Re: [codec] WGLC #2 for draft-ietf-codec-opus-10

["Benjamin M. Schwartz" <[e-mail]>]

On 11/09/2011 07:30 AM, Erik Norvell wrote:
> The current situation is that there are no constraints on the encoder other than that it should produce a readable bitstream. It may change the quality in any direction. In the interest of maximizing the implementation freedom, the same should apply for the decoder.

The above suggestion is impossible as a matter of logic, because our
purpose is interoperability.

If the decoder is normatively specified, then an encoder may be designed
so that the output of a compliant decoder sounds like the original input
audio.  This is our current situation.

If the encoder is normatively specified, then a decoder may be designed so
that its output sounds like the original input audio encoded by a
compliant encoder.

If neither the encoder nor the decoder is normatively specified, then
neither can be designed with any expectation of what sort of output will
be produced.

Without any normative specification, it would be trivial to build a
compliant but non-interoperable encoder-decoder pair.  Such an encoder
would write valid Opus bitstreams, but these bitstreams would only sound
like the original input audio when played through the corresponding
decoder.  Similarly, the decoder would not produce correct audio output
when connected to other encoder implementations.

Of course, this is all old news.  We have already reached consensus on
this issue, which is why it is all explicitly built into the document that
guides our process here:

http://datatracker.ietf.org/doc/draft-ietf-codec-guidelines/?include_text=1

See Sections 4.1 and 4.3.

--Ben