<div dir="ltr">Hi,<br><br>I have written a small program to convert webm (vorbis) audio to aac format, using FFmpeg libraries - C++ (on Windows using 32 bit Zeranoe FFmpeg builds). After writing this program, I find it is sometimes converting files as per expectation, and at other times, results in larger duration files, and audio playback is broken/awkward as well. <br>
<br>This code appears to be working fine for mp3, which also uses FLTP format (same as vorbis), so technically both look similar.<br><br>Please see below sample code I am using:<br><br> ////////////////////////////////////////////////<br>
#include "stdafx.h"<br> <br> #include <iostream><br> #include <fstream><br> <br> #include <string><br> #include <vector><br> #include <map><br> <br> #include <deque><br>
#include <queue><br> <br> #include <math.h><br> #include <stdlib.h><br> #include <stdio.h><br> #include <conio.h><br> <br> extern "C"<br> {<br> #include "libavcodec/avcodec.h"<br>
#include "libavformat/avformat.h"<br> #include "libavdevice/avdevice.h"<br> #include "libswscale/swscale.h"<br> #include "libavutil/dict.h"<br> #include "libavutil/error.h"<br>
#include "libavutil/opt.h"<br> #include <libavutil/fifo.h><br> #include <libavutil/imgutils.h><br> #include <libavutil/samplefmt.h><br> #include <libswresample/swresample.h><br>
}<br> <br> AVFormatContext* fmt_ctx= NULL;<br> int audio_stream_index = -1;<br> AVCodecContext * codec_ctx_audio = NULL;<br> AVCodec* codec_audio = NULL;<br> AVFrame* decoded_frame = NULL;<br>
uint8_t** audio_dst_data = NULL;<br> int got_frame = 0;<br> int audiobufsize = 0;<br> AVPacket input_packet;<br> int audio_dst_linesize = 0;<br>
int audio_dst_bufsize = 0;<br> SwrContext * swr = NULL;<br> <br> AVOutputFormat * output_format = NULL ;<br> AVFormatContext * output_fmt_ctx= NULL;<br> AVStream * audio_st = NULL;<br>
AVCodec * audio_codec = NULL;<br> double audio_pts = 0.0;<br> AVFrame * out_frame = avcodec_alloc_frame();<br> <br> int audio_input_frame_size = 0;<br>
<br> uint8_t * audio_data_buf = NULL;<br> uint8_t * audio_out = NULL;<br> int audio_bit_rate;<br> int audio_sample_rate;<br> int audio_channels;<br>
<br> int decode_packet();<br> int open_audio_input(char* src_filename);<br> int decode_frame();<br> <br> int open_encoder(char* output_filename);<br> AVStream *add_audio_stream(AVFormatContext *oc, AVCodec **codec,<br>
enum AVCodecID codec_id);<br> int open_audio(AVFormatContext *oc, AVCodec *codec, AVStream *st);<br> void close_audio(AVFormatContext *oc, AVStream *st);<br> void write_audio_frame(uint8_t ** audio_src_data, int audio_src_bufsize);<br>
<br> int open_audio_input(char* src_filename)<br> {<br> int i =0;<br> /* open input file, and allocate format context */<br> if (avformat_open_input(&fmt_ctx, src_filename, NULL, NULL) < 0)<br>
{<br> fprintf(stderr, "Could not open source file %s\n", src_filename);<br> exit(1);<br> }<br> <br> // Retrieve stream information<br> if(avformat_find_stream_info(fmt_ctx, NULL)<0)<br>
return -1; // Couldn't find stream information<br> <br> // Dump information about file onto standard error<br> av_dump_format(fmt_ctx, 0, src_filename, 0);<br> <br> // Find the first video stream<br>
for(i=0; i<fmt_ctx->nb_streams; i++)<br> {<br> if(fmt_ctx->streams[i]->codec->codec_type==AVMEDIA_TYPE_AUDIO)<br> {<br> audio_stream_index=i;<br> break;<br>
}<br> }<br> if ( audio_stream_index != -1 )<br> {<br> // Get a pointer to the codec context for the audio stream<br> codec_ctx_audio=fmt_ctx->streams[audio_stream_index]->codec;<br>
<br> // Find the decoder for the video stream<br> codec_audio=avcodec_find_decoder(codec_ctx_audio->codec_id);<br> if(codec_audio==NULL) {<br> fprintf(stderr, "Unsupported audio codec!\n");<br>
return -1; // Codec not found<br> }<br> <br> // Open codec<br> AVDictionary *codecDictOptions = NULL;<br> if(avcodec_open2(codec_ctx_audio, codec_audio, &codecDictOptions)<0)<br>
return -1; // Could not open codec<br> <br> // Set up SWR context once you've got codec information<br> swr = swr_alloc();<br> av_opt_set_int(swr, "in_channel_layout", codec_ctx_audio->channel_layout, 0);<br>
av_opt_set_int(swr, "out_channel_layout", codec_ctx_audio->channel_layout, 0);<br> av_opt_set_int(swr, "in_sample_rate", codec_ctx_audio->sample_rate, 0);<br> av_opt_set_int(swr, "out_sample_rate", codec_ctx_audio->sample_rate, 0);<br>
av_opt_set_sample_fmt(swr, "in_sample_fmt", codec_ctx_audio->sample_fmt, 0);<br> av_opt_set_sample_fmt(swr, "out_sample_fmt", AV_SAMPLE_FMT_S16, 0);<br> swr_init(swr);<br>
<br> // Allocate audio frame<br> if ( decoded_frame == NULL ) decoded_frame = avcodec_alloc_frame();<br> int nb_planes = 0;<br> AVStream* audio_stream = fmt_ctx->streams[audio_stream_index];<br>
nb_planes = av_sample_fmt_is_planar(codec_ctx_audio->sample_fmt) ? codec_ctx_audio->channels : 1;<br> int tempSize = sizeof(uint8_t *) * nb_planes;<br> audio_dst_data = (uint8_t**)av_mallocz(tempSize);<br>
if (!audio_dst_data)<br> {<br> fprintf(stderr, "Could not allocate audio data buffers\n");<br> }<br> else<br> {<br> for ( int i = 0 ; i < nb_planes ; i ++ )<br>
{<br> audio_dst_data[i] = NULL;<br> }<br> }<br> }<br> }<br> <br> <br> int decode_frame()<br> {<br> int rv = 0;<br> got_frame = 0;<br>
if ( fmt_ctx == NULL )<br> {<br> return rv;<br> }<br> int ret = 0;<br> audiobufsize = 0;<br> rv = av_read_frame(fmt_ctx, &input_packet);<br> if ( rv < 0 )<br>
{<br> return rv;<br> }<br> rv = decode_packet();<br> // Free the input_packet that was allocated by av_read_frame<br> av_free_packet(&input_packet);<br> return rv;<br>
}<br> <br> int decode_packet()<br> {<br> int rv = 0;<br> int ret = 0;<br> <br> //audio stream?<br> if(input_packet.stream_index == audio_stream_index)<br> {<br> /* decode audio frame */<br>
rv = avcodec_decode_audio4(codec_ctx_audio, decoded_frame, &got_frame, &input_packet);<br> if (rv < 0)<br> {<br> fprintf(stderr, "Error decoding audio frame\n");<br>
//return ret;<br> }<br> else<br> {<br> if (got_frame)<br> {<br> if ( audio_dst_data[0] == NULL )<br> {<br>
ret = av_samples_alloc(audio_dst_data, &audio_dst_linesize, decoded_frame->channels,<br> decoded_frame->nb_samples, (AVSampleFormat)decoded_frame->format, 1);<br>
if (ret < 0)<br> {<br> fprintf(stderr, "Could not allocate audio buffer\n");<br> return AVERROR(ENOMEM);<br>
