If you visit KurzweilAI.net, the Web site operated by writer and inventor Ray Kurzweil, you'll meet Ramona, his alter ego. Ramona is an animated avatar who speaks in surprisingly lifelike fashion, answering any questions you may have about the site or the day's news and weather. At the moment, she's little more than a novelty item, but the technology that powers her, provided by LifeFX and eGain, may soon reinvent the way you interact with the Web.
LifeFX and eGain are just two of several companies working to bring avatars like Ramona to everyday Web sites and turn them into virtual sales people and customer service representatives. An avatar-also known as a bot, a stand-in, or just a character-could potentially talk you through the purchase of an item on an e-commerce site or provide help when you're trying to use that item.
In theory, when interacting with such digital personalities, you'll be more comfortable and more content than if you were reading the text of a help file or exchanging email with some unseen customer service rep. "People are more engaged when they're interacting with a humanistic persona," says Michael Zapata III of Lipsinc, a company that develops this type of avatars. "And when people are more engaged, that increases customer satisfaction."
LifeFX, an outfit that has long developed digital models for the medical and entertainment industries, builds lifelike 3-D faces that, when fed audio, seem to speak that audio. Kurzweil's Ramona consists of a LifeFX 3-D head, backed by an artificial intelligence and speech engine provided by eGain. When you type in a question, the eGain engine generates an answer. As it coverts the answer to audio, the LifeFX face moves accordingly.
The audio is far from lifelike: It sounds much like a robot from a 1950's science fiction film. But the movement of the face is quite impressive. "It's about much more than flapping lips," says Lucie Salhany, copresident and CEO of LifeFX. "We have to simulate the movement of the muscle structure underneath the face to make it look right." Alternatively, LifeFX faces can be used with recorded speech, so the face sounds as good as it looks.
Open your facemail
Another LifeFX stand-in is being used by KiwiLogic, a company that develops natural language systems, but widespread use of the technology for customer service purposes is still months off. As a secondary business, LifeFX is offering an email client, known as facemail, that makes use of similar animated stand-ins. When a friend or colleague uses the client to open email you've sent, a stand-in of your choice reads your message aloud.
Currently, the text is converted by speech via IBM's ViaVoice technology and consists of seven or eight different stand-ins to choose from. By April, LifeFX and its new partner, Kodak, will, for a small fee, create a customised stand-in from photographs you provide. By next year, you'll be able to send recorded audio rather than relying on the text-to-speech engine.
Jackie Fenn, a vice president and research fellow at Gartner, agrees that LifeFX stand-ins could bring a more comforting feel to the Internet but says that it's a bit too early to speculate about when or where the technology will make a large impact. "It's still at a point where people are trying to figure out how to use this kind of thing," she says. "We're very much in the experimental phase."
She also questions whether the technology's audio component is appropriate for all environments: "It's fine if you're alone in your study, but if you're in an office full of people, you might not want all your business read aloud."
