 |  Outbound VoiceXML Applications via HTTP | | TOC | | XML Grammars |  |
| VoiceXML 2.1 Development Guide | Home | Frameset Home |
|
<initial><initial> element.
<?xml version="1.0" encoding="UTF-8" ?>
<vxml version="2.1"
xmlns="http://www.w3.org/2001/vxml"
xmlns:voxeo="http://community.voxeo.com/xmlns/vxml">
<!-- Welcome Form -->
<!-- Plays welcome message -->
<form id="frm_welcome">
<block>
<prompt>
Welcome to the dead celebrity hotline
</prompt>
<!-- Move to the main form -->
<goto next="#frm_get_celeb" />
</block>
</form>
<!-- Main Form -->
<form id="frm_get_celeb">
<!-- Load the grammar file -->
<grammar src="mixed_init.srgs.xml" type="application/grammar-xml" root="CELEBRITY" />
<!-- Prompt the user for info -->
<initial name="init_get_celeb">
<prompt>
Please tell us which dead celebrity you saw, and where we can find him.
</prompt>
</initial>
<!-- Ask the user for the celebrities name -->
<field name="celebrity">
<prompt>Which celebrity did you see?</prompt>
</field>
<!-- ask the user where the celeb was -->
<field name="location">
<prompt>And where was he sighted at?</prompt>
</field>
<!-- All of the fields have been matched -->
<filled mode="all">
<!-- Goodbye -->
<exit />
</filled>
</form>
</vxml>
<filled> elements within each <field><initial> element above the fields.<filled> elements from each field and place a <filled mode="all"> within our main form. This will allow us to match *both* utterances before moving to the filled section. The <initial> element is really the key ingredient here. The initial element is what makes mixed initiative dialogs possible. Execution will visit the initial section first at attempt to match both of our utterances.<filled> section. Having the filler rules will significantly reduce the chance of having a <nomatch> condition met.
<?xml version= "1.0" encoding="UTF-8" ?>
<grammar xmlns="http://www.w3.org/2001/06/grammar"
xml:lang="en-US">
<!-- create the top level rule -->
<!-- this grammar will fill both subgrammars and put each result into a fancy variable -->
<!-- be sure to declare this as public so we can read it in the vxml script -->
<rule id="CELEBRITY" scope="public">
<!-- match one of the <item>'s between the <one-of> tags -->
<one-of>
<item>
<!-- both grammars were matched mixed-init style -->
<item><ruleref uri="#CELEB"/><tag>out.celebrity=rules.CELEB.celebrity;</tag></item>
<item><ruleref uri="#LOCATION"/><tag>out.location=rules.LOCATION.location</tag></item>
</item>
<item>
<!-- matched celeb only -->
<ruleref uri="#CELEB"/>
</item>
<item>
<!-- matched location only -->
<ruleref uri="#LOCATION"/>
</item>
</one-of>
</rule>
<!-- Celebrity grammar -->
<!-- we can use this by itself or mixed-init style -->
<rule id="CELEB" scope="public">
<one-of>
<item>
<!-- optional utterance prefixes go here -->
<!-- Load the celebrity filler subgrammar -->
<item repeat="0-1"> <!-- a filler match can be repeated once if desired -->
<ruleref uri="#FILL_CELEB"/>
</item>
<!-- choose one of these celebrities -->
<!-- return the info we need in out.celebrity -->
<one-of>
<item> elvis presley <tag> out.celebrity = "Elvis Presley"; </tag></item>
<item> buddy holly <tag> out.celebrity = "Buddy Holly"; </tag></item>
<item> john travolta <tag> out.celebrity = "John Travolta"; </tag></item>
<item> frank sinatra <tag> out.celebrity = "Frank Sinatra"; </tag></item>
<item> marty feldman <tag> out.celebrity = "Marty Feldman"; </tag></item>
<item> jim morrison <tag> out.celebrity = "Jim Morrison"; </tag></item>
</one-of>
</item>
</one-of>
</rule>
<!-- the location grammar -->
<!-- same as celebrity grammar, just different info to match -->
<rule id="LOCATION" scope="public">
<one-of>
<item>
<!-- optional utterance prefixes go here -->
<item repeat="0-1">
<ruleref uri="#FILL_LOC"/>
</item>
<one-of>
<item> flea market <tag> out.location = "Flea Market"; </tag></item>
<item> wrestling match <tag> out.location = "Wrestling Match"; </tag></item>
<item> las vegas <tag> out.location = "Las Vegas"; </tag></item>
<item> my bathroom <tag> out.location = "My Bathroom"; </tag></item>
<item> grace land <tag> out.location = "Graceland"; </tag></item>
<item> dairy queen <tag> out.location = "Dairy Queen"; </tag></item>
</one-of>
</item>
</one-of>
</rule>
<!-- filler grammar to prefix 'celeb' utterance' -->
<rule id="FILL_CELEB" scope="public">
<one-of>
<item> i saw </item>
<item> could swear it was </item>
<item> it was </item>
<item> holy moley </item>
</one-of>
</rule>
<!-- filler grammar to prefix 'location' utterance' -->
<rule id="FILL_LOC" scope="public">
<one-of>
<item> in the </item>
<item> at the </item>
<item> was at the </item>
<item> ditty bopping at </item>
<item> big as life near the </item>
<item> looking fatter in person at </item>
<item> hanging out at </item>
<item> eating like a pig at </item>
</one-of>
</rule>
</grammar>
<initial> element handles the initial utterance from the user. If it matches both fields, it will move execution to the <filled> section. If it does not match, it will move to the <nomatch> section. Since filled is somewhat self explanatory, we will go over some special code for the nomatch section. The initial element has an attribute called "name" which will also create a variable with the same name as the value of the name attribute. We can think of this variable as a switch. When we set the name variable to "true", the form will use the two individual fields instead of the initial section. This allows us to ask the user for input in separate pieces if they can't match the mixed initiative style grammar from the initial section. For example:
<initial name="init_get_celeb">
<prompt>
Please tell us which dead celebrity you saw, and where we can find him.
