W3C home > Mailing lists > Public > www-voice@w3.org > January to March 2011

Re: Fwd: Re: DTMF, MRCP, but ... what???

From: Chris Davis <davisc@iivip.com>
Date: Tue, 08 Feb 2011 13:09:08 -0600
Message-ID: <4D5194D4.9040206@iivip.com>
To: luke@dashjr.org
CC: www-voice@w3.org
>>  Why would the MRCP server be providing a semantic interpretation of its own

That's just Nuance.

>>How should a conforming VoiceXML interpreter handle cases like this?

use the SWI_meaning tag and pull out the value as the semantic
interpretation.
so.. the following:

<SWI_meaning>
    {SWI_literal:1 2 3 4}
</SWI_meaning>

results in the input var being:

"1 2 3 4"

Regards,
Chris

--------------------------
Our MRCP server is yielding the following NLSML result for a basic grammar
with no semantic interpretation tags:
	<?xml version='1.0'?>
	<result>
	<interpretation grammar="session:144235924" confidence="100">
	<input mode="dtmf">
     	1 2 3 4
	</input>
	<instance>
	<SWI_literal>
	1 2 3 4
	</SWI_literal>
	<SWI_grammarName>
	session:144235540
	</SWI_grammarName>
	<SWI_meaning>
	{SWI_literal:1 2 3 4}
	</SWI_meaning>
	</instance>
	</interpretation>
	</result>

Carefully reading over the NLSML and VoiceXML specifications, it seems correct
to assign the full<instance>  contents to the VXML variable as an ECMAScript
object structure. However, the application of course is expecting the raw
input string '1 2 3 4', not an object.

Why would the MRCP server be providing a semantic interpretation of its own,
rather than omitting it? How should a conforming VoiceXML interpreter handle
cases like this?

Thanks,

Luke

-- 
Chris Davis
Interact Incorporated R&D
512-502-9969x117
Received on Tuesday, 8 February 2011 19:09:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 February 2011 19:09:52 GMT