[MPlayer-users] advice on mencoder options

[HR]

[...]
> [mailto:mplayer-users-bounces at mplayerhq.hu] On Behalf Of 
> Dominik 'Rathann' Mierzejewski
[...]
> > Consider using vbr=4 for better sound. Then tune the 
> quality with q= and
> > aq= to get an average of 128, since that's your preference. 
> I'd guess
> > something like q=6 and aq=2 gets you close. q=5 might 
> overshoot 128, but
> > this ofcourse depends on the input. Just add/subtract any 
> spare/overshot
> > bits to vbitrate.
> 
> Could you elaborate a bit on that? I'm sure many people could 
> use a more
> detailed explanation of q and aq parameters than:
> [...]
>        q=<0-9>
>               quality (0-highest, 9-lowest) (only for VBR)

q reference: http://lame.sourceforge.net/doc/html/switchs.html#V

> 
>        aq=<0-9>
>               algorithmic quality (0-best/slowest, 9-worst/fastest)

aq reference: http://lame.sourceforge.net/doc/html/switchs.html#q

Don't mind the confusion of the parameter names; I didn't make the
confusion, and I think I've got it right. Read the other lame
documentation too - I don't know anything that isn't in the docs.

The lame docs state that the file size is totally unpredictable when
using vbr. However, for any practical purposes (ie. encoding the audio
of a DVD movie), the size is predictable to well within +- 10%, which is
good enough for targeting an approx average bitrate. q=6 has kept the
abr in the 100-115 range for a variety of my movie encodings (not
concerts or musicals, but plain movies).

Really *explaining* the algorithm quality is beyond my capability, and
probably holds only a theoretical interest for acoustics professionals.
aq=2 is recommended by the lame project. If you don't care about speed,
then aq=0 probably works just fine, and might even improve quality.

HR