</prompt>
<!-- User did not say what we needed -->
<!-- after the first failure -->
<nomatch count="1">
<prompt>
Okay, I'll ask you for information one piece at a time.
</prompt>
<!-- set the initial elements name to true -->
<!-- this will make the app prompt the user for each piece of information -->
<assign name="init_get_celeb" expr="true" />
<!-- start over -->
<reprompt/>
</nomatch>
</initial>
<nomatch> section that will be triggered only once, and it will disable our mixed initiative input mode. After this section is met, the user will have to enter information one piece of information at a time because we set "init_get_celeb" (the initial section's name) to true. Also, it is worth noting that we have not mapped any fields to the grammar slots. This is because the individual subgrammars have the same names as our fields, and our form was told which grammars to use at the form level.<filled mode="all">? Well, execution will now move to the filled section and since "all" was specified, we have two variables available: "celebrity" and "location" filled with their respective data returned from our grammar. What we are going to do with the data is provide some custom messages that will be played to the user based on their input.
<!-- Ask the user for the celebrities name -->
<field name="celebrity">
<prompt>Which celebrity did you see?</prompt>
</field>
<!-- ask the user where the celeb was -->
<field name="location">
<prompt>And where was he sighted at?</prompt>
</field>
<!-- All of the fields have been matched -->
<filled mode="all">
<!-- John Travolta is the best actor ever -->
<if cond="celebrity == 'John Travolta'">
<prompt>
True, John Travolta is still alive, but his career sure isn't.
</prompt>
</if>
<!-- play the results to the user -->
<prompt>
Thank you for your report.
A team armed with butterfly nets and
</prompt>
<!-- Customize the response depending on which celeb was matched -->
<if cond="celebrity == 'Elvis Presley'">
<prompt>
a peanut butter and banana sandwich
</prompt>
<elseif cond="celebrity == 'Buddy Holly'" />
<prompt>
a pair of contact lenses
</prompt>
<elseif cond="celebrity == 'John Travolta'" />
<prompt>
a script for battlefield earth, part two
</prompt>
<elseif cond="celebrity == 'Frank Sinatra'" />
<prompt>
a double martini, and two teenage girls
</prompt>
<elseif cond="celebrity == 'Marty Feldman'" />
<prompt>
an all you can eat shrimp platter
</prompt>
<elseif cond="celebrity == 'Jim Morrison'" />
<prompt>
L S D
</prompt>
</if>
<!-- put the location into the mix -->
<prompt>
for bait will be deployed by helicopter to
<value expr="location" />
in an attempt to recapture
<value expr="celebrity" />
<break time="1000ms" />
Goodbye
</prompt>
<!-- Goodbye -->
<exit />
</filled>
<initial> element and how to implement form level grammars. If you want to try this sample code out for yourself, feel free to download the code below.
