Recherche avancée

Sur d’autres sites (11787)

• Best way to pipe audio and video chunks from within python to ffmpeg

  8 mai 2016, par basilikum

  Problem

  I’m getting audio and video chunks from a third-party server and I would like to pipe those chunks to ffmpeg to create a WebM live stream according to these instructions :

  http://wiki.webmproject.org/adaptive-streaming/instructions-to-do-webm-live-streaming-via-dash

  Here they are using input from webcam and microphone but I need to use the data chunks, so the ffmpeg command would look somehow like this :

      cmd = [
  
          "ffmpeg",
  
          "-f", "flv", "-i", "video.fifo",
  
          "-f", "s16le", "-ar", "16000", "-ac", "1", "-i", "audio.fifo",
  
          "-map", "0:0",
  
          "-pix_fmt", "yuv420p",
  
          "-c:v", "libvpx-vp9",
  
          "-s", "640x480", "-keyint_min", "40", "-g", "40", "-speed", "6",
  
          "-tile-columns", "4", "-frame-parallel", "1", "-threads", "8",
  
          "-static-thresh", "0", "-max-intra-rate", "300",
  
          "-deadline", "realtime", "-lag-in-frames", "0",
  
          "-error-resilient", "1",
  
          "-b:v", "3000k",
  
          "-f", "webm_chunk",
  
          "-header", self.video_header,
  
          "-chunk_start_index", "1",
  
          "video_360_%d.chk",
  
          "-map", "1:0",
  
          "-c:a", "libvorbis",
  
          "-b:a", "16k", "-ar", "16000",
  
          "-f", "webm_chunk",
  
          "-audio_chunk_duration", "2000",
  
          "-header", self.audio_header,
  
          "-chunk_start_index", "1",
  
          "audio_171_%d.chk"
  
      ]

  As you can see, I am using a "video.fifo" and "audio.fifo" file, because I thought it would be a good idea to pipe the chunks in via a named pipe, but I can’t get it to work. Here is what I’m doing :

  p = subprocess.Popen(cmd)
  
  fa = os.open("video.fifo", os.O_WRONLY)
  
  fv = os.open("audio.fifo", os.O_WRONLY)

  So I’m starting the subprocess first, so that it opens the fifo files for reading. After that, I should be able to open them for writing but I am not. More specifically, I am able to open the first one, but not the second one. So maybe that has something to do with how ffmpeg handles its inputs if there are more than one, but I just don’t know.

  Question

  How can I either solve the problem of non openable named pipes or how can I achieve what I wanted to achieve without named pipes.

• Manifest error pops up when trying to download certain videos for discord.py music bot

  30 novembre 2020, par bork

  I'm trying to make a discord music bot. I'm using youtube-dl to retrieve the info and ffmpeg to play the audio out. My bot has no problems downloading videos and everything is working fine. But when I tried downloading certain videos, this error popped up :

  


  [dash @ 0x7fe45a801200] Manifest too large: 65055
https://manifest.googlevideo.com/api/manifest/dash/expire/1606336104/ei/CGq-X4G6Htauz7sPrZuCsA8/ip/14.192.212.39/id/64a943d43b8f53eb/source/youtube/requiressl/yes/playback_host/r5---sn-h5mpn-30ae.googlevideo.com/mh/Go/mm/31%2C29/mn/sn-h5mpn-30ae%2Csn-30a7rn7l/ms/au%2Crdu/mv/m/mvi/5/pl/24/hfr/all/as/fmp4_audio_clear%2Cwebm_audio_clear%2Cwebm2_audio_clear%2Cfmp4_sd_hd_clear%2Cwebm2_sd_hd_clear/initcwndbps/533750/vprv/1/mt/1606314038/fvip/5/keepalive/yes/beids/23927369/itag/0/sparams/expire%2Cei%2Cip%2Cid%2Csource%2Crequiressl%2Chfr%2Cas%2Cvprv%2Citag/sig/AOq0QJ8wRQIhAOn6Br0QsuXc-3unfhYdzXVXcydzVWioIQlKvv2U4i3OAiB6ApoiqFoPvE3YKYGPbRiId_bHQYO8zsawGGPMidYGAA%3D%3D/lsparams/playback_host%2Cmh%2Cmm%2Cmn%2Cms%2Cmv%2Cmvi%2Cpl%2Cinitcwndbps/lsig/AG3C_xAwRQIhANmxfRNBI4wSJo6trsKkq8GinQ-ADMxgHRmelBwM-GEAAiAafey9YRrZz1h1S6PzV3u0S6IsZUKscGrrGP9Pofv2uQ%3D%3D: Invalid data found when processing input


  


  After trying to download other videos, which have no problem at all, I found out that the videos showing these errors have an extra step that downloads MPD manifest. I would try to download videos that are way larger and it would work, but it's just these certain videos, with the duration of about 7-10 minutes, that would have these errors. I'm really lost.

  


  This is my code for playing the videos in the voice channels :

  


  voice = get(client.voice_clients, guild = ctx.guild)
with YoutubeDL(ydl_opts) as ydl:
                info = ydl.extract_info(url[playlist], download = False)

URL = info['formats'][0]['url']
voice.play(FFmpegPCMAudio(URL, **FFMPEG_OPTIONS))


  


• Can't load MPEG file with MSE

  11 juillet 2019, par Léandre

  Here is my problem : I want to reproduce this example of MSE use, but when I change the mp4 file, it doesn’t work. I have this error :

  Failed to execute 'endOfStream' on 'MediaSource': The MediaSource's readyState is not 'open'. at SourceBuffer.<anonymous></anonymous>

  I saw here that I needed to fragment my mp4 file, wich I did. Here is my mp4 file header :

  [ftyp] size=8+28
  
     major_brand = mp42
  
     minor_version = 0
  
     compatible_brand = isom
  
     compatible_brand = iso2
  
     compatible_brand = avc1
  
     compatible_brand = mp41
  
     compatible_brand = mp42
  
  [moov] size=8+3841
  
     [mvhd] size=12+96

  And this is the mp4 file header of the file demo : (this file works)

  [ftyp] size=8+20
  
     major_brand = mp42
  
     minor_version = 1
  
     compatible_brand = mp42
  
     compatible_brand = avc1
  
     compatible_brand = iso5
  
  [moov] size=8+1871
  
     [mvhd] size=12+96
  
        timescale = 1000
  
        duration = 60095 

  Even when I change the minor value with MP4Box -brand mp42:1 fragmented-video.mp4, it doesn’t work.

  I also tried to separate video data and audio data ( view in FFMPEG H.264 encoding for HTML5 (and ultimately MPEG-DASH) ), but it didn’t solve the problem.

  (I tried on Chromium)

  If you have any ideas to solve this, it will be great !

  Thanks