VoiceXML Review - Letter from the Editor

Volume 3, Issue 2 - March/April 2003

March 2003

Dear Readers,

A colleague just reminded us that it was 4 years ago yesterday (March 2, 1999), that AT&T, Lucent and Motorola announced the creation of the VoiceXML Forum. (IBM joined as a founding member shortly thereafter.) It's hard to imagine that four years have gone by so quickly! Yet, at the same time, when we survey how far things have come along since the Spring of 1999, then it seems like a long time! At that time, VoxML was in its infancy, and the 0.9 VoiceXML specification didn't appear until months later. Now in 2003, the VoiceXML 2.0 specification is a W3C candidate recommendation, and there are literally hundreds of companies offering VoiceXML-related products and services! If you are interested in the history of VoiceXML, you might want to take a look at the recently updated introduction to the tutorials on www.voicexml.org, written by Jim Ferrans.

The VoiceXML Forum intends to celebrate the arrival of Spring with its annual Spring Users Group Meeting, being held in conjunction with AVIOS / Speech TEK Expo 2003. The meeting will be held on Thursday, April 3 at the Fairmont in San Jose, California. The meeting will feature both a business and a technical track, so there is something there for everybody. We have assembled an excellent program of speakers, who will be debating and presenting topics ranging from Open Standards IVR - The CPE vs. ASP Debate to VoiceXML's future role in X+V (multimodal!) Also featured will be a variety of live demonstrations, including commercially deployed VoiceXML applications, as well as cutting edge X+V multimodal tools and applications. In addition, key members of the W3C Voice Browser working group will be present to provide an overview of what's going on within the standards area. You simply can't afford to miss this event. Be sure to select the VoiceXML User Group Meeting option when registering for AVIOS.

This issue of the VoiceXML review is rather special, in that we are going pause and take a close look at a variety of non-commercial VoiceXML projects that have sprouted up on various parts of the globe. Some of these projects are open source, while others are open binary. All of them demonstrate the advantages of using VoiceXML in solving real-world problems. They also confirm the fact that it's possible with a small group of bright people to build and deploy a real VoiceXML platform without requiring huge budgets and armies of programmers.

Pavel Cenek, previously with the University, Brno, in Czech Republic and now with the Norwegian applied research institute Norut IT, brings us up-to-date with his project: Elvira. Elvira is a flexible and extensible implementation of VoiceXML that Pavel and his colleagues have put together for their research in human language technologies. Elvira's component-based architecture has made it particularly useful in extending VoiceXML to handle statistic gathering, dialogue strategy evaluation, and exploring multimodal interfaces.

This month, our faithful First Words columnist Rob Marchand takes us through the handling of complex recognition results in VoiceXML. However, true to his column's moniker, Rob manages to make sense of these rather recent changes/additions to the 2.0 specifications without getting you lost in the forest. By the way, Rob happens to be the e-zine's longest standing columnist. If you are just getting started in VoiceXML, I'd encourage you to go back through the archives and work your way through the articles Rob has penned over the past couple of years. Rob's columns build quite nicely upon each other and it makes for an excellent tutorial.

The PublicVoiceXML project, recently referred to as the "jBoss of the VoiceXML World" by one observer, is an open source VoiceXML implementation, funded in part by the Information Society Technology Program of the European Commission. Dr. Roland Alton-Scheidl brings us an overview of this remarkable project (the platform was used to report the general election results in Austria last fall!) Of particular interest in this article is an insightful discussion of the open source concept, its potential economic impact, and its importance within the European community. Also noteworthy, is the team's intent to submit an implementation report to the W3C based on the recently published Implementation Report tests.

The third article we selected for publication in this issue is the Open VXI open source VoiceXML implementation offered by SpeechWorks in partnership with Carnegie Mellon University. This open source VoiceXML project, as far as we know, has been around the longest, and is one you've likely heard about. The article, written by Brian Eberman, gives an excellent architectural overview and then dives into detail regarding how one goes about integrating the interpreter into the speech resources and call control subsystem. Plans are underway to evolve a new version that implements the current VoiceXML 2.0 candidate recommendation.

In our First Words column, Rob Marchand polishes off his ongoing discussion on handling complex recognition results, showing how a form level grammar can be used to introduce mixed-initiative dialogues, vs. the more elementary directed dialogues discussed so far.

We get a fair share of questions from folks about how you to do task "X" or "Y" when writing a VoiceXML application in JSP, ASP, or one of your other favorite web development paradigms. The simple answer is of course, that you do it the same way you do it when generating markup for your favorite HTML browser, only now you generate VoiceXML markup! Nevertheless, this month, Matt Oshy tackles one of these standard "meat and potatoes" web development questions (i.e. how do I access data from an Access database for my VoiceXML app?) with vengeance. This seemingly innocent question was all Matt needed to unleash his arsenal of finely honed XML skills in a frenzy of creative coding. From XML to XSLT, to VoiceXML and SRGS, along with a pinch of JavaScript (or should we say JScript?) - you'll find it all here!

Sincerely,

Jonathan Engelsma
Editor-in-Chief
VoiceXML Review

back to the top