Re: [codec] WG Review: Internet Wideband Audio Codec (codec)

Here's what I've seen (maybe some other ITU-T attendees can comment).

The charter of SG16 (and its "questions" - like WGs in the IETF) is handled
somewhat differently than the IETF charter process.  The entire ITU-T
charter and organization is re-approved every study period (3 years), and
the organization and leadership is kept for the entire period.  Each
question has its own focus, specified in the charter.  This does not include
specific work items.  When new work is proposed, first the question agrees
to add them to its work program. Then the work program is approved by the
working party (one layer up), and finally the full study group (in a plenary
session). This happens fairly rapidly (within the same meeting session).

The ITU-T questions that do audio codecs first identify the application need
(reasonably detailed requirements) for the proposed codec, and makes sure it
does not duplicate ITU-T previous work (and other work the members know
about).  If there is agreement to proceed, the codec is added to the ITU-T
work program.  The approved work program is published, and usually sent to
other interested SDOs in liason statements.

Later on, the group works out even more specific requirements (called "terms
of reference") and develops a test plan for candidates. Often there are two
batteries of tests - a "lower bar" for qualifying reasonable candidates, and
a "higher bar" for choosing between the candidates that pass the first test.

Then it calls for proposals, accepts candidates, and starts the selection
process and characterization/testing.

Stephen Botzko
Polycom


On Thu, Jan 21, 2010 at 10:20 AM, Jean-Marc Valin <
jean-marc(_dot_)valin(_at_)octasic(_dot_)com> wrote:

Hi,

Actually, maybe we can look at how other SDOs are handling this issue.
Considering that ITU-T, 3GPP/3GPP2 and (to a lesser extent) MPEG all
standardise codecs in the same space, how do these SDOs coordinate? For
example, does the ITU-T SG16 have some text in it's "charter" that says "we
will communicate the requirements to 3GPP and other SDOs to see if they
have codecs that fill these requirements"? Can someone more familiar than I
am with the ITU-T/3GPPx/MPEG/... explain how this issue is being handled?

       Jean-Marc

Adrian Farrel wrote:

Richard,

I think I agree...

It's not clear to me why SDOs need to be involved in the process of
determining whether existing codecs satisfy the requirements.


However, no-one can make the determination without requirements to make

an

evaluation against.

And to be sure that all the candidates are in the melting pot, it is at
worst harmless to poll the other SDOs for their input and suggestions.

I would expect that one of the tasks of this WG is to coordinate and
document (i.e. make) the evaluation.

Cheers,
Adrian

Information on standard codecs -- including their technical and legal
aspects -- is pretty widely available.  And if information about a

 codec

isn't generally available (e.g., if standards are being closely  held),
then that codec fails to meet the requirements by definition --
there's a
requirement that it by widely implementable, which requires  its
specification to be widely available.

I've only been following this discussion off and on, but I don't  really
see anyone really challenging the requirements in the current  draft
charter, and I don't really see anyone proposing codecs that  meet those
requirements. Unless one of those two changes, it seems  evident that

the

requirements are not being satisfied, so we should  just move on with
forming the WG.

--Richard



On Jan 21, 2010, at 8:39 AM, Adrian Farrel wrote:

[snip]

What I try to say is that first the requirements must be set, only
then
will it be possible for representatives of other SDOs to determine

if

already standarddized codecs (or codecs under standardization)  meet
them.


I agree.  Obviously no one (inside or outside the IETF) can tell
exactly
how existing codecs in other SDOs relate to this work until the
detailed
requirements are locked down.

Also, I think the burden is mostly on CODEC to make this  assessment.
Other
SDOs may offer their views in liason statements, and can respond  with
their
own work programs.  But in the end it would be up the IETF to
decide if
there is too much overlap.


Right, and this is surely easy to achieve and good project  management,
anyway.

Document the requirements to a reasonable level of detail.
Circulate the requirements explicitly requesting suggestions.
Evaluate the suggestions and give reasons for rejecting existing
Codecs.
Go on and develop a new Codec if required.

It does not follow that people cannot start work on a new Codec  before
completion of the third step, but the WG would be premature  to adopt a
Codec solution draft before having formally surveyed the  landscape.

The first step has to be done anyway, and I don't see that it can be
considered as slowing down the development of a solution since it is
impossible to build a solution without knowing the requirements. The
second step might add a few weeks to the cycle. The third step, if  we
are to believe the comments in this thread, will not take long.

So why does anyone object to such a process?

As to whether this sequence of steps should be codified in the

 charter,

my experience is that if you don't write down a process, it  is very
hard
to get interoperable implementations.

Thanks,
Adrian


_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf


_______________________________________________
codec mailing list
codec(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/codec

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf