This item is a Priority 3 based on
http://www1.ietf.org/mail-archive/web/speechsc/current/msg01755.html.

I have marked it deferred to track it as an item for future examination.



From Jerry Carter on 12 Apr 2006:
---------------------------------



A weight parameter should be supported on RECOGNIZE.

 

Suppose that a recognition against two weighted grammars is required.  The
following message sequence would be necessary.

 

DEFINE-GRAMMAR grammar_1

DEFINE-GRAMMAR grammar_2

RECOGNIZE against a text/grammar-ref-list of grammars and corresponding weights

 

MRCP clients often default to this approach even for simple applications in
environments such as VoiceXML 2.x where weights are allowed.  MRCP clients are
often required to use this approach when handling complex applications which
make extensive use of weights.  The required use of DEFINE-GRAMMAR is
inefficient and exacerbates the problems discussed previously on the list
<http://www1.ietf.org/mail-archive/web/speechsc/current/msg01704.html>.

 

A more efficient approach is sending a single RECOGNIZE with a multipart-mime
body containing both grammar definitions and their corresponding weights.

 

---

 

The current specification might allow for a RECOGNIZE with a multipart-mime body
containing both grammar definitions followed by a block of
text/grammar-ref-list.  If this use is intended, then the text in 9.9 should
clearly call out this case and should include an example.  Even then, allowing a
weight in the headers for each grammar would be much simpler and clearer.