Commit 1f2bacc8 authored by Michael Niedermayer's avatar Michael Niedermayer

Merge commit 'e7dfaf16'

* commit 'e7dfaf16':
  libavfilter: example audio filtering program

Conflicts:
	.gitignore
	configure
	doc/Makefile
Merged-by: 's avatarMichael Niedermayer <michaelni@gmx.at>
parents 9e329185 e7dfaf16
......@@ -38,6 +38,7 @@
/doc/examples/avio_reading
/doc/examples/avcodec
/doc/examples/demuxing_decoding
/doc/examples/filter_audio
/doc/examples/filtering_audio
/doc/examples/filtering_video
/doc/examples/metadata
......
......@@ -1248,6 +1248,7 @@ EXAMPLE_LIST="
avio_reading_example
avcodec_example
demuxing_decoding_example
filter_audio_example
filtering_audio_example
filtering_video_example
metadata_example
......@@ -2409,6 +2410,7 @@ zmq_filter_deps="libzmq"
avio_reading="avformat avcodec avutil"
avcodec_example_deps="avcodec avutil"
demuxing_decoding_example_deps="avcodec avformat avutil"
filter_audio_example_deps="avfilter avutil"
filtering_audio_example_deps="avfilter avcodec avformat avutil"
filtering_video_example_deps="avfilter avcodec avformat avutil"
metadata_example_deps="avformat avutil"
......
......@@ -39,6 +39,7 @@ DOCS = $(DOCS-yes)
DOC_EXAMPLES-$(CONFIG_AVIO_READING_EXAMPLE) += avio_reading
DOC_EXAMPLES-$(CONFIG_AVCODEC_EXAMPLE) += avcodec
DOC_EXAMPLES-$(CONFIG_DEMUXING_DECODING_EXAMPLE) += demuxing_decoding
DOC_EXAMPLES-$(CONFIG_FILTER_AUDIO_EXAMPLE) += filter_audio
DOC_EXAMPLES-$(CONFIG_FILTERING_AUDIO_EXAMPLE) += filtering_audio
DOC_EXAMPLES-$(CONFIG_FILTERING_VIDEO_EXAMPLE) += filtering_video
DOC_EXAMPLES-$(CONFIG_METADATA_EXAMPLE) += metadata
......
/*
* copyright (c) 2013 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
* libavfilter API usage example.
*
* @example filter_audio.c
* This example will generate a sine wave audio,
* pass it through a simple filter chain, and then compute the MD5 checksum of
* the output data.
*
* The filter chain it uses is:
* (input) -> abuffer -> volume -> aformat -> abuffersink -> (output)
*
* abuffer: This provides the endpoint where you can feed the decoded samples.
* volume: In this example we hardcode it to 0.90.
* aformat: This converts the samples to the samplefreq, channel layout,
* and sample format required by the audio device.
* abuffersink: This provides the endpoint where you can read the samples after
* they have passed through the filter chain.
*/
#include <inttypes.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include "libavutil/channel_layout.h"
#include "libavutil/md5.h"
#include "libavutil/opt.h"
#include "libavutil/samplefmt.h"
#include "libavfilter/avfilter.h"
#include "libavfilter/buffersink.h"
#include "libavfilter/buffersrc.h"
#define INPUT_SAMPLERATE 48000
#define INPUT_FORMAT AV_SAMPLE_FMT_FLTP
#define INPUT_CHANNEL_LAYOUT AV_CH_LAYOUT_5POINT0
#define VOLUME_VAL 0.90
static int init_filter_graph(AVFilterGraph **graph, AVFilterContext **src,
AVFilterContext **sink)
{
AVFilterGraph *filter_graph;
AVFilterContext *abuffer_ctx;
AVFilter *abuffer;
AVFilterContext *volume_ctx;
AVFilter *volume;
AVFilterContext *aformat_ctx;
AVFilter *aformat;
AVFilterContext *abuffersink_ctx;
AVFilter *abuffersink;
AVDictionary *options_dict = NULL;
uint8_t options_str[1024];
uint8_t ch_layout[64];
int err;
/* Create a new filtergraph, which will contain all the filters. */
filter_graph = avfilter_graph_alloc();
if (!filter_graph) {
fprintf(stderr, "Unable to create filter graph.\n");
return AVERROR(ENOMEM);
}
/* Create the abuffer filter;
* it will be used for feeding the data into the graph. */
abuffer = avfilter_get_by_name("abuffer");
if (!abuffer) {
fprintf(stderr, "Could not find the abuffer filter.\n");
return AVERROR_FILTER_NOT_FOUND;
