Voice XML: Chapter 7 Exercises

VoiceXML: Introduction to Developing
Speech Applications

7-1 Identify words and phrases for a yes/no grammar by using each of the following techniques:

A. Construct two lists of words by writing down all of the words you can think of that mean "yes" and all of the words you can think of that mean "no."

B. Ask two friends to repeat (A)

C. Conduct a simple wizard of Oz experiment with ten subjects by asking them a yes/no question and recording their responses.

D. Create a simple VoiceXML program that asks a caller five yes/no questions, and records the caller responses (use the <record> tag). Encourage ten subjects to call the application and capture their responses.

<?xml version="1.0"?> <vxml version="2.0"> <!-- Exercise 7.1.d program to solicit answers to yes/no quesitons --> <form id = "data_collection"> <block> <prompt> please answer the following question by speaking the answer <break size="small"/> </prompt> </block> <record name="answer_1" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> Is four plus four equal eight? </prompt> </record> <record name="answer_2" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> Is three plus two equal six? </prompt> </record> <record name="answer_3" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> Does the month of February have thirty one days? </prompt> </record> <record name="answer_4" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> Does the month of January have thirty one days </prompt> </record> <record name="answer_5" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> Do you approve of the new european currency, the euro? </prompt> </record> <block> <prompt> Thank you for answering these questions. Goodbye <break size="small"/> </prompt> <submit next="write_answers_to_disk.cgi" namelist="answer_1 answer_2 answer_3 answer_4 answer_5" /> </block> </form> </vxml>

F. Compare the lists from (A) - (D). Order the techniques by the effort expended to obtain the lists. Order the techniques by the number of words each technique produces. Recommend a general approach for constructing a list of words and phrases for a grammar.

G. Retrieve the yes/no grammars from web sites with reusable grammars. (See the companion web site for a list of companies supplying reusable grammars.) Are there words in their grammars that are missing from your lists? Are there words in your list that are missing form their grammars? Do you recommend any changes in the general approach you recommended in (F)?

 

Name	URL	Reusable Grammars	Yes/No Grammar
BeVocal Café	www.bevocal.com	-	yes, yeah, yup, yep. ya no, nope
Hey Anita FreeSpeech	www.heyanita.com	freespeech.heyanita.com /samples/grammar.asp	yes, yeah, yup, sure, okay, correct, right no, nope
VoiceGenie Developer Workshop	www.voicegenie.com	developer.voicegenie.com /vg_grammar.php	yes right sure okay yup correct true yeah no incorrect wrong false nope
Voxeo Community	www.voxeo.com	-	-

7-2 Write the following grammars using the ABNF grammar notation for your national language:

A. Months of the year. (In English, these include "January," "February, etc" .)

<grammar roo t= "month"> <rule id = "month"> <one-of> <item>January</item> <item>February</item> <item>March</item> <item>April</item> <item>May</item> <item>June</item> <item>July</item> <item>August</item> <item>September</item> <item>October</item> <item>November</item> <item>December</item> </one-of> </rule> </grammar>

B. Days in a month. (In English, these include "one," "two," ..., "thirty-one.")

<grammar root = "day"> <rule id = "day"> <one-of> <relref id = "ones" </relref> <relref id = "tens" </relref> <relref id = "twenties" </relref> <relref id = "thirties" </relref> </one-of> </rule> <rule id = "ones"> <one-of> <item>one</item> <item>two</item> <item>three</item> <item>four</item> <item>five</item> <item>six</item> <item>seven</item> <item>eight</item> <item>none</item> </one-of> </rule> <rule id = "tens"> <one-of> <item>ten</item> <item>eleven</item> <item>twelve</item> <item>thirteen</item> <item>fourtreen</item> <item>fifteen</item> <item>sixteen</item> <item>seventeen</item> <item>eighteen</item> <item>nineteen</item> </one-of> </rule> <rule id = "twenties"> <one-of> <item>twenty</item> <item>twenty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "thirties"> <one-of> <item>thirty</item> <item>twenty one </item> </one-of> </rule>

C. Years. (In English, these include "nineteen ninety-nine," "two thousand," "two thousand and one, etc" .)

<grammar root = "year"> <rule id = "day"> <rule id = "ones"> <one-of> <item>one</item> <item>two</item> <item>three</item> <item>four</item> <item>five</item> <item>six</item> <item>seven</item> <item>eight</item> <item>none</item> </one-of> </rule> <rule id = "tens"> <one-of> <item>ten</item> <item>eleven</item> <item>twelve</item> <item>thirteen</item> <item>fourtreen</item> <item>fifteen</item> <item>sixteen</item> <item>seventeen</item> <item>eighteen</item> <item>nineteen</item> </one-of> </rule> <rule id = "twenties"> <one-of> <item>twenty</item> <item>twenty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "thirties"> <one-of> <item>thirty</item> <item>twenty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "fourties"> <one-of> <item>fourty</item> <item>fourty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "fifties"> <one-of> <item>fifty</item> <item>fifty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "sixties"> <one-of> <item>sixty</item> <item>sixty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "seventies"> <one-of> <item>seventy</item> <item>seventy <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "eighties"> <one-of> <item>eighty</item> <item>eighty <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "nineties"> <one-of> <item>ninety</item> <item>ninety <relref id = "ones" </relref> </item> </one-of> </rule> <rule id = "hundreds"> <one-of> <item>hundred</item> <item> <relref id = "ones" </relref> </item> <item> <relref id = "tens" </relref> </item> <item> <relref id = "thirties" </relref> </item> <item> <relref id = "fourties" </relref> </item> <item> <relref id = "fifties" </relref> </item> <item> <relref id = "sixties" </relref> </item> <item> <relref id = "seventies" </relref> </item> <item> <relref id = "eighties" </relref> </item> <item> <relref id = "nineties" </relref> </item> </one-of> </rule> <rule id = "years"> <one-of> <item>nineteen</item> <item>nineteen<relref id = "hundreds" </relref> </item> <item>two thousand <relref id = "hundreds" </relref> </item> </one-of> </rule> </grammar>

D. Dates. The caller may say the values for month, day and year in any order.

<grammar root = "date"> <rule id = "mdy"> <relref id = "month" </relref> <relref id = "day" </relref> <relref id = "year </relref> </rule> <rule id = "dmy"> <relref id = "day" </relref> <relref id = "month" </relref> <relref id = "year </relref> </rule> <rule id = "date"> <one-of> <relref id = "mdy" </relref> <relref id = "dmy" </relref> </one-of> </rule> </grammar>

7-3 VoiceXML has build-in grammars for Boolean, date, digits, currency, number, phone, and time. However, the definition of these grammars has not been standardized.

A. Retrieve definition for one of the build-in grammars from at least two different VoiceXML venders. Compare the two grammars and identify any differences.

Most venders do not specify the definition of built-in grammars. After using the following program to test the words accepted by a field with type Boolean

<?xml version="1.0"?> <vxml version="2.0"> <!-- evaluate use of "boolean" --> <form id = "data_collection"> <field name="test" type = "boolean"> <prompt>answer? </prompt> </field> <block> <prompt> answer is <value expr="test"/> </prompt> </block> </form> </vxml> The BeVocal engine accepts "true" but not "sure." The TellMe engine accepts "sure," but not" true."

B. If an application is to be ported across multiple VoiceXML interpreters, do you recommend using default grammars? Why?

No. Avoid using built in grammars because the resulting applications may not respond in the same way if they are ported to another engine. Replace build-in grammars by specified grammars to ensure portability.

7-4 Many callers are reluctant to speak to a computer. Write a VoiceXML form field that solicits the caller's name. Include five noresponse error handlers that encourage the caller to speak his or her name.

<?xml version="1.0"?> <vxml version="2.0"> <!-- Exercise 7.4 --> <form> <field name="UserName"> <prompt>Say your name</prompt> <catch event="noinput" count= "1"> <prompt> Sorry, I did not hear you, please say your name </prompt> </catch> <catch event="noinput" count= "2"> <prompt> I need your name to access your file, please say your name </prompt> </catch> <catch event="noinput" count= "3"> <prompt> Even though I am a computer program, I can understand you when you speak. please say your name. </prompt> </catch> <catch event="noinput" count= "4"> <prompt> It is really important that you speak your name now. </prompt> </catch> <catch event="noinput" count= "5"> <prompt> My name is Computer Application number 413, but everyone calls me Joan. What is your name? </prompt> </catch> <catch event="noinput" count= "6"> <prompt> Goodbye </prompt> <exit/> </catch> <grammar type="application/grammar+xml" version="1.0"> <rule id = "aUser" scope = "public"> <one-of> <item>fred</item> <item>sam</item> </one-of> </rule> </grammar> </field> </form> </vxml>

7-5 Many proper names are difficult for a speech recognition engine to understand. In these cases, many developers record the name for later manual transcription. Write a VoiceXML form field that solicits the caller's name. Include 3 nomatch error handlers, with the third error handler recording the caller's name.

<?xml version="1.0"?> <vxml version="2.0"> <!-- Exercise 7.5 --> <form id = "form1"> <field name="UserName"> <prompt>Say your name</prompt> <catch event="nomatch" count= "1"> <prompt> Sorry, I did not understand you, please say your name again </prompt> </catch> <catch event="nomatch" count= "2"> <prompt> Sorry, I still did not understand you. What is your name? </prompt> </catch> <catch event="nomatch" count= "3"> <goto next = "#form2"/> </catch> <grammar type="application/grammar+xml" version="1.0"> <rule id = "aUser" scope = "public"> <one-of> <item>fred</item> <item>sam</item> </one-of> </rule> </grammar> <filled> <submit next="write_answers_to_disk.cgi" namelist="UserName" /> <exit/> </filled> </field> </form> <form id = "form2"> <record name="answer_1" maxtime="10s" dtmfterm="true" beep="true" finalsilence="1s"> <prompt> This time I will record your name. Please say your name again </prompt> <filled> <submit next="write_answers_to_disk.cgi" namelist="UserName" /> <exit/> </filled> </record> </form> </vxml>

a