Cisco Cisco Unified Customer Voice Portal 10.0(1) Developer's Guide

OICE

LEMENT

: F

ORM

ISCO

CVP V

OICE

XML 3.1

Element Specifications

Notes:

•

The voice browser requires the top-level slot name in a grammar (whether inline or external) to match the field-level slot attribute
(or the field name attribute if the slot attribute is unspecified), in order for the field name variable (and hence the CVP VoiceXML
“value” element data) to be defined. For inline grammars, the CVP VoiceXML Form element automatically generates the
grammar slot name based on the slot attribute or the field name. For custom-defined grammars that are referenced from an
external source, it is the developer’s job to configure “

Field Name

” and/or “

Field Slot

” correctly so that the above requirement

is met.

•

For a grammar that returns either different slots for different inputs or multiple slots per utterance, there are two ways to configure
the Form element to store the slot/value pair information:

Leave the “

slot_element_data

” setting empty. The Form element will create element data named “nbestInterpretationX”

(where X is from 1 to the length of the n-best list) that contains a string that uses delimiters “+” and “:” to separate the multiple
slot names from their values. For example: “+Slot1:value1+Slot2:value2...”. A developer would then need to parse this string
in a subsequent element to obtain the different slot name and value pairs.

Configure the “

slot_element_data

” setting with the names for all the slots that can be returned. The Form element will

create a new set of n-best element data to store the recognition results for each slot listed in that setting. The element data will
be named as <SLOT_ELEMENT_DATAX> (where “SLOT_ELEMENT_DATA” is a string identical to the setting value and
X is from 1 to the length of the n-best list). For example, if “

slot_element_data

” have two values “city” and “state” and

there are three n-best results triggered, then a total of six element data in the names of “city1”, “city2”, “city3”, “state1”,
“state2”, and “state3” will be created to store each of the n-best values for the “city” and “state” slots. Note that if n-best
processing is disabled by setting

maxnbest

to 1, then only one interpretation will be returned per input and thereby only one

element data per slot (“city1” and “state1”) will be created.

Element Data

Name

Type

Notes

value

string This stores the value of the VXML field name variable.

value_confidence

float This stores the confidence score of the captured Form utterance. When n-best recognition is enabled, this stores the

confidence score of the top hypothesis in the n-best list.

<SLOT_ELEMENT_DATA1>
<SLOT_ELEMENT_DATA2>
…

<SLOT_ELEMENT_DATAX

string A separate set of element data stores the interpretation values for each filled slot of captured n-best utterances.

While the maximum number of <

SLOT_ELEMENT_DATAX>

values is equal to the

maxnbest

setting value, the

actual number of these values available is dependent on speech recognition at runtime, where
<

SLOT_ELEMENT_DATA1>

holds the slot value of the top hypothesis in the n-best list and

SLOT_ELEMENT_DATAX>

holds the slot value of the last hypothesis. Note that if the

slot_element_data