Skip to content

matroskademux: Wrong titles

Currently when processing mkv files gstreamer extracts wrong title data for video, audio and subtitles track.

Steps to reproduce:

wget https://sample-videos.com/video123/mkv/240/big_buck_bunny_240p_1mb.mkv
wget https://raw.githubusercontent.com/spektraldevelopment/spektralvideo/master/video/bigbuckbunny/bigbuckbunny.srt
mkvmerge -o test.mkv big_buck_bunny_240p_1mb.mkv bigbuckbunny.srt
mkvpropedit --edit info --set title=MYTITLE test.mkv
mkvpropedit --edit track:a1 --set name=MYAUDIO test.mkv
mkvpropedit --edit track:s1 --set name=MYSUBS test.mkv

Now run gst-discoverer-1.0 -v test.mkv to get the following output:

Properties:
  Duration: 0:12:00.000000000
  Seekable: yes
  Live: no
  Tags: 
      encoder: Lavf53.24.2
      container format: Matroska
      datetime: 2021-01-12T23:48:29Z
      title: MYTITLE, MYSUBS, MYTITLE, MYAUDIO, MYAUDIO, MYAUDIO, MYTITLE
      bitrate: 0
      extended comment: DURATION=00:11:58.000000000
      language code: und
      audio codec: MPEG-4 AAC audio
      video codec: H264
  container: video/x-matroska
    subtitles: text/x-raw, format=(string)pango-markup
      Tags:
        encoder: Lavf53.24.2
        container format: Matroska
        datetime: 2021-01-12T23:48:29Z
        title: MYTITLE, MYSUBS
        bitrate: 0
        extended comment: DURATION=00:11:58.000000000
        language code: und
      
      Codec:
        text/x-raw, format=(string)pango-markup
      Stream ID: ff3671f6a224945258488b71b8ebdeba0e4b99364950d64ba1d50521e71ee181/003:9292031702051792385
      Language: und
    audio: audio/mpeg, mpegversion=(int)4, framed=(boolean)true, stream-format=(string)raw, codec_data=(buffer)11b0, level=(string)4, base-profile=(string)lc, profile=(string)lc, channels=(int)6, rate=(int)48000
      Tags:
        encoder: Lavf53.24.2
        container format: Matroska
        datetime: 2021-01-12T23:48:29Z
        title: MYTITLE, MYAUDIO, MYAUDIO, MYAUDIO
        bitrate: 384600
        extended comment: DURATION=00:00:12.095000000
        audio codec: MPEG-4 AAC audio
        language code: und
      
      Codec:
        audio/mpeg, mpegversion=(int)4, framed=(boolean)true, stream-format=(string)raw, codec_data=(buffer)11b0, level=(string)4, base-profile=(string)lc, profile=(string)lc, channels=(int)6, rate=(int)48000
      Stream ID: ff3671f6a224945258488b71b8ebdeba0e4b99364950d64ba1d50521e71ee181/002:002
      Language: und
      Channels: 6 (front-left, front-right, front-center, lfe1, rear-left, rear-right)
      Sample rate: 48000
      Depth: 32
      Bitrate: 384600
      Max bitrate: 0
    video: video/x-h264, level=(string)2, profile=(string)main, codec_data=(buffer)014d4014ffe10016674d4014da0507ec0440000003004000000c83c50aa801000468ef3c80, stream-format=(string)avc, alignment=(string)au, width=(int)320, height=(int)240, pixel-aspect-ratio=(fraction)1/1, framerate=(fraction)25/1, chroma-format=(string)4:2:0, bit-depth-luma=(uint)8, bit-depth-chroma=(uint)8, parsed=(boolean)true
      Tags:
        encoder: Lavf53.24.2
        container format: Matroska
        datetime: 2021-01-12T23:48:29Z
        title: MYTITLE
        bitrate: 309588
        extended comment: DURATION=00:00:12.080000000
        video codec: H264
        language code: und
      
      Codec:
        video/x-h264, level=(string)2, profile=(string)main, codec_data=(buffer)014d4014ffe10016674d4014da0507ec0440000003004000000c83c50aa801000468ef3c80, stream-format=(string)avc, alignment=(string)au, width=(int)320, height=(int)240, pixel-aspect-ratio=(fraction)1/1, framerate=(fraction)25/1, chroma-format=(string)4:2:0, bit-depth-luma=(uint)8, bit-depth-chroma=(uint)8, parsed=(boolean)true
      Stream ID: ff3671f6a224945258488b71b8ebdeba0e4b99364950d64ba1d50521e71ee181/001:001
      Width: 320
      Height: 240
      Depth: 24
      Frame rate: 25/1
      Pixel aspect ratio: 1/1
      Interlaced: false
      Bitrate: 309588
      Max bitrate: 0

As can be seen, while title for the video stream gets extracted correctly, it is wrong for the audio (MYTITLE, MYAUDIO, MYAUDIO, MYAUDIO), subtitles (MYTITLE, MYSUBS) and the whole file (MYTITLE, MYSUBS, MYTITLE, MYAUDIO, MYAUDIO, MYAUDIO, MYTITLE). Moreover, the last one seems to have always randomly changed order of items after the first "MYTITLE" when re-running gst-discoverer, which completely puzzles me.

Another scenario is when the mkv file has only the audio stream name specified and gstreamer sets it as title of the entire file, or when it has both title of the file and audio stream specified and gstreamer sets them to "MYAUDIO, MYTITLE, MYAUDIO, MYAUDIO, MYTITLE" and "MYAUDIO, MYTITLE, MYAUDIO, MYAUDIO".

I managed to bisect which commit introduced this issue: 4c10a1ba

However, reverting isn't a complete solution to the problem as before it gstreamer would set every title tag to the file title, disregarding audio and subtitle titles completely. It is a big improvement in behaviour though - if there is no title set for a video, gstreamer doesn't take it from audio title or subtitle titles and other programs can set it based on the filename.