Source Code
| ANNOTATIONS: EXISTING POSTS |
zweit
|
|
| Hi,
After I have changed the value of "count" to "2" in <nomatch> tag and seen the effect, I understood what role "count" played, or have you mentioned it in the former lessons? regard ------- Secondt |
|
Michael.Book
|
|
| Howdy Secondt,
The tutorials, in the interest of brevity, simply do not detail every element and attribute used in the sample code. However, this VoiceXML documentation set does include handy breakdowns of each VoiceXML element and its respective attributes. Just find the element that has sparked your curiosity under the "ELEMENTS" menu at the bottom of the frame on the left, and give it a click. Detailed goodness awaits... I hope this helps... Have Fun, ~ Michael |
|
sambhav
|
|
| Hello
I am developing an application where number of fields is more than two in a form. say four. Can anybody suggest how I can achieve mixed initiative approach in that case? I want to ensure that user can fill one,two,three or all four attribute at a time. Thanks a lot :-) |
|
voxeoJohn
|
|
| Hello,
We would be glad to help in this regard, have you had a chance to give the above tutorial a try? Are you having an issue getting this to work? If you are we ask that you please submit debugger output (via our support ticketing system) so we can offer our assistance getting your application to work. We'll be standing by for your update at this time. Regards, John Dyer Customer Engineer Voxeo Support |
|
sambhav
|
|
| Hello
I worked the above example with four fields. Here is the code. ***Grammar File*** <?xml version= "1.0" encoding="UTF-8" ?> <!-- Whole grammar in a readable form 1. ROOT -> COLOR FRUIT FLOWER CUISINE |COLOR_FRUIT |FRUIT_FLOWER |FLOWER_CUISINE |COLOR |FRUIT |FLOWER |CUISINE 2. COLOR_FRUIT -> COLOR FRUIT 3. FRUIT_FLOWER -> FRUIT FLOWER 4. FLOWER_CUISINE -> FLOWER CUISINE 5. COLOR -> color 6. FRUIT -> fruit 7. CUISINE -> cuisine /* This grammar suggests that caller can provide all four field in one go or just two of them, or one at a time Number of combinations of providing fields can be higher, but here we have simplified our need */ --> <grammar xmlns="http://www.w3.org/2001/06/grammar" xml:lang="en-US"> <rule id="ROOT" scope="public"> <one-of> <!-- matched color followed by fruit flower and cuisine --> <item> <item> <ruleref uri="#COLOR"/> <tag>out.color=rules.COLOR.color;</tag> </item> <item> <ruleref uri="#FRUIT"/> <tag>out.fruit=rules.FRUIT.fruit</tag> </item> <item> <ruleref uri="#FLOWER"/> <tag>out.flower=rules.FLOWER.flower</tag> </item> <item> <ruleref uri="#CUISINE"/> <tag>out.cuisine=rules.CUISINE.cuisine</tag> </item> </item> <!-- matched color followed by fruit --> <item> <ruleref uri="#COLOR"/> <tag>out.color=rules.COLOR.color;</tag> <ruleref uri="#FRUIT"/> <tag>out.fruit=rules.FRUIT.fruit;</tag> </item> <!-- matched fruit followed by flower --> <item> <ruleref uri="#FRUIT"/> <tag>out.fruit=rules.FRUIT.fruit;</tag> <ruleref uri="#FLOWER"/> <tag>out.flower=rules.FLOWER.flower;</tag> </item> <!-- matched flower followed by cuisine --> <item> <ruleref uri="#FLOWER"/> <tag>out.flower=rules.FLOWER.flower;</tag> <ruleref uri="#CUISINE"/> <tag>out.cuisine=rules.CUISINE.cuisine</tag> </item> <item> <!-- matched color only --> <ruleref uri="#COLOR"/> </item> <item> <!-- matched fruit only --> <ruleref uri="#FRUIT"/> </item> <item> <!-- matched flower only --> <ruleref uri="#FLOWER"/> </item> <item> <!-- matched cuisine only --> <ruleref uri="#CUISINE"/> </item> </one-of> </rule> <!-- Color grammar --> <!-- we can use this by itself or mixed-init style --> <rule id="COLOR" scope="public"> <one-of> <item> <!-- optional utterance prefixes go here --> <!-- Load the color filler subgrammar --> <item repeat="0-1"> <!-- a filler match can be repeated once if desired --> <ruleref uri="#FILL_COLOR"/> </item> <!-- choose one of these colors --> <!-- return the info we need in out.color --> <one-of> <item> red <tag> out.color = "red"; </tag> </item> <item> blue <tag> out.color = "blue"; </tag> </item> <item> green <tag> out.color = "green"; </tag> </item> <item> yellow <tag> out.color = "yellow"; </tag> </item> <item> violet <tag> out.color = "violet"; </tag> </item> <item> magenta <tag> out.color = "magenta"; </tag> </item> </one-of> </item> </one-of> </rule> <!-- the fruit grammar --> <rule id="FRUIT" scope="public"> <one-of> <item> <!-- optional utterance prefixes go here --> <item repeat="0-1"> <ruleref uri="#FILL_FRUIT"/> </item> <one-of> <item> mango <tag> out.fruit = "mango"; </tag> </item> <item> banana <tag> out.fruit = "banana"; </tag> </item> <item> apple <tag> out.fruit = "apple"; </tag> </item> <item> strawberry <tag> out.fruit = "strawberry"; </tag> </item> </one-of> </item> </one-of> </rule> <rule id="FLOWER" scope="public"> <one-of> <item> <!