Recherche avancée

Sur d’autres sites (6140)

• Cannot stream properly on YouTube using ffmpeg with Python

  1er mars 2018, par Francesco

  I have a few problems related to sending ffmpeg stream to YouTube.
  1. If I comment out "time.sleep(sleep_time)" the stream goes 3x fast(basing on what ?) ;
  2. If I uncomment the ’-r’ args the stream lags ;
  3. If I don’t add the background music the stream doesn’t start on YouTube, the problem persists if I uncomment the ’-an’ args.
  4. When the background music ends the stream stops working on YouTube.

  frame = cv2.imread('STATIC_IMAGE.jpg')
  
  fps = 25
  
  sleepTime = 1 / fps
  
  
  
  height, width, channels = frame.shape
  
  
  
  command = [
  
      'ffmpeg',
  
  
  
      #OpenCV image.
  
      #'-re',
  
      #'-threads', '0',
  
      '-f', 'rawvideo', #image2pipe
  
      '-vcodec','rawvideo',
  
      '-s', str(width) + 'x' + str(height),
  
      '-pixel_format', 'bgr24',
  
      #'-r', str(fps),
  
      #'-an',
  
      '-i', '-',
  
  
  
      #Background music
  
      #'-stream_loop', '-1',
  
      #'-re',
  
      #'-r', str(fps),
  
      '-i', 'music.mp3',
  
  
  
      #Output (actual stream)
  
      #'-r', str(fps),
  
      #'-crf', '25',
  
      #'-g', '39',
  
      #'-g', '2',
  
      #'-ac', '2',
  
      #'-c:a', 'aac',
  
      #'-b:a', '128k',
  
      #'-ar', '44100',
  
      #'-an',
  
      #'-b:v', '300k',
  
      #'-c:v', 'libx264',
  
      #'-bufsize', '600k',
  
      #'-maxrate', '300k',
  
      #'-qmin', '32',
  
      #'-qmax', '64',
  
      #'-vcodec', 'libx264',
  
      #'-pixel_format', 'yuv420p',
  
      '-vcodec', 'h264',
  
      '-pixel_format', 'h264',
  
      '-f', 'flv',
  
      'rtmp://a.rtmp.youtube.com/live2/STREAM_KEY'
  
  ]
  
  
  
  import subprocess as sp
  
  proc = sp.Popen(command, stdin=sp.PIPE, shell=False)
  
  
  
  while True:
  
      proc.stdin.write(frame.tostring())
  
  
  
      time.sleep(sleepTime)

  Any idea of why these problems happen ?
  May the "while True" be the problem ?

• Seeking Guidance on Audio Development Libraries for Playing Various Music Formats on iOS [closed]

  3 décembre 2023, par FaiChou

  I am exploring audio development and am overwhelmed by the numerous libraries available :

  


  

  • AVFoundation

  


  • MediaPlayer

  


  • AVFAudio

  


  • Audio Engine

  


  • CoreAudio

  


  • AudioToolbox

  


  • Audio Unit

  


  


  Which of these should I use to play both network or local music files (flac/wav/mp3/aac/wma/ogg) ? What are their specific roles ?

  


  My understanding is that I should use ffmpeg for demuxing to unpackage, then decode the stream to get PCM samples, possibly using Audio Engine for playback. Is this a software decoding method ? For Apple-supported formats like mp3, can I bypass ffmpeg for decoding ? Which library would be most suitable in this case ?

  


  Additionally, how do I extract metadata (song name, artist, cover, lyrics, etc.) from music files ? Apple's own solutions seem to support only ID3 or iTunes tags. How should I handle other formats like wav/flac ?

  


  The two links below are the resources I've referenced, but they seem outdated. My search for newer, reliable explanations has been fruitless.

  


  

  • whats-the-difference-between-the-apple-audio-frameworks

  


  • Core Audio Essentials

  


  


• Ffmpeg - padding/margins/offset in amix filter. Overlay conversation to music

  10 juin 2022, par user19313832

  Use filter_complex and amix to overlay conversation with music.

  


  There is a video file with music, and an audio file with a conversation.
When they start talking, the volume of the music fades. When the conversation ends, the volume of the music increases again. It works.

  


  But there is a problem that the volume of the music decreases only after they started talking. It is required that the volume of the music decrease even before they start talking, with an indent that can be specified. That is, when they have not said anything yet, the volume of the music decreases, for example, in a second when they start talking. Is there a solution ?