}
abuffer_ctx = avfilter_graph_alloc_filter(filter_graph, abuffer, "src");
if (!abuffer_ctx) {
fprintf(stderr, "Could not allocate the abuffer instance.\n");
return AVERROR(ENOMEM);
}
/* Set the filter options through the AVOptions API. */
av_get_channel_layout_string(ch_layout, sizeof(ch_layout), 0, INPUT_CHANNEL_LAYOUT);
av_opt_set (abuffer_ctx, "channel_layout", ch_layout, AV_OPT_SEARCH_CHILDREN);
av_opt_set (abuffer_ctx, "sample_fmt", av_get_sample_fmt_name(INPUT_FORMAT), AV_OPT_SEARCH_CHILDREN);
av_opt_set_q (abuffer_ctx, "time_base", (AVRational){ 1, INPUT_SAMPLERATE }, AV_OPT_SEARCH_CHILDREN);
av_opt_set_int(abuffer_ctx, "sample_rate", INPUT_SAMPLERATE, AV_OPT_SEARCH_CHILDREN);
/* Now initialize the filter; we pass NULL options, since we have already
* set all the options above. */
err = avfilter_init_str(abuffer_ctx, NULL);
if (err < 0) {
fprintf(stderr, "Could not initialize the abuffer filter.\n");
return err;
}
/* Create volume filter. */
volume = avfilter_get_by_name("volume");
if (!volume) {
fprintf(stderr, "Could not find the volume filter.\n");
return AVERROR_FILTER_NOT_FOUND;
}
volume_ctx = avfilter_graph_alloc_filter(filter_graph, volume, "volume");
if (!volume_ctx) {
fprintf(stderr, "Could not allocate the volume instance.\n");
return AVERROR(ENOMEM);
}
/* A different way of passing the options is as key/value pairs in a
* dictionary. */
av_dict_set(&options_dict, "volume", AV_STRINGIFY(VOLUME_VAL), 0);
err = avfilter_init_dict(volume_ctx, &options_dict);
av_dict_free(&options_dict);
if (err < 0) {
fprintf(stderr, "Could not initialize the volume filter.\n");
return err;
}
/* Create the aformat filter;
* it ensures that the output is of the format we want. */
aformat = avfilter_get_by_name("aformat");
if (!aformat) {
fprintf(stderr, "Could not find the aformat filter.\n");
return AVERROR_FILTER_NOT_FOUND;
}
aformat_ctx = avfilter_graph_alloc_filter(filter_graph, aformat, "aformat");
if (!aformat_ctx) {
fprintf(stderr, "Could not allocate the aformat instance.\n");
return AVERROR(ENOMEM);
}
/* A third way of passing the options is in a string of the form
* key1=value1:key2=value2.... */
snprintf(options_str, sizeof(options_str),
"sample_fmts=%s:sample_rates=%d:channel_layouts=0x%"PRIx64,
av_get_sample_fmt_name(AV_SAMPLE_FMT_S16), 44100,
(uint64_t)AV_CH_LAYOUT_STEREO);
err = avfilter_init_str(aformat_ctx, options_str);
if (err < 0) {
av_log(NULL, AV_LOG_ERROR, "Could not initialize the aformat filter.\n");
return err;
}
/* Finally create the abuffersink filter;
* it will be used to get the filtered data out of the graph. */
abuffersink = avfilter_get_by_name("abuffersink");
if (!abuffersink) {
fprintf(stderr, "Could not find the abuffersink filter.\n");
return AVERROR_FILTER_NOT_FOUND;
}
abuffersink_ctx = avfilter_graph_alloc_filter(filter_graph, abuffersink, "sink");
if (!abuffersink_ctx) {
fprintf(stderr, "Could not allocate the abuffersink instance.\n");
return AVERROR(ENOMEM);
}
/* This filter takes no options. */
err = avfilter_init_str(abuffersink_ctx, NULL);
if (err < 0) {
fprintf(stderr, "Could not initialize the abuffersink instance.\n");
return err;
}
/* Connect the filters;
* in this simple case the filters just form a linear chain. */
err = avfilter_link(abuffer_ctx, 0, volume_ctx, 0);
if (err >= 0)
err = avfilter_link(volume_ctx, 0, aformat_ctx, 0);
if (err >= 0)
err = avfilter_link(aformat_ctx, 0, abuffersink_ctx, 0);
if (err < 0) {
fprintf(stderr, "Error connecting filters\n");
return err;
}
/* Configure the graph. */
err = avfilter_graph_config(filter_graph, NULL);
if (err < 0) {
av_log(NULL, AV_LOG_ERROR, "Error configuring the filter graph\n");
return err;
}
*graph = filter_graph;
