Enigma 2: Missing audio track PTS in some MPEG-2 TS recordings? (Playback problems in VLC)

  • Some .ts files recorded with the DM 7025 seem to exhibit strange problems when played back in VLC (0.8.6a):

    • The audio track is present, but silent.
    • The video periodically breaks up in blocks and automatically skips forward – several minutes at a time.
    • The seek bar (at the bottom of the VLC window) works erratically.

    At first, I suspected this was a problem with VLC itself, since Media Player Classic (and the DM 7025) seemed to play back the same files fine. However, the VLC folks now claim that the problem is caused by missing PTS information in the audio stream, and that it is either the broadcaster’s fault, or possibly a bug in DM 7025 (Enigma 2).


    • • •


    The curious thing about this problem is that it consistently only occurs on certain multiplexes and channels – not all of them. Nonetheless, only the DM 7025 seems to be affected. Other DVR-type devices (or recordings made with ordinary DVB cards from the same source) do not seem to suffer from this problem.


    In my case, the problem is present on local DVB-T channels broadcast by YLE – the Finnish equivalent of BBC. If you take a look at the above-mentioned VLC bug report, the original poster seems to have the same problem with some Polish DVB-S channels. I have also seen reports from my fellow countrymen who seem to have the same problem with their DM 7025s, YLE’s channels, and VLC, so it’s not just me.


    • • •


    Would someone of the developers please take a closer look at this problem and see if the VLC folks are right? (I can provide sample .ts files on request; just e-mail me and ask.) It would also help if people who have seen this problem themselves would post information about the channels on which it occurs, and where they get those channels from.

    znark

    4 Mal editiert, zuletzt von Jukka Aho ()

  • Zitat

    It would also help if people who have seen this problem themselves would post information about the channels on which it occurs, and where they get those channels from.


    I have the silent audio problem on Film4 (and Film4+) on Astra 2D 28.2.


    I have to demux any Film4 recording with ProjectX and mux it again to be able to play it in VLC. Even then the playback times are reported incorrectly (e.g. VLC shows a total time of 30 minutes for a 2 hour film) I always thought it was a VLC-Problem, as demuxing the files worked flawless.

  • A sample file that demonstrates the problem (80 MB) is now available for download (for the time being, at least.) It was originally recorded with Enigma2 v2.2 but then truncated to 80 MB with dd since the original recording was over 600 MB in size.


    As you can see yourself by downloading the file, it plays fine on Enigma2 itself, but acts weird in VLC and some other players. (Media Player Classic seems to play it back well, though.) This also affects streaming – if I’m streaming live tv from the DM 7025 to VLC, the stream is missing audio as well.


    My skills at analyzing MPEG-2 transport streams and their technical validity are limited, so I cannot confirm whether the comment about “missing PTS” on the VLC bugtrack discussion holds true or not. But I would like to get to the bottom of this problem, so if any better-informed person can take a look at the sample stream and identify what exactly is wrong with it, please do so.


    If desired, I could also provide full, raw dumps of the entire multiplex. (Can this be done with the DM 7025 hardware itself, btw? Does it allow direct unfiltered access to the raw DVB stream?)

    znark

    3 Mal editiert, zuletzt von Jukka Aho ()

  • Zitat

    Originally posted by tmbinc
    I've added my comment to the bug entry, though i don't have a clue what's wrong.


    Thanks for your effort, anyway. I was kind of hoping it would be something obvious and relatively simple to pin-point, but it now appears it isn’t. :frowning_face: I wonder where we could get some kind of MPEG specialist to take a hard look into the deep bowels that sample file and determine where the fault lies, exactly.


    I think I’m going to do as I already suggested above and record a raw sample clip of the whole multiplex, too. Perhaps it would offer some further clues to solving this puzzle.

    znark

    Einmal editiert, zuletzt von Jukka Aho ()