[Vorbis-dev] I heard about aoTuV...

[Gregory Maxwell]

On Fri, May 2, 2008 at 8:04 PM, Sebastian Pipping
<webmaster at hartwork.org> wrote:
> ... and was wondering how it differs from the official
>  libvorbis and why there is need for a fork.
>
>  Could somebody shed some light on the past and background
>  of the libvorbis/aoTuV story please? Both on- and offlist
>  mails are welcome.
>
>  Thanks in advance,
>  Sebastian

I'm sure someone with more direct involvement in these matters can
provide a better answer... but here is a quick one:

There really isn't much to say...  Aoyoume does fantastic work on the
aoTuV encoder. clearly improving it..

By maintaining a separate copy he has more freedom to explore options
and change things.  He also gets the recognition he deserves for the
hard work that he is doing mostly alone.

There are different pressures on the reference encoder...
compatibility with all the decoders out there (Not that aoTuV would be
at all incompatible with a compliant decoder, but there are some
broken decoders out there, so any change has a little risk), much
wider distribution, a large established base of users who are happy
with the performance.

Occasionally Aoyoume's results are merged into the slower moving
reference encoder. In the past when this happened there was time
consuming careful review done and a few extra improvements were made.

If your interested in using Vorbis is at the lowest bitrates you'll
certainly want to checkout the aoTuV branch since listening tests have
demonstrated that it tends to outperform the reference encoder at
those bitrates.

In an open world we all benefit from diversity because there is no
incentive to make things incompatible (quite the opposite).   If I
could change something about this situation it would only be the wish
that there were two more groups of people working on their own open
encoders.. incrementally improving the state of the art, and learning
from each others efforts.