}<br> /* TODO: extend return code of the av_samples_* functions so that this call is not needed */<br> audio_dst_bufsize = av_samples_get_buffer_size(NULL, audio_st->codec->channels,<br>
decoded_frame->nb_samples, (AVSampleFormat)decoded_frame->format, 1);<br> <br> //int16_t* outputBuffer = ...;<br> swr_convert(swr, audio_dst_data, out_frame->nb_samples,<br>
(const uint8_t **)(decoded_frame->data), decoded_frame->nb_samples);<br> //swr_convert( swr, audio_dst_data, out_frame->nb_samples, (const uint8_t**) decoded_frame->extended_data, decoded_frame->nb_samples );<br>
}<br> /* copy audio data to destination buffer:<br> * this is required since rawaudio expects non aligned data */<br> //av_samples_copy(audio_dst_data, decoded_frame->data, 0, 0,<br>
// decoded_frame->nb_samples, decoded_frame->channels, (AVSampleFormat)decoded_frame->format);<br> }<br> }<br> }<br> return rv;<br> }<br> <br> <br>
int open_encoder(char* output_filename )<br> {<br> int rv = 0;<br> <br> /* allocate the output media context */<br> AVOutputFormat *opfmt = NULL;<br> <br> avformat_alloc_output_context2(&output_fmt_ctx, opfmt, NULL, output_filename);<br>
if (!output_fmt_ctx) {<br> printf("Could not deduce output format from file extension: using MPEG.\n");<br> avformat_alloc_output_context2(&output_fmt_ctx, NULL, "mpeg", output_filename);<br>
}<br> if (!output_fmt_ctx) {<br> rv = -1;<br> }<br> else<br> {<br> output_format = output_fmt_ctx->oformat;<br> }<br> <br> /* Add the audio stream using the default format codecs<br>
* and initialize the codecs. */<br> audio_st = NULL;<br> <br> if ( output_fmt_ctx )<br> {<br> if (output_format->audio_codec != AV_CODEC_ID_NONE)<br> {<br> audio_st = add_audio_stream(output_fmt_ctx, &audio_codec, output_format->audio_codec);<br>
}<br> <br> /* Now that all the parameters are set, we can open the audio and<br> * video codecs and allocate the necessary encode buffers. */<br> if (audio_st)<br> {<br>
rv = open_audio(output_fmt_ctx, audio_codec, audio_st);<br> if ( rv < 0 ) return rv;<br> }<br> <br> av_dump_format(output_fmt_ctx, 0, output_filename, 1);<br> /* open the output file, if needed */<br>
if (!(output_format->flags & AVFMT_NOFILE))<br> {<br> if (avio_open(&output_fmt_ctx->pb, output_filename, AVIO_FLAG_WRITE) < 0) {<br> fprintf(stderr, "Could not open '%s'\n", output_filename);<br>
rv = -1;<br> }<br> else<br> {<br> /* Write the stream header, if any. */<br> if (avformat_write_header(output_fmt_ctx, NULL) < 0)<br>
{<br> fprintf(stderr, "Error occurred when opening output file\n");<br> rv = -1;<br> }<br> }<br> }<br>
}<br> <br> return rv;<br> }<br> <br> AVStream *add_audio_stream(AVFormatContext *oc, AVCodec **codec,<br> enum AVCodecID codec_id)<br> {<br> AVCodecContext *c;<br> AVStream *st;<br>
<br> /* find the audio encoder */<br> *codec = avcodec_find_encoder(codec_id);<br> if (!(*codec)) {<br> fprintf(stderr, "Could not find codec\n");<br> exit(1);<br> }<br>
<br> st = avformat_new_stream(oc, *codec);<br> if (!st) {<br> fprintf(stderr, "Could not allocate stream\n");<br> exit(1);<br> }<br> st->id = 1;<br> <br>
c = st->codec;<br> <br> /* put sample parameters */<br> c->sample_fmt = AV_SAMPLE_FMT_S16;<br> c->bit_rate = audio_bit_rate;<br> c->sample_rate = audio_sample_rate;<br>
