VoiceXML: What's Everyone Talking About?
VoiceXML can help that automation take place. Though not a new technology, VoiceXML is gaining steam as more and more corporations shift to an IP-based infrastructure. Basically, VoiceXML is an XML grammar (and not specifically a language) to create applications that make Web-based information accessible to users via voice and telephone.
Connecting the Dots
There have been many efforts to bridge the gap between computer and telephone-using consumers. The greatest distance is between WAP (Wireless Application Protocol), a text-based technology, and VoiceXML, which use voice technology. In some ways the two technologies are similar in terms of how they are implemented, but while WAP is purely text that's appropriate for a PDA or cellular phone, VoiceXML builds on a familiar ground already used in the customer-service world: voice. There have been proprietary voice-recognition systems (or Interactive Voice Response systems, as they were once called) used for decades now, so customers are familiar with them. VoiceXML builds on this popularity with tools to bring Web content directly to the consumer.
Let's say you're calling Delta Airlines on its Song low-fare subsidiary, and you need to either make a reservation or change one. Using the 800/FlySong line, users can find flights, check schedules, compare fares and track baggage. Most airlines still requires assistance from a human customer-server rep at some point in these processes, but the Tellme Networks setup delivers almost all relevant information via VoiceXML.
Similarly, we're seeing voice portals some national, some localized rise as a business model. Most of these voice portals are powered by VoiceXML to some degree. Basically, these voice portals empower users to access Web data (like listings, stock quotes, phone numbers, and directions) via a natural-language interface. The numbers are with voice portals: There are more than a billion telephones in use around the world, while Internet access is limited to only 250 million PCs or so. And while PDAs may someday displace cell phones to an extent, it's still hard to imagine a scenario more user-friendly for a consumer than being able to call in a question to a VoiceXML-based system.
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.