The purpose of VoiceXML is to present the caller - or user - with information, and prompt the user for a response. The
simplest example of this is a single form, with a single field. Like HTML, VoiceXML handles user input using forms,
however, instead of <input> like HTML, VoiceXML forms generally contain one or more <field> elements.
Like HTML forms, the data of a VoiceXML <form> can be submitted to a URL. The <submit> element can be nested
in a <catch> element or other similar event related elements, but for now we'll just place the <submit> element
inside of a simple <block> element so it is executed automatically.
Each <field> has a name attribute which defines how that field is accessed in the VoiceXML, and what name is used
when the data is submitted to a URL.
Inside the <field> you can nest a <prompt>, which - as you'd expect - prompts the user to provide a value for the
given field.
How does the user provide that data? User input is captured using VoiceXML's <grammar> element. The most basic
grammar is simple speech recognition. While we won't explore it now, each <grammar> must have a root rule defined
as an attribute, and a matching child <rule>. We'll define our grammar as being <one-of> a set of values, and
provide all the valid <items> as children. We'll also explicitly set the <grammar> mode attribute to voice,
limiting this to speech recognition only.
But what about those handy - likely now virtual - buttons a user can press? A <grammar> element is also use to capture
them. To support both speech and DTMF (dialed digits), we'll add another <grammar> element, this time using a mode
of dtmf. Like the last one, we'll define a set of <items> the user should be selecting <one-of>. However, we'll
add a <tag> element to control the value of the selection. Should a user press 1, we don't want to send that 1 to our
URL. Instead we want to send what that digit represents, and we can define that by setting the property of the out
object matching the name of our <field>.
Example: Full VXML
Now, once the user has provided a value for <field>, the <block> element is executed, and the data is sent to the
defined URL. That data can be evaluated and used to generate the next VoiceXML document.
Example: PHP
Want to dig a little deeper into VoiceXML? Check out the W3C's documentation of <form>, <field>,
and <submit>.