[livesupport-dev] What should we log
  • Due to http://code.campware.org/projects/livesupport/ticket/505,
    I assume that files played throught htmlUI have not beeing logged, because
    this is just local playback.
  • 11 Comments sorted by
  • Sebastian G
  • If some of the radio-active guys can give me some informally input about the
    needs, I will write the use-case szenario.


    > -----Urspr
  • On Tue, 6 Dec 2005, Sebastian G
  • This raise the question if these (recording the broadcast or having logfiles
    and analyze methods) are 2 worlds, or must they be combined:
    Having a log which display broadcasted items, with the possibility to click
    on one and listen the sound.

    Which format is common to record sound? Especially, how the files at the end
    are dividet? By day or hour, by playlist, by single song (which would losse
    transition info)?





    > -----Urspr
  • This is something I've talked with Akos about as well. In many countries,
    a simple text log file is not enough; you have to have a sound recording
    of the broadcast. But this is something where a text log could be combined
    with the sound file using SMIL. SMIL allows you to annotate audio; in this
    case, the annotation would be the in and out points of files, so that in a
    long (let's say 4-hour) recording, you could put the in and out points at
    the points in the broadcast that correspond to the log, but have all that
    in a single SMIL file.

    That's my idea of how it could be done at least.

    doug






    Sebastian G
  • On Tue, 6 Dec 2005, Sebastian G
  • On Tue, 6 Dec 2005 Douglas.Arellanes@mdlf.org wrote:

    > This is something I've talked with Akos about as well. In many countries,
    > a simple text log file is not enough; you have to have a sound recording
    > of the broadcast. But this is something where a text log could be combined
    > with the sound file using SMIL. SMIL allows you to annotate audio; in this
    > case, the annotation would be the in and out points of files, so that in a
    > long (let's say 4-hour) recording, you could put the in and out points at
    > the points in the broadcast that correspond to the log, but have all that
    > in a single SMIL file.

    This is *not* accepted. Must be one big audio file over here. Because the
    same thing could apply for television with 'text-tv'. For text-tv you
    generaly have 20 minutes of broadcast repeated. You must record that 20
    minutes for two weeks long (ok it differs per day, but you get the
    point) Smile


    Stefan
  • Stefan,

    I think we're in general agreement. Czech stations use VHS tapes, which
    generally record in 12-hour mode, so the file could be of indeterminate
    length. How long the file is is something the user can decide, based on
    legal requirements and on what their hard disks can store.

    My point was that by making use of the work Akos has already done on
    implementing SMIL, we could combine the text and sound files in a new way,
    providing automatically annotated sound files, essentially putting tags on
    timestamps in the sound file that correspond to text entries in the log.
    This would help to ensure proof of broadcast by providing not only the log
    of the item, but also its actual sound output.

    doug






    Stefan de Konink
    12/06/2005 10:39 PM
    Please respond to livesupport-dev


    To: livesupport-dev@campware.org
    cc:
    Subject: Re: AW: AW: [livesupport-dev] What should we log


    On Tue, 6 Dec 2005 Douglas.Arellanes@mdlf.org wrote:

    > This is something I've talked with Akos about as well. In many
    countries,
    > a simple text log file is not enough; you have to have a sound recording
    > of the broadcast. But this is something where a text log could be
    combined
    > with the sound file using SMIL. SMIL allows you to annotate audio; in
    this
    > case, the annotation would be the in and out points of files, so that in
    a
    > long (let's say 4-hour) recording, you could put the in and out points
    at
    > the points in the broadcast that correspond to the log, but have all
    that
    > in a single SMIL file.

    This is *not* accepted. Must be one big audio file over here. Because the
    same thing could apply for television with 'text-tv'. For text-tv you
    generaly have 20 minutes of broadcast repeated. You must record that 20
    minutes for two weeks long (ok it differs per day, but you get the
    point) Smile


    Stefan
  • Douglas.Arellanes@mdlf.org wrote:
    > I think we're in general agreement. Czech stations use VHS tapes, which
    > generally record in 12-hour mode, so the file could be of indeterminate
    > length. How long the file is is something the user can decide, based on
    > legal requirements and on what their hard disks can store.
    Ok, that was the situation here too. VHS tapes and replacing them Surprised
    Memories Smile

    > My point was that by making use of the work Akos has already done on
    > implementing SMIL, we could combine the text and sound files in a new
    > way, providing automatically annotated sound files, essentially putting
    > tags on timestamps in the sound file that correspond to text entries in
    > the log. This would help to ensure proof of broadcast by providing not
    > only the log of the item, but also its actual sound output.
    If you place them in an ogg container, why don't you add the information
    to it 'on the fly'?

    We currently use our stations internet stream encoder (gstreamer) as
    primary source for Video and Audio logging. Every hour the client gets a
    killall and because it runs from within a bashscript a new time is
    added. I agree, not quiet professional, though it works sweet with ogg.


    Stefan
  • Stefan de Konink wrote:
    > The 'proof-of-broadcast' playlist of our current broadcast system, is just
    > artist-title. That is generally speaking not enough. For local stations
    > this is not an issue because there is another rights scheme for us.

    yes, how I see it, broof of broadcast and broadcast archive is two
    separate issues. of course they coincide based on time, and could be
    presented us such from a user point of view, but at first I won't mix
    the two..


    Akos
  • Stefan de Konink wrote:
    > We currently use our stations internet stream encoder (gstreamer) as
    > primary source for Video and Audio logging. Every hour the client gets a
    > killall and because it runs from within a bashscript a new time is
    > added. I agree, not quiet professional, though it works sweet with ogg.

    we do that same at Tilos Radio, only we use darkice instead of gstreamer...


    Akos