c->channels = audio_channels;<br> <br> // some formats want stream headers to be separate<br> if (oc->oformat->flags & AVFMT_GLOBALHEADER)<br> c->flags |= CODEC_FLAG_GLOBAL_HEADER;<br>
<br> return st;<br> }<br> <br> int open_audio(AVFormatContext *oc, AVCodec *codec, AVStream *st)<br> {<br> int ret=0;<br> AVCodecContext *c;<br> <br> st->duration = fmt_ctx->duration;<br>
c = st->codec;<br> <br> /* open it */<br> ret = avcodec_open2(c, codec, NULL) ;<br> if ( ret < 0)<br> {<br> fprintf(stderr, "could not open codec\n");<br> return -1;<br>
//exit(1);<br> }<br> <br> if (c->codec->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE)<br> audio_input_frame_size = 10000;<br> else<br> audio_input_frame_size = c->frame_size;<br>
int tempSize = audio_input_frame_size *<br> av_get_bytes_per_sample(c->sample_fmt) *<br> c->channels;<br> return ret;<br> }<br> <br> void close_audio(AVFormatContext *oc, AVStream *st)<br>
{<br> avcodec_close(st->codec);<br> }<br> <br> void write_audio_frame(uint8_t ** audio_src_data, int audio_src_bufsize)<br> {<br> AVFormatContext *oc = output_fmt_ctx;<br> AVStream *st = audio_st;<br>
if ( oc == NULL || st == NULL ) return;<br> AVCodecContext *c;<br> AVPacket pkt = { 0 }; // data and size must be 0;<br> int got_packet;<br> <br> av_init_packet(&pkt);<br> c = st->codec;<br>
<br> out_frame->nb_samples = audio_input_frame_size;<br> int buf_size = audio_src_bufsize *<br> av_get_bytes_per_sample(c->sample_fmt) *<br> c->channels;<br> avcodec_fill_audio_frame(out_frame, c->channels, c->sample_fmt,<br>
(uint8_t *) *audio_src_data,<br> buf_size, 1);<br> avcodec_encode_audio2(c, &pkt, out_frame, &got_packet);<br> if (!got_packet)<br> {<br> }<br> else<br> {<br>
if (pkt.pts != AV_NOPTS_VALUE)<br> pkt.pts = av_rescale_q(pkt.pts, st->codec->time_base, st->time_base);<br> if (pkt.dts != AV_NOPTS_VALUE)<br> pkt.dts = av_rescale_q(pkt.dts, st->codec->time_base, st->time_base);<br>
if ( c && c->coded_frame && c->coded_frame->key_frame)<br> pkt.flags |= AV_PKT_FLAG_KEY;<br> <br> pkt.stream_index = st->index;<br> pkt.flags |= AV_PKT_FLAG_KEY;<br>
/* Write the compressed frame to the media file. */<br> if (av_interleaved_write_frame(oc, &pkt) != 0)<br> {<br> fprintf(stderr, "Error while writing audio frame\n");<br>
exit(1);<br> }<br> }<br> av_free_packet(&pkt);<br> }<br> <br> <br> void write_delayed_frames(AVFormatContext *oc, AVStream *st)<br> {<br> AVCodecContext *c = st->codec;<br>
int got_output = 0;<br> int ret = 0;<br> AVPacket pkt;<br> pkt.data = NULL;<br> pkt.size = 0;<br> av_init_packet(&pkt);<br> int i = 0;<br> for (got_output = 1; got_output; i++)<br>
{<br> ret = avcodec_encode_audio2(c, &pkt, NULL, &got_output);<br> if (ret < 0)<br> {<br> fprintf(stderr, "error encoding frame\n");<br> exit(1);<br>
}<br> static int64_t tempPts = 0;<br> static int64_t tempDts = 0;<br> /* If size is zero, it means the image was buffered. */<br> if (got_output)<br> {<br>
if (pkt.pts != AV_NOPTS_VALUE)<br> pkt.pts = av_rescale_q(pkt.pts, st->codec->time_base, st->time_base);<br> if (pkt.dts != AV_NOPTS_VALUE)<br> pkt.dts = av_rescale_q(pkt.dts, st->codec->time_base, st->time_base);<br>
