outpicref->linesize[0] + s->xpos, 0, outlink->w - s->xpos);
memset(s->outpicref->data[1] + i * s->outpicref->linesize[1] + s->xpos, 128, outlink->w - s->xpos);
memset(s->outpicref->data[2] + i * s->outpicref->linesize[2] + s->xpos, 128, outlink->w - s->xpos);
}
} else {
for (i = s->xpos; i < outlink->h; i++) {
memset(s->outpicref->data[0] + i * s->outpicref->linesize[0], 0, outlink->w);
memset(s->outpicref->data[1] + i * s->outpicref->linesize[1], 128, outlink->w);
memset(s->outpicref->data[2] + i * s->outpicref->linesize[2], 128, outlink->w);
}
}
ret = ff_filter_frame(outlink, s->outpicref);
s->outpicref = NULL;
/* fill RDFT input with the number of samples available */
for (ch = 0; ch < s->nb_display_channels; ch++) {
const int16_t *p = (int16_t *)fin->extended_data[ch];
for (n = 0; n < s->win_size; n++)
s->rdft_data[ch][n] = p[n] * s->window_func_lut[n];
}
return ret;
/* run RDFT on each samples set */
for (ch = 0; ch < s->nb_display_channels; ch++)
av_rdft_calc(s->rdft, s->rdft_data[ch]);
}
#define RE(y, ch) s->rdft_data[ch][2 * (y) + 0]
#define IM(y, ch) s->rdft_data[ch][2 * (y) + 1]
#define MAGNITUDE(y, ch) hypot(RE(y, ch), IM(y, ch))
static void calc_magnitudes(ShowSpectrumContext *s)
{
int ch, y, h = s->orientation == VERTICAL ? s->h : s->w;
for (ch = 0; ch < s->nb_display_channels; ch++) {
float *magnitudes = s->magnitudes[ch];
for (y = 0; y < h; y++)
magnitudes[y] = MAGNITUDE(y, ch);
}
}
static void acalc_magnitudes(ShowSpectrumContext *s)
{
int ch, y, h = s->orientation == VERTICAL ? s->h : s->w;
for (ch = 0; ch < s->nb_display_channels; ch++) {
float *magnitudes = s->magnitudes[ch];
for (y = 0; y < h; y++)
magnitudes[y] += MAGNITUDE(y, ch);
}
}
static void scale_magnitudes(ShowSpectrumContext *s, float scale)
{
int ch, y, h = s->orientation == VERTICAL ? s->h : s->w;
for (ch = 0; ch < s->nb_display_channels; ch++) {
float *magnitudes = s->magnitudes[ch];
for (y = 0; y < h; y++)
magnitudes[y] *= scale;
}
}
static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
@ -380,27 +424,9 @@ static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
const double w = s->win_scale;
int h = s->orientation == VERTICAL ? s->channel_height : s->channel_width;
int ch, plane, n, x, y;
av_assert0(insamples->nb_samples == s->win_size);
/* fill RDFT input with the number of samples available */
for (ch = 0; ch < s->nb_display_channels; ch++) {
const int16_t *p = (int16_t *)insamples->extended_data[ch];
for (n = 0; n < s->win_size; n++)
s->rdft_data[ch][n] = p[n] * s->window_func_lut[n];
}
/* run RDFT on each samples set */
for (ch = 0; ch < s->nb_display_channels; ch++)
av_rdft_calc(s->rdft, s->rdft_data[ch]);
int ch, plane, x, y;
/* fill a new spectrum column */
#define RE(y, ch) s->rdft_data[ch][2 * (y) + 0]
#define IM(y, ch) s->rdft_data[ch][2 * (y) + 1]
#define MAGNITUDE(y, ch) hypot(RE(y, ch), IM(y, ch))
/* initialize buffer for combining to black */
if (s->orientation == VERTICAL) {
for (y = 0; y < outlink->h; y++) {
@ -417,6 +443,7 @@ static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
}