*src = abuffer_ctx;
*sink = abuffersink_ctx;
return 0;
}
/* Do something useful with the filtered data: this simple
* example just prints the MD5 checksum of each plane to stdout. */
static int process_output(struct AVMD5 *md5, AVFrame *frame)
{
int planar = av_sample_fmt_is_planar(frame->format);
int channels = av_get_channel_layout_nb_channels(frame->channel_layout);
int planes = planar ? channels : 1;
int bps = av_get_bytes_per_sample(frame->format);
int plane_size = bps * frame->nb_samples * (planar ? 1 : channels);
int i, j;
for (i = 0; i < planes; i++) {
uint8_t checksum[16];
av_md5_init(md5);
av_md5_sum(checksum, frame->extended_data[i], plane_size);
fprintf(stdout, "plane %d: 0x", i);
for (j = 0; j < sizeof(checksum); j++)
fprintf(stdout, "%02X", checksum[j]);
fprintf(stdout, "\n");
}
fprintf(stdout, "\n");
return 0;
}
/* Construct a frame of audio data to be filtered;
* this simple example just synthesizes a sine wave. */
static int get_input(AVFrame *frame, int frame_num)
{
int err, i, j;
#define FRAME_SIZE 1024
/* Set up the frame properties and allocate the buffer for the data. */
frame->sample_rate = INPUT_SAMPLERATE;
frame->format = INPUT_FORMAT;
frame->channel_layout = INPUT_CHANNEL_LAYOUT;
frame->nb_samples = FRAME_SIZE;
frame->pts = frame_num * FRAME_SIZE;
err = av_frame_get_buffer(frame, 0);
if (err < 0)
return err;
/* Fill the data for each channel. */
for (i = 0; i < 5; i++) {
float *data = (float*)frame->extended_data[i];
for (j = 0; j < frame->nb_samples; j++)
data[j] = sin(2 * M_PI * (frame_num + j) * (i + 1) / FRAME_SIZE);
}
return 0;
}
int main(int argc, char *argv[])
{
struct AVMD5 *md5;
AVFilterGraph *graph;
AVFilterContext *src, *sink;
AVFrame *frame;
uint8_t errstr[1024];
float duration;
int err, nb_frames, i;
if (argc < 2) {
fprintf(stderr, "Usage: %s <duration>\n", argv[0]);
return 1;
}
duration = atof(argv[1]);
nb_frames = duration * INPUT_SAMPLERATE / FRAME_SIZE;
if (nb_frames <= 0) {
fprintf(stderr, "Invalid duration: %s\n", argv[1]);
return 1;
}
avfilter_register_all();
/* Allocate the frame we will be using to store the data. */
frame = av_frame_alloc();
if (!frame) {
fprintf(stderr, "Error allocating the frame\n");
return 1;
}
md5 = av_md5_alloc();
if (!md5) {
fprintf(stderr, "Error allocating the MD5 context\n");
return 1;
}
/* Set up the filtergraph. */
err = init_filter_graph(&graph, &src, &sink);
if (err < 0) {
fprintf(stderr, "Unable to init filter graph:");
goto fail;
}
/* the main filtering loop */
for (i = 0; i < nb_frames; i++) {
/* get an input frame to be filtered */
err = get_input(frame, i);
if (err < 0) {
fprintf(stderr, "Error generating input frame:");
goto fail;
}
/* Send the frame to the input of the filtergraph. */
err = av_buffersrc_add_frame(src, frame);
if (err < 0) {
av_frame_unref(frame);
fprintf(stderr, "Error submitting the frame to the filtergraph:");
goto fail;
}
/* Get all the filtered output that is available. */
while ((err = av_buffersink_get_frame(sink, frame)) >= 0) {
/* now do something with our filtered frame */
err = process_output(md5, frame);
if (err < 0) {
fprintf(stderr, "Error processing the filtered frame:");
goto fail;
}
av_frame_unref(frame);
}
if (err == AVERROR(EAGAIN)) {
/* Need to feed more frames in. */
continue;
} else if (err == AVERROR_EOF) {
/* Nothing more to do, finish. */
break;
} else if (err < 0) {
/* An error occurred. */
fprintf(stderr, "Error filtering the data:");
goto fail;
}
}
avfilter_graph_free(&graph);
av_frame_free(&frame);
av_freep(&md5);
return 0;
fail:
av_strerror(err, errstr, sizeof(errstr));
fprintf(stderr, "%s\n", errstr);
return 1;
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment