Studio 10.5 Form Element: VoiceKeyword bug?

[janinegraves]

Hi,

Has anyone used Studio 10.5 Form element with VoiceKeyword settings?

I'm getting an error from Nuance10 that the grammar VXMLServer creates is not valid when it tries to use the speech reco grammar.

And the DTMF Keypress setting causes NoMatch when I enter DTMF.

Has anyone else seen this?

Does anyone know if there's an ES for this?

Here are the details:

Studio Form Element:

Voice Keyword: janine [janine]

Voice Keyword: graves [graves]

VXMLServer Grammar created:

	<field name="foundation_fld" modal="false">
	<property name="inputmodes" value="dtmf voice" />
	<grammar mode="voice" version="1.0" xml:lang="en-US" root="ROOTVOICE">
	<rule id="ROOTVOICE" scope="public">
	<one-of>
	<item>
	janine
	<tag><![CDATA[<foundation_fld  "janine">]]></tag>
	</item>
	<item>
	graves
	<tag><![CDATA[<foundation_fld  "graves">]]></tag>
	</item>
	</one-of>
	</rule>
	</grammar>

Nuance Log:

20151012082740218| 4200| 0| FGBLNBGCAAACNBJCAAAAAAAB| MyApp| ** WARNING **| -9|  SWI_ERROR_INVALID_ARG| invalid argument| SWItrcGetEventStreamSWIrcndTod |

20151012083300598| 4120| 0| FGBLNCKCAAAGPMBCAAAAAAAB| MyApp| ** ERROR => SVC AFFECT **| 4512|  SWI_JS_SCRIPT_EXCEPTION| JS error report - is exception| ECMA-script | invalid XML tag syntax while processing script '<unknown>'

20151012083300598| 4120| 0| FGBLNCKCAAAGPMBCAAAAAAAB| MyApp| ** ERROR => SVC AFFECT **| 4503|  SWI_JS_SCRIPT_ERROR| Error with JS script| ECMA-script | Error compiling function {<foundation_fld  "janine">} in (parser\SWIjs2.cpp:3998).

20151012083300598| 4120| 0| FGBLNCKCAAAGPMBCAAAAAAAB| MyApp| RecognizerErrorUnwind|| SWI error GRAMMAR_ERROR(13) detected in parser\SWIjs2.cpp:3999

20151012083300598| 4120| 0| FGBLNCKCAAAGPMBCAAAAAAAB| MyApp| RecognizerErrorUnwind|| SWI error GRAMMAR_ERROR(13) detected in parser\SWIjs2.cpp:4095

[SEAN NILSEN]

Hi Janine,

Were you able to work around this issue? I'm running into the same thing.

Thanks!

[neilfluhr · Accepted solution]

This is a confirmed defect in Unified CVP version 10.5. The fix is in version 11. The only option with 10.5 is to work-around it by converting all of the "DTMF Keypress" form parameters to an external grammar.

[hesam yavari]

...same issue here... we dont have the option of upgrading our client v11 at this moment in time.

[ewindgat]

Yes, the defect number is CSCux0379. It was fixed in 11.0. There is no workaround available.

[janinegraves]

There's a lengthy workaround - and that is to switch from using the inline grammar (Settings named Voice Keyword and DTMF Keypress) to using an external grammar that resides in a file on the media server, and then using the Form element's settings named Voice Grammar and DTMF Grammar settings with a URL that the recognizer can use to retrieve the grammar.

This can be done, it just takes a little work up front. But if you have 10.5 customers with ASR and Form elements, you may want to know how to do this.

1. You have to enable VXML Debug Logging

A) Go to Project/Properties/CallStudio/General.

B) Set the Language: en-US  (or whatever your default is)

C) Also in that pane, Next to the Loggers box, press Add...   to add the Debug Logger info

Name: VxmlLogs

Class: com.audium.logger.application.debug.ApplicationDebugLogger

E) Save and Deploy the Studio application to VxmlServer.

F) On VXMLServer, you must now manually modify the file

C:\Cisco\CVP\VXMLServer:\applications\AppName\data\application\settings.xml

change this line:   <voice_browser>cisco_nuance_01</voice_browser>

to this:                   <voice_browser>cisco_osr_02</voice_browser>

G) Run the updateApp.bat script in the folder C:\Cisco\CVP\VXMLServer:\applications\AppName\admin\updateApp.bat to load this modified app into memory.

H) Now call in and you'll have valid grammars created.

I) After the phone call, open the VXML Debug Log that contains the valid grammar(s). It'll be in

C:\Cisco\CVP\VXMLServer:\applications\AppName\logs\VxmlLogs

scroll down in that file until you find the grammars created by the Form element.

You'll copy the voice grammar and the dtmf grammar into separate text files.

In each of the text files you create that contains the grammar, you MUST add this xml header on the VERY FIRST LINE and at the first position in the file:  <?xml version="1.0"?>         ********there was a typo here before, but this is right now*******

J)Save the files with a .xml extension and put them somewhere on your media server.

K) Then modify the form element with URI's that point to the VoiceGrammar and DTMFGrammar files you just created.

L)Test by putting the URI into a web browser and ensure each grammar is retrievable.

Now save, deploy, and test your app by calling in. It should work.

M) Once it works, remove the VXML Debug Logging from Project/Properties - this is super important if you want to have a job in a week! :-)