Some Questions on VoiceXML 2.0

By Matt Oshry

In this monthly column, an industry expert will answer common questions about VoiceXML and related technologies. Readers are encouraged to submit questions about VoiceXML, including development, voice-user interface design, and speech technology in general, or how VoiceXML is being used commercially in the marketplace. If you have a question about VoiceXML, e-mail it to and be sure to read future issues of VoiceXML Review for the answer.

Q. I have numerous references to recorded audio on a different host than my VoiceXML content. I'm reluctant to hard-code the path to the audio server throughout my VoiceXML documents since the host may change. Does VoiceXML provide me with any alternatives?

A: Your best bet is to use the expr attribute on your audio tags and reference a variable declared at application scope. You initialize the variable to the base URI of the audio and use the JavaScript concatenation operator (+) to combine the variable with a static string that specifies the path to the specific audio file. If you move your audio to a different host, you need only update the application-scoped variable.

In the following example, the first document represents the application root. This document declares a variable, baseAudioPath, that specifies the base URI to your recorded audio. The second document references the application root document and contains a block of audio elements that reference the application-scoped variable.

Q. I read about the xml:base attribute in Section 1.5.1 of the VoiceXML 2.0 specification ( http://www.w3.org/TR/voicexml20/#dml1.5.1). Can't I just use the xml:base attribute to specify the base URI to my recorded audio?

A. You could use xml:base to specify the URI to your recorded audio, but there are significant caveats you should be aware of.

According to the specification, the xml:base attribute "defines the base URI for all relative URI references throughout the document." That means that, in addition to audio elements, your script, grammar, subdialog, choice, link, goto, object, and submit elements are candidates for xml:base resolution too if they specify relative URIs. Of course, fetchaudio and application attributes that specify relative URIs will also be resolved using the xml:base URI.

Here's a simple example:

   xmlns="http://www.w3.org/2001/vxml"
   xml:base="http://audio.acmeairlines.org/"
   application="root.vxml">

Let's say the document is hosted at "http://reservations.acmeairlines.org/index.vxml". When the interpreter parses and executes the document, it observes the xml:base attribute and resolves the application attribute of the vxml element to "http://audio.acmeairlines.org/root.vxml". If the application root document happens to be located on the audio Web server (unlikely), the interpreter continues to execute the document and resolves the relative URI of the first audio file to "http://audio.acmeairlines.org/welcome.wav". The second audio element gets resolved to something like "http://audio.acmeairlines.org/ads/ad7.vxml". The actual URI will depend on what number the user-defined JavaScript function GetAdIndex() returns. Finally, the interpreter attempts to navigates to "http://audio.acmeairlines.org/mainmenu.vxml", once again resolving the relative URI associated with the next attribute using the xml:base attribute.

In conclusion, if your audio is hosted on a different server than your application, you're better off defining a variable with that base URI at application scope and referencing it via the expr attribute of your audio elements.

back to the top

Copyright © 2001-2002 VoiceXML Forum. All rights reserved.
The VoiceXML Forum is a program of the
IEEE Industry Standards and Technology Organization (IEEE-ISTO).