Recherche avancée

Sur d’autres sites (7002)

• Bad file descriptor decoding mp3 to pipe with ffmpeg

  1er février 2017, par Pete Bleackley

  Can any FFMpeg gurus help me with the following ?

  I’m trying to convert a podcast to PCM and downsample it to 16KHz mono before feeding it to a speech recognition system for transcription. The command line

  ffmpeg -i http://media.blubrry.com/conlangery/content.blubrry.com/conlangery/Conlangery01.mp3 -f s16le -ac 1 -ar 16000 pipe:0

  fails with

  av_interleaved_write_frame() : Bad file descriptor

  What is the problem here and how do I fix it ?

  EDIT

  Full error message is

  ffmpeg version N-83189-gd5d474aea5-static http://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2017 the FFmpeg developers
  
    built with gcc 5.4.1 (Debian 5.4.1-4) 20161202
  
    configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc-5 --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gray --enable-libass --enable-libfreetype --enable-libfribidi --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxvid --enable-libzimg
  
    libavutil      55. 44.100 / 55. 44.100
  
    libavcodec     57. 75.100 / 57. 75.100
  
    libavformat    57. 62.100 / 57. 62.100
  
    libavdevice    57.  2.100 / 57.  2.100
  
    libavfilter     6. 69.100 /  6. 69.100
  
    libswscale      4.  3.101 /  4.  3.101
  
    libswresample   2.  4.100 /  2.  4.100
  
    libpostproc    54.  2.100 / 54.  2.100
  
  Input #0, mp3, from 'http://media.blubrry.com/conlangery/content.blubrry.com/conlangery/Conlangery01.mp3':
  
    Metadata:
  
      track           : 1
  
      album           : Conlangery Podcast
  
      title           : Conlangery 01
  
      artist          : George Corley
  
      date            : 2011
  
    Duration: 00:44:58.08, start: 0.025057, bitrate: 128 kb/s
  
      Stream #0:0: Audio: mp3, 44100 Hz, stereo, s16p, 128 kb/s
  
      Metadata:
  
        encoder         : LAME3.98r
  
  Output #0, s16le, to 'pipe:0':
  
    Metadata:
  
      track           : 1
  
      album           : Conlangery Podcast
  
      title           : Conlangery 01
  
      artist          : George Corley
  
      date            : 2011
  
      encoder         : Lavf57.62.100
  
      Stream #0:0: Audio: pcm_s16le, 16000 Hz, mono, s16, 256 kb/s
  
      Metadata:
  
        encoder         : Lavc57.75.100 pcm_s16le
  
  Stream mapping:
  
    Stream #0:0 -> #0:0 (mp3 (native) -> pcm_s16le (native))
  
  Press [q] to stop, [?] for help
  
  av_interleaved_write_frame(): Bad file descriptor
  
  Error writing trailer of pipe:0: Bad file descriptorsize=       1kB time=00:00:00.02 bitrate= 256.0kbits/s speed=95.2x    
  
  video:0kB audio:1kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.000000%
  
  Conversion failed!
  
  Traceback (most recent call last):
  
    File "./PodcastTranscriber.py", line 206, in <module>
  
      PodcastTranscriber('http://conlangery.com/feed/',upload)()        
  
    File "./PodcastTranscriber.py", line 101, in __call__
  
      self.process(item)
  
    File "./PodcastTranscriber.py", line 136, in process
  
      (audio,errors)=mp3.run(stdout=subprocess.PIPE)
  
    File "/usr/local/lib/python2.7/site-packages/ffmpy.py", line 105, in run
  
      raise FFRuntimeError(self.cmd, self.process.returncode, out[0], out[1])
  
  ffmpy.FFRuntimeError: `ffmpeg -i http://media.blubrry.com/conlangery/content.blubrry.com/conlangery/Conlangery01.mp3 -f s16le -ac 1 -ar 16000 pipe:0` exited with status 1
  
  </module>

  The invoking code is

  def process(self,item):
  
      """Downloads the audio and transcribes it"""
  
      audio_url=None
  
      print item['title']
  
      for link in item.links:
  
          if link['rel']=='enclosure':
  
              audio_url=link['href']
  
      if audio_url is not None:
  
          pubDate=to_date(item.published_parsed)
  
          if self.lastUploadDate is None or pubDate>self.lastUploadDate:
  
              self.lastUploadDate=pubDate
  
          mp3=ffmpy.FFmpeg(inputs={audio_url:None},
  
                           outputs={'pipe:0':['-f','s16le','-ac','1','-ar','16000']})
  
          (audio,errors)=mp3.run(stdout=subprocess.PIPE)
  
          sphinx=subprocess.Popen(['java','-jar','transcriber.jar'],
  
                                  stdin=audio,
  
                                  stdout=subprocess.PIPE)
  
          wiki=threading.Thread(target=self.callback,args=(item,sphinx.stdout))
  
          wiki.start()
  
          #mp3.start()
  
          #mp3.join()
  
          sphinx.stdin.close()
  
          wiki.join()
  
          sphinx.wait()

• How to correctly calculate which segments are ready to be downloaded using MPEG-DASH

  24 avril 2019, par igal k

  What i’m trying to do ?

  Write a simple MPEG-DASH client using the SegmentTemplate pattern to calculate which segments are ready to be downloaded for a live source.

  A picture taken using chrome’s debugging tools at a moment X showing an mpd request(8af651fd747.....mpd) and the actual segments fetched respectfully to that request.

  

  Given the following MPD

  <mpd availabilitystarttime="2019-04-24T06:43:32Z" maxsegmentduration="PT4.096S" minbuffertime="PT4.096S" minimumupdateperiod="PT15.835S" profiles="urn:mpeg:dash:profile:isoff-live:2011" publishtime="2019-04-24T11:14:01Z" suggestedpresentationdelay="PT11.878S" timeshiftbufferdepth="PT65.536S" type="dynamic" xmlns="urn:mpeg:dash:schema:mpd:2011">
  
    <location>https://content-aaps1.uplynk.com/channel/8af651fd7473474f86a05ffb0a1c8972.mpd?rmt=wv&amp;amp;cid=8af651fd7473474f86a05ffb0a1c8972&amp;amp;oid=600e5c27541344a1bf3818617ad712ce&amp;amp;prettydash=1&amp;amp;exp=1556091088&amp;amp;rn=4138683939&amp;amp;tc=1&amp;amp;ct=c&amp;amp;sig=5fb7f0c18f3f1d2ad4fdee53c02c1e1ed904bc5e8474f4ebf886d209ff7f21c9&amp;amp;pbs=05b6594bcf4b4728ac1094976a80194d</location>
  
    <period start="PT2826.240S">
  
      <adaptationset maxframerate="30" maxheight="720" maxwidth="1280" mimetype="video/mp4" segmentalignment="true" startwithsap="1">
  
        <representation bandwidth="2604473" codecs="avc1.64001e" framerate="30" height="360" scantype="progressive" width="640">
  
          <baseurl>https://x-default-stgec.uplynk.com/aapm/slices/8c1/600e5c27541344a1bf3818617ad712ce/8c1027496a964b049f1bd5895f8f0412/</baseurl>
  
          <segmenttemplate duration="368640" initialization="https://x-default-stgec.uplynk.com/aapm/slices/8c1/600e5c27541344a1bf3818617ad712ce/8c1027496a964b049f1bd5895f8f0412/$RepresentationID$_init.mp4?pbs=05b6594bcf4b4728ac1094976a80194d&amp;amp;_jt=l&amp;amp;chid=8af651fd7473474f86a05ffb0a1c8972" media="$RepresentationID$$Number%08d$.m4f?pbs=05b6594bcf4b4728ac1094976a80194d&amp;amp;_jt=l&amp;amp;chid=8af651fd7473474f86a05ffb0a1c8972" presentationtimeoffset="254361599" startnumber="690" timescale="90000"></segmenttemplate>
  
        </representation>
  
      </adaptationset>
  
    </period>
  
    <utctiming schemeiduri="urn:mpeg:dash:utc:http-iso:2014" value="https://content-aaps1.uplynk.com/misc/utcservertime"></utctiming>
  
  </mpd>

  I see that the next segment request should be #3955

  What i have tried so far

      period.end = 1556104456;
  
      period.start = 2826;
  
      availability_start_time = 1556088212;
  
      max_segment_duration = 4;
  
      time_shift_buffer_depth = 65

  So, first of all, i read DASH-IF-IOP 4.3 section 4.3.4.2 page #82 and implemented the following code :

     int k1 = 1;
  
  int period_duration = period.end - (period.start + data_.availability_start_time);
  
  int k2 = ceil((float)period_duration / (float)data_.max_segment_duration);
  
  double duration = ((float)representation.duration / (float)representation.timeScale);
  
  size_t live_edge = std::min(
  
      (int)floor((float)((data_.publish_time - data_.availability_start_time - period.start) / duration)), k2);
  
  
  
  size_t oldest = std::max(k1, (int)floor((float)((data_.publish_time - data_.availability_start_time - period.start -
  
                                                      data_.time_shift_buffer_depth) /
  
                                                  duration)));

  after calculating everything : k1=1, k2=3355, live_edge=3272 and oldest=3256

  Also tried using ffmpeg’s dashdec.c

  for min_segment :

  if (c->is_live && pls->fragment_duration)
  
      {
  
          num = pls->first_seq_no + (((get_current_time_in_sec() - c->availability_start_time) - c->time_shift_buffer_depth) * pls->fragment_timescale) / pls->fragment_duration;
  
      }

  for max_segment :

  num = pls->first_seq_no + (((get_current_time_in_sec() - c->availability_start_time)) * pls->fragment_timescale) / pls->fragment_duration;

  after a small modification :

  size_t pmax = (((data_.publish_time - data_.availability_start_time))) / duration;
  
  size_t pmin = ((data_.publish_time - data_.availability_start_time) - data_.time_shift_buffer_depth) / duration;

  pmin=3946 pmax=3961

  in the ffmpeg example, i had to manually remove the first_seq_no variable because it looked like i doubled added the SegmentTemplate@StartNumber.

  even after succeeding in this task, how do i exactly build the request list of Segment(NOW) ----> Segment(LIVE_EDGE)

• lavu : Drop old deprecated AVOption API

  5 septembre 2015, par Hendrik Leppkes

  lavu : Drop old deprecated AVOption API
  Deprecated in 2011/10
  

  • [DH] libavutil/opt.c
  • [DH] libavutil/opt.h
  • [DH] libavutil/version.h