for (ch = 0; ch < s->nb_display_channels; ch++) {
float *magnitudes = s->magnitudes[ch];
float yf, uf, vf;
/* decide color range */
@ -471,7 +498,7 @@ static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
float *out = &s->combine_buffer[3 * row];
/* get magnitude */
float a = w * MAGNITUDE(y, ch);
float a = w * magnitudes[y];
/* apply scale */
switch (s->scale) {
@ -600,7 +627,7 @@ static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
s->xpos = 0;
if (s->orientation == HORIZONTAL && s->xpos >= outlink->h)
s->xpos = 0;
if (s->sliding != FULLFRAME || s->xpos == 0) {
if (!s->single_pic && (s->sliding != FULLFRAME || s->xpos == 0)) {
ret = ff_filter_frame(outlink, av_frame_clone(s->outpicref));
if (ret < 0)
return ret;
@ -609,6 +636,38 @@ static int plot_spectrum_column(AVFilterLink *inlink, AVFrame *insamples)
return s->win_size;
}
#if CONFIG_SHOWSPECTRUM_FILTER
static int request_frame(AVFilterLink *outlink)
{
ShowSpectrumContext *s = outlink->src->priv;
AVFilterLink *inlink = outlink->src->inputs[0];
unsigned i;
int ret;
ret = ff_request_frame(inlink);
if (ret == AVERROR_EOF && s->sliding == FULLFRAME && s->xpos > 0 &&
s->outpicref) {
if (s->orientation == VERTICAL) {
for (i = 0; i < outlink->h; i++) {
memset(s->outpicref->data[0] + i * s->outpicref->linesize[0] + s->xpos, 0, outlink->w - s->xpos);
memset(s->outpicref->data[1] + i * s->outpicref->linesize[1] + s->xpos, 128, outlink->w - s->xpos);
memset(s->outpicref->data[2] + i * s->outpicref->linesize[2] + s->xpos, 128, outlink->w - s->xpos);
}
} else {
for (i = s->xpos; i < outlink->h; i++) {
memset(s->outpicref->data[0] + i * s->outpicref->linesize[0], 0, outlink->w);
memset(s->outpicref->data[1] + i * s->outpicref->linesize[1], 128, outlink->w);
memset(s->outpicref->data[2] + i * s->outpicref->linesize[2], 128, outlink->w);
}
}
ret = ff_filter_frame(outlink, s->outpicref);
s->outpicref = NULL;
}
return ret;
}
static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
{
AVFilterContext *ctx = inlink->dst;
@ -631,6 +690,11 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
if (ret < 0)
goto fail;
av_assert0(fin->nb_samples == s->win_size);
run_rdft(s, fin);
calc_magnitudes(s);
ret = plot_spectrum_column(inlink, fin);
av_frame_free(&fin);
av_audio_fifo_drain(s->fifo, s->skip_samples);
@ -672,3 +736,154 @@ AVFilter ff_avf_showspectrum = {
.outputs = showspectrum_outputs,
.priv_class = &showspectrum_class,
};
#endif // CONFIG_SHOWSPECTRUM_FILTER
#if CONFIG_SHOWSPECTRUMPIC_FILTER
static const AVOption showspectrumpic_options[] = {
{ "size", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "4096x2048"}, 0, 0, FLAGS },
{ "s", "set video size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = "4096x2048"}, 0, 0, FLAGS },
{ "mode", "set channel display mode", OFFSET(mode), AV_OPT_TYPE_INT, {.i64=COMBINED}, 0, NB_MODES-1, FLAGS, "mode" },
{ "combined", "combined mode", 0, AV_OPT_TYPE_CONST, {.i64=COMBINED}, 0, 0, FLAGS, "mode" },
{ "separate", "separate mode", 0, AV_OPT_TYPE_CONST, {.i64=SEPARATE}, 0, 0, FLAGS, "mode" },
