How VoiceXML Works
VoiceXML is deceptively simple. Basically, when you build a VoiceXML application, you're creating a tree that guides a caller through a series of questions mostly requiring a simple response. The application converts the responses (using voice-recognition technology) into text, leading to the next level of the tree.
Because VoiceXML is XML designed for a specific purpose, you can either program it directly using an XML editor or combine it with a Java wrapper. For example: There are a number of development tools that generate VoiceXML applications, including Cafe from BeVocal.
The Future of VoiceXML
The VoiceXML Forum continues to enhance VoiceXML. Work on version 3.0 has been going on for months. The next direction for VoiceXML applications to move is toward location-based services. TellMe currently offers localized information based on the Caller ID of the caller. In the future, data will be localized based on GPS or E9111 positioning data.
But the real challenges to VoiceXML won't take place on the language-development side, but rather how society feels about providing information to a machine over the telephone. There are still many people who feel awkward about having a "conversation" with an anonymous voice over the telephone, and for those people a human voice will be the only interaction they want. In the end, VoiceXML applications won't completely replace the call center but they certainly can become an important part of customer service.