[vorbis-dev] Renegade MPEG member: Intro and volunteerism

[Eric Scheirer]

Hi nice Vorbis people,

I've been following the work on Ogg and Vorbis for some
time, and I've been very impressed with the work so far.
I'd like to introduce myself and volunteer for some work
as it's needed.  I'm very excited about the prospect of
developing open-source audio coding technologies that
are not encumbered by patents, and think this has the
potential to be very important work going into the 
future.  Some of you may have read my comments on 
Slashdot expressing skepticism that doing this was possible
at all.  While I'm still concerned about possible 
patent exposure in Vorbis (see below), it's a credit
to Monty that the work has progressed as far as it has.

I'm presenting finishing my Ph.D. in the Machine Listening
Group at the MIT Media Lab, where I do research on all 
kinds of digital-audio processing and psychoacoustics
topics.  I was a member of the MPEG committee from 1996
through 1999 and was one of the primary authors and an
Editor of the MPEG-4 audio standard.  The parts of 
MPEG-4 that I developed and contributed are now the only
patent-free and public-domain parts of the standard.
(The parts called "Structured Audio", for downloadable 
sound synthesis models, and "AudioBIFS", for downloadable
audio post-production and mixing).

There are several directions I can volunteer to work on,
and I'd like opinions about where my resources would be
most useful.  It looks to me like the basic psychoacoustics,
coding, and programming is well in hand, although I could
pitch in there if needed.

1. Formalizing a "standards-like" specification for Ogg and
a Vorbis decoder.  This would be useful for enabling 
interoperability between PC-based decoders that could just
use libvorbis and decoders on other platforms, like 
portables, that could not.  Such a specification would say,
as for the MP3 specification, exactly what an Ogg bitstream
looks like and how to decode Vorbis-coded data into sound.
I think this is very important if we view that people 
will want to build other devices to deal with Ogg/Vorbis in
the future.

2. Conducting formal listening tests of Ogg/Vorbis compared
to MP3, Real, AAC, and other codecs at given maximum bitrates.
I was one of the authors, and the primary statistical analyst,
of the official MPEG-2 and MPEG-4 formal listening tests,
so I know a bit about how to run these things.  The problem
here is that real money is needed, because we'd need real
audio engineers to participate as subjects, and probably 
studio time and space (we don't have good enough facilities
at MIT) to conduct the tests.

3. Helping to research patent coverage as it applies to Vorbis.
I have done private consulting on patent-coverage issues in
audio systems in the past, and while I am not a lawyer, I could
help keep things moving.  I think this is a crucial practical
issue because I'm sure MPEG patent-holders will sue to prevent
Vorbis distribution regardless of the merits.  It's best to
have done as much discovery as we can before that time comes.

One other thing that I'll probably do also for fun is to 
develop cross-coders using libvorbis to encode MPEG-4 Structured
Audio (SA) bitstreams.  This will enable Vorbis coding to be used
in strictly standards-compliant applications (important especially
for some government contracts).  SA is a general format that
includes the decoder source code along with each bitstream,
so can be used to store any format of coded data at all (see
my papers on "Generalized Audio Coding" on my WWW site for more
details on this idea).  Since SA is in the public-domain, there
are no openness implications to doing this.

Congratulations to Monty and everyone else that has been working
on Vorbis on your new-found fame!

Best to all,

 -- Eric

+-----------------+
|  Eric Scheirer  | A-7b5 D7b9|G-7 C7|Cb   C-7b5 F7#9|Bb  |B-7 E7|
|eds at media.mit.edu|      < http://sound.media.mit.edu/~eds >
|  617 253 1750   | A A/G# F#-7 F#-/E|Eb-7b5 D7b5|Db|C7b5 B7b5|Bb|
+-----------------+

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/