if ( c && c->coded_frame && c->coded_frame->key_frame)<br> pkt.flags |= AV_PKT_FLAG_KEY;<br> <br> pkt.stream_index = st->index;<br> /* Write the compressed frame to the media file. */<br>
ret = av_interleaved_write_frame(oc, &pkt);<br> }<br> else<br> {<br> ret = 0;<br> }<br> av_free_packet(&pkt);<br> }<br> }<br>
<br> int main(int argc, char **argv)<br> {<br> /* register all formats and codecs */<br> av_register_all();<br> avcodec_register_all();<br> avformat_network_init();<br> avdevice_register_all();<br>
int i =0;<br> int ret=0;<br> char src_filename[90] = "test_a.webm";<br> char dst_filename[90] = "output.aac";<br> open_audio_input(src_filename);<br> if ( codec_ctx_audio->bit_rate == 0 ) codec_ctx_audio->bit_rate = 112000;<br>
audio_bit_rate = codec_ctx_audio->bit_rate;<br> audio_sample_rate = codec_ctx_audio->sample_rate;<br> audio_channels = codec_ctx_audio->channels;<br> open_encoder( dst_filename );<br>
while(1)<br> {<br> int rv = decode_frame();<br> if ( rv < 0 )<br> {<br> break;<br> }<br> <br> if (audio_st)<br> {<br> audio_pts = (double)audio_st->pts.val * audio_st->time_base.num /<br>
audio_st->time_base.den;<br> }<br> else<br> {<br> audio_pts = 0.0;<br> }<br> if ( codec_ctx_audio )<br> {<br> if ( got_frame)<br>
{<br> write_audio_frame( audio_dst_data, audio_dst_bufsize );<br> }<br> }<br> if ( audio_dst_data[0] )<br> {<br> av_freep(&audio_dst_data[0]);<br>
audio_dst_data[0] = NULL;<br> }<br> printf("\naudio_pts: %.3f", audio_pts);<br> }<br> while(1)<br> {<br> if ( audio_dst_data && audio_dst_data[0] )<br>
{<br> av_freep(&audio_dst_data[0]);<br> audio_dst_data[0] = NULL;<br> }<br> ret = av_samples_alloc(audio_dst_data, NULL, codec_ctx_audio->channels,<br>
decoded_frame->nb_samples, AV_SAMPLE_FMT_S16, 0);<br> ret = swr_convert(swr, audio_dst_data, out_frame->nb_samples,NULL, 0);<br> if ( ret <= 0 ) break;<br> write_audio_frame( audio_dst_data, audio_dst_bufsize );<br>
}<br> write_delayed_frames( output_fmt_ctx, audio_st );<br> av_write_trailer(output_fmt_ctx);<br> close_audio( output_fmt_ctx, audio_st);<br> swr_free(&swr);<br> avcodec_free_frame(&out_frame);<br>
getch();<br> return 0;<br> }<br><br>"test_a.webm" input file results in longer duration (40 second output), and if I change it to "jet.webm", it is converted fine.<br><br>Both input files are approximately 18 second duration.<br>
<br>For reference, these files can be downloaded from links below:<br><br><a href="http://www.filedropper.com/testa">http://www.filedropper.com/testa</a> ,<br><a href="http://www.filedropper.com/jet">http://www.filedropper.com/jet</a><br>
<br>Alternatively, they are zipped and uploaded elsewhere as well:<br><br><a href="http://www.files.com/shared/52c3eefe990ea/test_audio_files.zip">http://www.files.com/shared/52c3eefe990ea/test_audio_files.zip</a><br><br>
Could someone kindly guide on what I am doing wrong here?<br><br>Thanks in advance...<br><br>p.s. These files are taken/extracted from different online sources/demos; also posted on SO: <a href="http://stackoverflow.com/questions/20867959/ffmpeg-library-webm-vorbis-audio-to-aac-conversion">http://stackoverflow.com/questions/20867959/ffmpeg-library-webm-vorbis-audio-to-aac-conversion</a><br>
</div>