Re: hypothetical: 'simplistic' A/V codec


well, I have a straightforward exmplanation for this:
lossy predictive filtering works fairly well for audio, and in my
experience, can get results comprable to those of MDCT based transforms, but
is a whole lot less complicated.

This I don't understand. Lossy predictive filtering is badly inferior to
MDCT based codecs at anything but very high bitrates. And you can write
very simple MDCT based codecs which are not going to be any more
complicated than a prediction based codec, and will perform much better.

It's not because Vorbis (or AAC, or MP3) is needlessy complicated that
your codec would have to be too.

--
GCP
.