Recherche avancée

Sur d’autres sites (10333)

• Streaming converted movie with mp4 container in NodeJS, movie playing very fast

  20 septembre 2016, par Mustafa

  I have used stream-transcoder module to convert a file make it a stream. So the file is not stored, it is on the fly.

  app.get("/video", function(req,res){
  
      res.writeHead(200, {'Content-Type': 'video/mp4'});
  
      var src = "movie.avi";
  
  
  
      var Transcoder = require('stream-transcoder');
  
      var stream = fs.createReadStream(src);
  
      new Transcoder(stream)
  
          .maxSize(1280, 720)
  
          .videoCodec('h264')
  
          .videoBitrate(800 * 1000)
  
          .fps(25)
  
          .sampleRate(44100)
  
          .channels(2)
  
          .audioBitrate(128 * 1000)
  
          .format('mp4')
  
          .on('finish', function() {
  
              console.log("finished");
  
          })
  
          .stream().pipe(res);
  
  });

  It works nicely, it is fast, but too fast, the audio is played at the same speed, however the video does not respect the frame rate, whatever is recieved from ffmpeg is immeidately shown, fastly. Additionally, it does not show the total time, I believe it is the problem. I need to somehow specify the length, framerate, but I could not find enough information on that. I thought the stream recieved from ffmpeg should contain that. And I could not find respective headers for that in HTTP.

  Here are the flags that this stream-transcoder module uses for MP4 :

  [ '-i',
  
    '-',
  
    '-vf',
  
    'scale=min(trunc(1280/hsub)*hsub\\,trunc(a*720/hsub)*hsub):min(trunc(720/vsub)*vsub\\,trunc(1280/a/vsub)*vsub)',
  
    '-vcodec',
  
    'h264',
  
    '-b:v',
  
    800000,
  
    '-r',
  
    25,
  
    '-ar',
  
    44100,
  
    '-ac',
  
    2,
  
    '-ab',
  
    128000,
  
    '-f',
  
    'mp4',
  
    '-movflags',
  
    'frag_keyframe+faststart',
  
    'pipe:1' ]

  When I use VP8 encoder and WebM, it works nicely, the time is displayed, video plays normal speed.

• Noise when decoding mp3 using libavcodec on iOS

  11 janvier 2014, par Trenskow

  I have this methods. It first sets up an audio input if necessary. Then it reads a frame, and returns the frame converted using an AudioConverterRef (wrapped in an KFInlineConverterUnit class) to my Core Audio graph. At the end of the method it tears down the input, if audio output has stopped.

  - (NSData *)readNextChunk {
  
  
  
      /* Setup audio input */
  
      if (_formatContext == NULL && self.runLoopActive) {
  
  
  
          static dispatch_once_t onceToken;
  
          dispatch_once(&onceToken, ^{
  
              av_register_all();
  
          });
  
  
  
          avformat_network_init();
  
  
  
          av_init_packet(&_packet);
  
  
  
          _formatContext = avformat_alloc_context();
  
  
  
          [self handleError:avformat_open_input(&_formatContext, [[_url absoluteString] cStringUsingEncoding:NSUTF8StringEncoding], NULL, NULL)];
  
  
  
          [self handleError:avformat_find_stream_info(_formatContext, NULL)];
  
  
  
          _audioStreamIndex = -1;
  
          for (NSInteger index = 0 ; index < _formatContext->nb_streams ; index++)
  
              if (_formatContext->streams[index]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
  
                  _audioStreamIndex = index;
  
                  break;
  
              }
  
  
  
          if (_audioStreamIndex > -1) {
  
  
  
              _codec = avcodec_find_decoder(_formatContext->streams[_audioStreamIndex]->codec->codec_id);
  
  
  
              _codecContext = _formatContext->streams[_audioStreamIndex]->codec;
  
  
  
              [self handleError:avcodec_open2(_codecContext, _codec, NULL)];
  
