[mythtv] [PATCH] -- picking wrong AC-3 audio stream to play

Mark Anderson markjanderson at mail.com
Thu Feb 17 20:14:32 UTC 2005

----- Original Message ----- 
From: "Brad Templeton" <brad+mydev at templetons.com>
To: "Development of mythtv" <mythtv-dev at mythtv.org>
Sent: Friday, February 18, 2005 6:05 AM
Subject: Re: [mythtv] [PATCH] -- picking wrong AC-3 audio stream to play

> On Thu, Feb 17, 2005 at 01:11:35PM -0500, Doug Larrick wrote:
>> Brad Templeton wrote:
>> > Again, am I missing something?  When does the player play DVDs?  Or is
>> > this something planned for the future, switching from mplayer for that
>> > function?
>> 1. I think you can specify to use the internal player for MythVideo.  I
>> don't use MythVideo at all, so I don't know.
>> 2. I believe the key phrase used before was "perfect dvd rip"; you'd
>> have to do an archive search to find that discussion.
> Thanks.   In looking more at the code, I note that what it does is
> immediately grab the first (meaning last because it is counting down)
> AC-3 codec stream it finds.   Only if it finds no such stream does it
> try to pick the stream from another codec with the most channels.
I don't believe the code actually knows the number of channels in the ac3 so 
it can't apply the channels test to ac3. It **should** select the "first" 
ac3 track, or the MPEG track with the highest number of channels. As I 
mentioned in my reply to your [users] post, the count up seems to be the 
right way, but I am still concerned that I don't understand why the original 
author wrote the loop as a count down. There  must have been a reason right? 
I don't have my myth box handy so I can't look into it at the momement.

I always expected this code to be extended to honour language settings by 
someone who needs it. I don't get multiple languages so I couldn't test it 
easily anyway.

Using bitrate as well as channels in the MPEG test would also be good, but 
would probably only yield value on rare occasions.

Also, note that PS recordings, like TS, in 0.17 record all audio pids 
associated with a video pid.

> My patch changed this so it picks the "first" (as in lowest numbered)
> AC-3 stream instead of the last.
> What is confusing it is that the PBS shows have 3 streams, all with
> an AC-3 codec.   I am not sure what's in the other streams, I just
> know that the first one (0.1) is the real one, and the last one 0.3 (that 
> myth
> picks) is silent.  I don't know what's on 0.2.  Presumably
> their long term intention is SAP style stuff.
>    Input #0, mpegts, from 
> 'myth://':
>    Stream #0.0[0x41]: Video: mpeg2video, 704x480, 29.97 fps, 3975 kb/s
>    Stream #0.1[0x44]: Audio: ac3, 48000 Hz, stereo, 384 kb/s
>    Stream #0.2[0x45]: Audio: ac3, 48000 Hz, stereo, 192 kb/s
>    Stream #0.3[0x46]: Audio: ac3, 48000 Hz, stereo, 192 kb/s
> In this case, the only way to pick is by order, or by attributes.  Bitrate
> would work here but I don't know if that's universal.
> Unfortunately the true long term solution adds yet more UI, in that myth's
> player, like a DVD player, might need to have a menu to let a user select
> the audio stream if multilingual streams are arriving.
> ---- side note on video
> This recording came from PBS-Encore.  It is Nova, this one is in
> widescreen but not HD.   The recording at 1080i is overkill and 6 gigs per
> hour.  The recording at 480i above is much smaller at 1.6gb, but is much
> blurrier.   This is in part because it's letterboxed in a 4:3 window, 
> which
> seems really dumb but it seems as though it's even more than that.  The
> 1080i recording is effectively EDTV, 720x480p, but upsampled to 1080i.
> An annoying choice here -- for best quality, a 6.3 gb file, or accept less
> quality at 1.6gb.  Probably down the road the right thing to do is
> to transcode down to 960x540p or something and get a small file but
> remain progressive and keep all the resolution.  (the deinterlace is
> trivial if you go to 540p)


mythtv-dev mailing list
mythtv-dev at mythtv.org

More information about the mythtv-dev mailing list