[mythtv] HDTV: Multiple audio streams

[Daniel Thor Kristjansson]

No, I have a big deadline at the end of this month so I don't have any
time for hobby programming at the moment. I did capture a myth recording
that has mangled audio though, it's in English. I'll try to chop off a
short piece and post it somewhere. It's very fuzzy and low volume and
then just turns to noise after a few seconds of playing. It seems to be
repeatable too, the last two episodes of "Gilmore Girls" in HDTV seems
to have this problem.

-- Daniel

On Thu, 4 Mar 2004, John P. Poet wrote:

]
]Daniel,
]
]Have you made any progress with this?
]
]I am should have my taxes figured out by Sunday, and should be able to
]resume looking into this.  Just wondering if you have investigated any more.
]
]
]John
]
]
]On Tue, 24 Feb 2004, Jason Hoos wrote:
]
]> There may be sufficient information in the PMT itself to determine which
]> stream to take.  As you say, there is some extra information associated with
]> each stream in the PMT.  On the streams I have, the PMTs at least seem to
]> include the ISO 639 language code, which RewritePMT currently completely
]> ignores.  Currently (well, last I looked at it anyways) RewritePMT just
]> grabs the PID of the first audio stream it finds; my guess is that the
]> station you're having trouble with shuffles the order that the streams are
]> listed in in the PMT every now and then.  I don't have any stations around
]> here that I know of that have any overdub or vision-impaired audio streams,
]> so I don't know what information those streams would be labeled with.
]>
]> I don't know how reliable this information is.  The one channel I looked at
]> had the audio stream labeled as being Spanish, even though my ears told me
]> it was English.  I'm not sure if the PMT was just flat-out wrong, or if the
]> stream could have been in Spanish normally and just not for this show.  The
]> video stream was labeled as English, so who knows.  I'm thinking the PMT for
]> this station might actually be wrong, since I'm pretty sure this station
]> broadcasts Spanish on a different subchannel entirely.  On the other hand,
]> this station only has one audio stream per subchannel, so perhaps if
]> RewritePMT looked for the stream with the right language and other
]> descriptors, and failing that just took the first one it found, things might
]> work right most of the time...
]>
]> Jason
]>
]> P.S. If you need some information on PMT structure let me know.
]>
]>
]> > -----Original Message-----
]> > From: Daniel Thor Kristjansson [mailto:danielk at cat.nyu.edu]
]> > Sent: Monday, February 23, 2004 10:18 PM
]> > To: Jason Hoos
]> > Subject: RE: Multiple audio streams
]> >
]> >
]> >
]> > Well I'm looking for an example now. I can't seem to find one when I
]> > need it; but the Simpsons seem to be the main offender. Basically I
]> > would get something like the stream for the blind or Spanish. I think
]> > these are all AC3 streams, from my reading of the standard this is all
]> > that is allowed with ATSC. The standard does define some labeling for
]> > the different stream types under the AC-3 audio service
]> > descriptor 0x81
]> > which can be in either the PMT or EIT tables. Each AC3 stream seams to
]> > be pointed to in the VCT table. Then the AC3 audio service descriptor
]> > has some pertinent information.
]> > From http://www.atsc.org/standards/a_52a.pdf (p 122-123)
]> >  bsmod    -- ?? I think this may be the type of stream, such as stream
]> >              for the blind.
]> >  full_svc -- 1 bit flag for whether this is complete or needs to be
]> >              mixed with another audio service
]> > if (bsmod<2) {
]> >  mainid  -- 3 bit field telling you which particular main service this
]> >             belongs to.
]> > } else {
]> >  asvcflags -- 8 bit flag field associating this with one or
]> > more of the
]> >               eight main services
]> > }
]> >  langcod/langcod2 -- language
]> >  text[]   -- description of service
]> >  ISO_639_language_code -- 24 bit ISO 639 language code
]> >
]> > But this seems to be supplimentary to the PID tables, right?
]> > I think it
]> > describes one of the eight main audio services described by the
]> > "service_location descriptor".
]> >
]> > The service_location descriptor might have (a_65b p 116)
]> >  PID_audio_1   AC-3 audio    Spanish
]> >  PID_audio_2   AC-3 audio    English
]> >  PID_audio_3   AC-3 audio    English
]> >  PID_audio_4   AC-3 audio    French
]> >  PID_video     MPEG-2 video  No lang.
]> >
]> > I'm assuming they mean in the VCT table (p 74). Apparently this is in
]> > the PSIP table to "minimize the time required for tuning" but
]> > PAT and PMT
]> > information is required for MPEG-2 compliance (p 120). But
]> > then they go
]> > on to say the PMT should always be processed because the PSIP will not
]> > cover cases of multiple audio tracks for the same language,
]> > which leads
]> > me to think that PMT is always the more correct thing to look at.
]> >
]> > Conclusion...
]> > First, we can to look at this VCT table for all the streams that match
]> > our target language, but we need to look at the PMT AC-3 table to find
]> > the bsmod service in our language that we want (i.e. bsmod for the
]> > sighted, for the blind, the "emergency" channel, effects, dialog,
]> > dialog for the hearing impaired, etc.)
]> >
]> > Does this make sense?
]> >
]> > -- Daniel
]> >
]>
]
]