{ "color", "set channel coloring", OFFSET(color_mode), AV_OPT_TYPE_INT, {.i64=INTENSITY}, 0, NB_CLMODES-1, FLAGS, "color" },
{ "channel", "separate color for each channel", 0, AV_OPT_TYPE_CONST, {.i64=CHANNEL}, 0, 0, FLAGS, "color" },
{ "intensity", "intensity based coloring", 0, AV_OPT_TYPE_CONST, {.i64=INTENSITY}, 0, 0, FLAGS, "color" },
{ "rainbow", "rainbow based coloring", 0, AV_OPT_TYPE_CONST, {.i64=RAINBOW}, 0, 0, FLAGS, "color" },
{ "moreland", "moreland based coloring", 0, AV_OPT_TYPE_CONST, {.i64=MORELAND}, 0, 0, FLAGS, "color" },
{ "nebulae", "nebulae based coloring", 0, AV_OPT_TYPE_CONST, {.i64=NEBULAE}, 0, 0, FLAGS, "color" },
{ "fire", "fire based coloring", 0, AV_OPT_TYPE_CONST, {.i64=FIRE}, 0, 0, FLAGS, "color" },
{ "scale", "set display scale", OFFSET(scale), AV_OPT_TYPE_INT, {.i64=LOG}, 0, NB_SCALES-1, FLAGS, "scale" },
{ "sqrt", "square root", 0, AV_OPT_TYPE_CONST, {.i64=SQRT}, 0, 0, FLAGS, "scale" },
{ "cbrt", "cubic root", 0, AV_OPT_TYPE_CONST, {.i64=CBRT}, 0, 0, FLAGS, "scale" },
{ "log", "logarithmic", 0, AV_OPT_TYPE_CONST, {.i64=LOG}, 0, 0, FLAGS, "scale" },
{ "lin", "linear", 0, AV_OPT_TYPE_CONST, {.i64=LINEAR}, 0, 0, FLAGS, "scale" },
{ "saturation", "color saturation multiplier", OFFSET(saturation), AV_OPT_TYPE_FLOAT, {.dbl = 1}, -10, 10, FLAGS },
{ "win_func", "set window function", OFFSET(win_func), AV_OPT_TYPE_INT, {.i64 = WFUNC_HANNING}, 0, NB_WFUNC-1, FLAGS, "win_func" },
{ "rect", "Rectangular", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_RECT}, 0, 0, FLAGS, "win_func" },
{ "bartlett", "Bartlett", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BARTLETT}, 0, 0, FLAGS, "win_func" },
{ "hann", "Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
{ "hanning", "Hanning", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
{ "hamming", "Hamming", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HAMMING}, 0, 0, FLAGS, "win_func" },
{ "blackman", "Blackman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BLACKMAN}, 0, 0, FLAGS, "win_func" },
{ "welch", "Welch", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_WELCH}, 0, 0, FLAGS, "win_func" },
{ "flattop", "Flat-top", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_FLATTOP}, 0, 0, FLAGS, "win_func" },
{ "bharris", "Blackman-Harris", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHARRIS}, 0, 0, FLAGS, "win_func" },
{ "bnuttall", "Blackman-Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BNUTTALL}, 0, 0, FLAGS, "win_func" },
{ "bhann", "Bartlett-Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHANN}, 0, 0, FLAGS, "win_func" },
{ "sine", "Sine", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_SINE}, 0, 0, FLAGS, "win_func" },
{ "nuttall", "Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_NUTTALL}, 0, 0, FLAGS, "win_func" },
{ "lanczos", "Lanczos", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_LANCZOS}, 0, 0, FLAGS, "win_func" },
{ "gauss", "Gauss", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_GAUSS}, 0, 0, FLAGS, "win_func" },
