[Vorbis] OGG compression optimization

[xiphmont at xiph.org]

> Well, it says that the method is patented, so that means it can't be used.

No, that just means _we_ wouldn't be able to do anything with it :-)
Don't put too much stock in patents-- you could patent yogurt assuming
the internet was somehow involved. :-)

I do have a criticism of what's being presented though; it is another
'electronic ear' algorithm, but unlike the others we have available,
there seems to no information on what it's actually doing.  Electronic
quality evaluation algorithms tend not to match up very well with
actual human ears. If they did, the problem of compression would be
much easier.  That's not even accounting for the issue of large
individual variation in humans.

In short, for anyone who cares enough to be interested in using a tool
like this, the tool is not likely to be trustworthy enough to be
useful.  Not to mention-- Vorbis already has psychoacoustic evaluation
built in. That's the whole point of the quality-based modes.  -q <n>
is supposed to produce identical quality for all inputs for any given
<n>.

Anyway, the writeup makes it sound very much like one of several
preexisting algorithms with a new marketing strategy, aka, digital
snake oil.  If that's not the case, Sevana, by all means do fill us in
:-)

Monty