[MPlayer-dev-eng] [PATCH] update ad_mpg123 in preparation to replace mp3lib

Thomas Orgis thomas-forum at orgis.org
Fri Mar 9 23:59:07 CET 2012


Am Fri, 9 Mar 2012 18:37:03 +0100
schrieb Alexander Strasser <eclipse7 at gmx.net>: 

>   Sorry, to have you waiting. While very valuable it is just a bit
> overwhelming.

I have to admit that libmpg123 as a rather "smart" library that would
like to handle things like stream parsing / resync and gapless decoding
itself, does make the use-case from MPlayer a bit non-trivial. Also I
could drop things like the bitrate estimation; but, well, I don't like
MPlayer claiming 2000 seconds of playback time for a 2 minute track.

Actually, there might be a dumbed-down partial libmpg123 in the future
that could make ad_mpg123 look simpler, but then one would loose
advanced things like gapless playback ... where we could start another
argument about if gapless decoding is wanted for video at all. Are
videos encoded so that the picture is in sync with the mp3 audio when
the encoder+decoder delay is left untouched or is cutting that cruft
expected?

We're talking about something around 20 ms shift in audio here. Does
that matter for a/v sync? I'm not sure right now. It's not nothing. So,
testing ad_mpg123 should include watching some old DivX;-) movies with
MP3 audio (encoded with LAME) and judging if sync is bettern/worse than
mp3lib.

> Maybe Ivan could comment on your latest patch and how
> it works out for him.

Yup, that would make sense.
 
>   BTW I am too in favor of removing mp3lib if this gets tested by a 
> broader audience (is in MPlayer SVN for a while) and no more problems
> occur.

Well, there is some real-world contact and nice confusion about those
two already:

http://linuxindetails.wordpress.com/2011/09/24/audio-codec-family-mpg123-afmmpg123-not-available/

So, you are not the only one who is overwhelmed, and be it by different
combinations of the letters "b", "g","i", "l", "m", "p" and some low
single-digit numbers;-)

Oh, and talking about the overwhelming world of detailed performance
measurements using MPlayer, supported by compiler weirdness: I am only
the messenger, not the one to blame;-)


Alrighty then,

Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20120309/e33dd9c7/attachment.asc>


More information about the MPlayer-dev-eng mailing list