  
  
          }
  
  
  
          if (self.timeOffset > .0) {
  
              int64_t timestamp = AV_TIME_BASE * self.timeOffset;
  
              av_seek_frame(_formatContext, -1, timestamp, AVSEEK_FLAG_ANY);
  
          }
  
  
  
      }
  
  
  
      /* Read and decode next frame */
  
  
  
      NSData *chunk = nil;        
  
      AVFrame *frame = avcodec_alloc_frame();
  
      while (self.runLoopActive && _formatContext != NULL && av_read_frame(_formatContext, &_packet) >= 0) {
  
  
  
          if (_packet.stream_index == _audioStreamIndex) {
  
  
  
              int gotFrame = 0;
  
              int len = avcodec_decode_audio4(_codecContext, frame, &gotFrame, &_packet);
  
              [self handleError:len];
  
  
  
              if (gotFrame) {
  
  
  
                  /* Setup a converter for my Audio Graph */
  
                  AudioStreamBasicDescription inputAudioDescription = [self audioDescriptionForCodecContext];
  
  
  
                  if (!_converter || memcmp(&inputAudioDescription, _converter.inputAudioDescription, sizeof(AudioStreamBasicDescription)) != 0)
  
                      _converter = [KFInlineConverterUnit newWithInputAudioDescription:inputAudioDescription
  
                                                                outputAudioDescription:self.audioDescription];
  
  
  
                  /* Convert frames */
  
                  if (inputAudioDescription.mFormatFlags & kAudioFormatFlagIsNonInterleaved) {
  
                      NSMutableArray *buffers = [@[] mutableCopy];
  
                      for (NSInteger i = 0 ; i < frame->channels ; i++) {
  
                          [buffers addObject:[NSData dataWithBytes:frame->data[i] length:frame->nb_samples * inputAudioDescription.mBytesPerFrame]];
  
                      }
  
                      chunk = [_converter convertBuffer:buffers];
  
                  } else
  
                      chunk = [_converter convertBuffer:@[[NSData dataWithBytes:frame->data[0] length:frame->nb_samples * inputAudioDescription.mBytesPerFrame]]];
  
  
  
                  if (_packet.data != NULL)
  
                      av_free_packet(&_packet);
  
  
  
                  break;
  
  
  
              }
  
  
  
          }
  
  
  
      }
  
  
  
      avcodec_free_frame(&frame);
  
  
  
      /* Tear down if output has stopped */
  
      if (!self.runLoopActive) {
  
  
  
          avformat_free_context(_formatContext);
  
          _formatContext = NULL;
  
  
  
          avformat_network_deinit();
  
  
  
      }
  
  
  
      return chunk;
  
  
  
  }

  Basically everything is fine. I am developing this for iOS, and when I test it in the iOS Simulator everything works smoothly. It plays all audio formats with no problems.

  The problem arises, when I run the code on an actual device. Then mp3 files are audible - but VERY distorted. I have tried everything in my mind to make this work, but I cannot figure what's going wrong.

  I can say that the problem does NOT lay in the converter. I've extracted the audio before it hits the decoder, and it sounds the same. So the converter just converts the distorted audio. So I think it might be an MP3 decoder issue.

  Anyone experienced this before ?

• ffmpeg error loading shared OpenCV libraries

  19 septembre 2017, par chronosynclastic

  I installed FFmpeg 3.3 from a PPA repository following the guide here. I already have OpenCV 3.3 installed on my Ubuntu 16.04 system with all the necessary libraries under /usr/local/lib. However, when I try to run ffmpeg, I get the error :

  ffmpeg: error while loading shared libraries: 
  
  libopencv_core.so.2.4: cannot open shared object file: No such file or directory 

  Does FFmpeg only work with OpenCV 2.4 libraries ? I have the corresponding library for OpenCV under /usr/local/lib/libopencv_core.so.3.3 but somehow FFmpeg looks for the version 2.4.