[Libav-user] FMP4 -> ADTS/AAC using libavformat/libavcodec
Ronak Patel
ronak2121 at yahoo.com
Fri Oct 4 15:41:19 EEST 2019
> On Oct 3, 2019, at 6:22 PM, Ronak via Libav-user <libav-user at ffmpeg.org> wrote:
>
> I am in a time crunch and would appreciate a prompt reply for this…I’m amazed that it is this complicated to append an ADTS header to MPEG4 Audio.
So it looks like I need to make an MPEG4 AudioSpecificConfig.
However I wonder what’s the point of using this muxer if I have to do that. I already specified all the parameters in the CodecParameters struct.
I’m thinking of abandoning FFmpeg to write the adts header and just do this myself.
This is just too complicated of a public api with too little documentation which provides almost no value.
There isn’t even a public helper function I can call to make that AudioSpecificConfig.
I would suggest these apis be added to Ffmpeg to make this a lot easier. If these apis already exist; I would suggest they be documented MUCH better.
>
>
>> On Oct 3, 2019, at 5:44 PM, Ronak via Libav-user <libav-user at ffmpeg.org> wrote:
>>
>> Hi,
>>
>> I’m writing a C++ program to validate the integrity of a Fragmented MP4 file containing AAC audio.
>> This program would parse the FMP4 file, read each audio packet, attachment ADTS headers, and then try to decode the AAC using libfdk_aac.
>>
>> I am using libavformat, and I am able to parse the MP4 atoms correctly, however I’m having issues figuring out the best way to attach the ADTS headers.
>>
>> My current idea is to use a libavformat mp4 demuxed and send that to an adts muxer. I’ve written all the code to do it now; but I’m now getting errors like so:
>>
>> [adts @ 0x7fae51001600] MPEG-4 AOT 0 is not allowed in ADTS
>>
>> I’m not sure what to do now.
>>
>> My code is as below:
>>
>> AudioAssetReader::AudioAssetReader(const AudioAssetReaderInput &input) throw (invalid_argument):
>> input(input), segmentsIter(input.getBeginSegmentsIter()), audioBufferSize(409600), adtsPacketBufferSize(409600) {
>> if (input.getSegments()->empty()) {
>> throw invalid_argument("A reader must read at least 1 segment of audio!");
>> }
>>
>> this->audioBuffer = new uint8_t[audioBufferSize];
>>
>> // create the main context used to parse the audio asset, segment by segment
>> bool isBufferWritable = false;
>> this->assetSegmentReaderContext = avio_alloc_context(this->audioBuffer, this->audioBufferSize, isBufferWritable, this, AudioAssetReader::readSegment, NULL, NULL);
>> this->assetParsingContext = avformat_alloc_context();
>> this->assetParsingContext->pb = this->assetSegmentReaderContext;
>> this->assetParsingContext->flags = AVFMT_FLAG_CUSTOM_IO;
>>
>> // create another context used to append ADTS headers to the parsed segments, if we need them
>> this->initializeADTSMuxerContext(input);
>> }
>>
>> void AudioAssetReader::initializeADTSMuxerContext(const AudioAssetReaderInput &input) {
>>
>> this->adtsPacketBuffer = new uint8_t[adtsPacketBufferSize];
>> this->adtsPacket = av_packet_alloc();
>>
>> // create another context used to append ADTS headers to the parsed segments, if we need them
>> bool isBufferWritable = true;
>> avformat_alloc_output_context2(&this->adtsMuxerContext, NULL, "adts", NULL);
>> this->adtsMuxerWriterContext = avio_alloc_context(this->adtsPacketBuffer, this->adtsPacketBufferSize, isBufferWritable, this, NULL, AudioAssetReader::writeADTSPacket, NULL);
>> this->adtsMuxerContext->pb = this->adtsMuxerWriterContext;
>>
>> // initialize an ADTS stream
>> AVCodecID aacCodec = AVCodecID::AV_CODEC_ID_AAC;
>> AVCodec *codec = avcodec_find_encoder(aacCodec);
>> AVStream *stream = avformat_new_stream(this->adtsMuxerContext, codec);
>> stream->id = this->adtsMuxerContext->nb_streams - 1;
>> stream->time_base.den = input.getSampleRateHz();
>> stream->time_base.num = 1;
>>
>> // configure the stream with the details of the AAC packets
>> AVCodecParameters *codecParameters = stream->codecpar;
>> codecParameters->codec_id = aacCodec;
>> codecParameters->bit_rate = input.getBitrateBps();
>> codecParameters->profile = this->getAACProfileForCodec(input.getCodec());
>> codecParameters->sample_rate = input.getSampleRateHz();
>> codecParameters->channels = input.getChannelCount();
>> codecParameters->codec_type = AVMEDIA_TYPE_AUDIO;
>> codecParameters->channel_layout = av_get_default_channel_layout(input.getChannelCount());
>>
>> uint8_t* header = new uint8_t[7];
>> memset(header, 1, 7);
>> //header[3] = 2 << 6;
>>
>> codecParameters->extradata = header;
>> codecParameters->extradata_size = 7;
>>
>> // write out a header to the muxer; this is a no-op for our purposes
>> const int result = avformat_write_header(this->adtsMuxerContext, NULL);
>> if (result < 0) {
>> char *message = new char[AV_ERROR_MAX_STRING_SIZE];
>> av_make_error_string(message, AV_ERROR_MAX_STRING_SIZE, result);
>> cerr<<endl<<"Failed to write the ADTS muxer header due to: "<<result<<": "<<message<<endl;
>> delete[] message;
>> }
>> }
>>
>> Ronak
>> _______________________________________________
>> Libav-user mailing list
>> Libav-user at ffmpeg.org
>> https://ffmpeg.org/mailman/listinfo/libav-user
>>
>> To unsubscribe, visit link above, or email
>> libav-user-request at ffmpeg.org with subject "unsubscribe".
>
> _______________________________________________
> Libav-user mailing list
> Libav-user at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/libav-user
>
> To unsubscribe, visit link above, or email
> libav-user-request at ffmpeg.org with subject "unsubscribe".
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ffmpeg.org/pipermail/libav-user/attachments/20191004/3145d58a/attachment.html>
More information about the Libav-user
mailing list