-- optional utterance prefixes go here --> <item repeat="0-1"> <ruleref uri="#FILL_FLOWER"/> </item> <one-of> <item> rose <tag>out.flower = "rose"; </tag> </item> <item> lily <tag>out.flower = "lily"; </tag> </item> <item> lotus <tag>out.flower = "lotus"; </tag> </item> </one-of> </item> </one-of> </rule> <rule id="CUISINE" scope="public"> <one-of> <item> <!-- optional utterance prefixes go here --> <item repeat="0-1"> <ruleref uri="#FILL_CUISINE"/> </item> <one-of> <item> indian <tag>out.cuisine = "indian"; </tag> </item> <item> chinese <tag>out.cuisine = "chinese"; </tag> </item> <item> italian <tag>out.cuisine = "italian"; </tag> </item> <item> mexican <tag>out.cuisine = "mexican"; </tag> </item> </one-of> </item> </one-of> </rule> <!-- filler grammar to prefix 'color' utterance' --> <rule id="FILL_COLOR" scope="public"> <one-of> <item> i like </item> <item> i love </item> <item> my favourite color is </item> <item> my color is </item> </one-of> </rule> <!-- filler grammar to prefix 'fruit' utterance' --> <rule id="FILL_FRUIT" scope="public"> <one-of> <item repeat="0-1"> and </item> <item> i like </item> <item> i love </item> <item> my favourite fruit is </item> </one-of> </rule> <!-- filler grammar to prefix 'flower' utterance' --> <rule id="FILL_FLOWER" scope="public"> <one-of> <item repeat="0-1"> and </item> <item> i like </item> <item> i love </item> <item> my favourite flower is </item> </one-of> </rule> <!-- filler grammar to prefix 'cuisine' utterance' --> <rule id="FILL_CUISINE" scope="public"> <one-of> <item repeat="0-1"> and </item> <item> i like </item> <item> i love </item> <item> my favourite cuisine is </item> </one-of> </rule> </grammar> ***Voicepage*** <%@ page session="false" %> <?xml version="1.0" encoding="UTF-8" ?> <vxml version="2.1"> <form id="experiment"> <grammar src="../grammars/four_variables.xml#ROOT" type="application/grammar-xml"/> <initial name="initial_prompt"> <prompt bargein="false"> What is your favourite color, fruit, flower and cuisine? </prompt> <nomatch count="1"> Okay, I'll ask you for information one piece at a time. <assign name="myInit" expr="true"/> <reprompt/> </nomatch> <noinput> <prompt bargein="false"> I could not hear anything.Lets do it again. </prompt> <reprompt/> </noinput> </initial> <field name="color"> <prompt bargein="false">What is your favourite color?</prompt> <grammar src="../grammars/four_variables.xml#COLOR" type="application/grammar-xml"/> </field> <field name="fruit"> <prompt bargein="false">And what is your favourite fruit?</prompt> <grammar src="../grammars/four_variables.xml#FRUIT" type="application/grammar-xml"/> </field> <field name="flower"> <prompt bargein="false">And flower?</prompt> <grammar src="../grammars/four_variables.xml#FLOWER" type="application/grammar-xml"/> </field> <field name="cuisine"> <prompt bargein="false">finally, your cuisine will be?</prompt> <grammar src="../grammars/four_variables.xml#CUISINE" type="application/grammar-xml"/> </field> <filled mode="any"> <prompt bargein="false"> your favourite color is <value expr="color"/>, fruit is <value expr="fruit"/>, flower is <value expr="flower"/> and cuisine is <value expr="cuisine"/> </prompt> </filled> </form> </vxml> *** code ends here *** The above code is working fine.I can say all four fields in one go or two at a time as suggested by grammar.If i provide two fields, i will asked for remaining two fields in two different prompts. Problem:- If i provide two attributes, I should be able to provide remaining two attributes in one go. How can I achieve that? Thanks :-) |
|
voxeoJeffK
|
|
| Hello,
This would be the normal, expected flow. The user's inputs are taken, and then the FIA chooses the empty fields to visit so that the correct prompting can drive the remaining inputs. It may be possible through some complex use of multiple <initial> blocks by selectively clearing their form item variable, cond or expr. The difficulty lies in that you have to account for each of the field input items. Which ones were filled, and which were not? If you go back to a single <initial> then you will get prompted for all of them! So for four choices you would need 24 <initial> blocks. Is it that you are required to capture all the remaining input items, or was there a misunderstanding concerning the behavior of <initial> in mixed dialogs? Regards, Jeff Kustermann Voxeo Support |
|
sambhav
|
|
| Hello
I need to capture remaining fields.Number of remaining fields at any time in a conversation will be decided by caller's earlier responses. I am clear with <initial> tag's behavior. Thanks :-) |
| login |
|