{ "orientation", "set orientation", OFFSET(orientation), AV_OPT_TYPE_INT, {.i64=VERTICAL}, 0, NB_ORIENTATIONS-1, FLAGS, "orientation" },
{ "vertical", NULL, 0, AV_OPT_TYPE_CONST, {.i64=VERTICAL}, 0, 0, FLAGS, "orientation" },
{ "horizontal", NULL, 0, AV_OPT_TYPE_CONST, {.i64=HORIZONTAL}, 0, 0, FLAGS, "orientation" },
{ NULL }
};
AVFILTER_DEFINE_CLASS(showspectrumpic);
static int showspectrumpic_request_frame(AVFilterLink *outlink)
{
ShowSpectrumContext *s = outlink->src->priv;
AVFilterLink *inlink = outlink->src->inputs[0];
int ret;
ret = ff_request_frame(inlink);
if (ret == AVERROR_EOF && s->outpicref) {
int samples = av_audio_fifo_size(s->fifo);
int consumed = 0;
int x = 0, sz = s->orientation == VERTICAL ? s->w : s->h;
int ch, spf, spb;
AVFrame *fin;
spf = s->win_size * (samples / ((s->win_size * sz) * ceil(samples / (float)(s->win_size * sz))));
spb = (samples / (spf * sz)) * spf;
fin = ff_get_audio_buffer(inlink, s->win_size);
if (!fin)
return AVERROR(ENOMEM);
while (x < sz) {
ret = av_audio_fifo_peek(s->fifo, (void **)fin->extended_data, s->win_size);
if (ret < 0) {
av_frame_free(&fin);
return ret;
}
av_audio_fifo_drain(s->fifo, spf);
if (ret < s->win_size) {
for (ch = 0; ch < s->nb_display_channels; ch++) {
memset(fin->extended_data[ch] + ret * sizeof(int16_t), 0,
(s->win_size - ret) * sizeof(int16_t));
}
}
run_rdft(s, fin);
acalc_magnitudes(s);
consumed += spf;
if (consumed >= spb) {
int h = s->orientation == VERTICAL ? s->h : s->w;
scale_magnitudes(s, 1. / (consumed / spf));
plot_spectrum_column(inlink, fin);
consumed = 0;
x++;
for (ch = 0; ch < s->nb_display_channels; ch++)
memset(s->magnitudes[ch], 0, h * sizeof(float));
}
}
av_frame_free(&fin);
s->outpicref->pts = 0;
ret = ff_filter_frame(outlink, s->outpicref);
s->outpicref = NULL;
}
return ret;
}
static int showspectrumpic_filter_frame(AVFilterLink *inlink, AVFrame *insamples)
{
AVFilterContext *ctx = inlink->dst;
ShowSpectrumContext *s = ctx->priv;
int ret;
ret = av_audio_fifo_write(s->fifo, (void **)insamples->extended_data, insamples->nb_samples);
av_frame_free(&insamples);
return ret;
}
static const AVFilterPad showspectrumpic_inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.filter_frame = showspectrumpic_filter_frame,
},
{ NULL }
};
static const AVFilterPad showspectrumpic_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_output,
.request_frame = showspectrumpic_request_frame,
},
{ NULL }
};
AVFilter ff_avf_showspectrumpic = {
.name = "showspectrumpic",
.description = NULL_IF_CONFIG_SMALL("Convert input audio to a spectrum video output single picture."),
.uninit = uninit,
.query_formats = query_formats,
.priv_size = sizeof(ShowSpectrumContext),
.inputs = showspectrumpic_inputs,
.outputs = showspectrumpic_outputs,
.priv_class = &showspectrumpic_class,
};
#endif // CONFIG_SHOWSPECTRUMPIC_FILTER

2
libavfilter/version.h
View File

@ -30,7 +30,7 @@
#include "libavutil/version.h"
#define LIBAVFILTER_VERSION_MAJOR 6
#define LIBAVFILTER_VERSION_MINOR 22
#define LIBAVFILTER_VERSION_MINOR 23
#define LIBAVFILTER_VERSION_MICRO 100
#define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \