We're still a few years away from being able to conduct two-way conversations with machines, according to experts. But voice recognition is proliferating in the areas of voice portals and personal assistant services for consumers and voice-activated dialing, automated attendant and telephony dictation for businesses.
"The Holy Grail of the speech recognition industry is to get to a truly natural dialogue with a caller ... and we're getting there," said Nancy Jamison, an analyst with Jamison Consulting.
Arial's long arm
One natural thing a caller might ask a live receptionist to do is look around for a person not at his or her desk.
Arial Systems is taking the automated attendant model in that direction. Using speech recognition technology from Lernout & Hauspie Speech Products, the ArialVoice browser enables callers to navigate a directory with their voice to get in touch with company employees anywhere in the office or at home via regular phone, mobile phone or pager.
ArialVoice recognises a caller's voice commands to search the ArialView online directory, a database that combines employee names, phone numbers and schedules with their current location and availability. It finds employees who have left their offices by tracking an infrared badge.
The system gives those receiving calls the option to take the call, hold the call until they're in a place to take it, let the caller leave a voice mail or direct the caller to other employees who can help.
"The purpose is to give the status and availability of the person you're trying to call and just facilitate the connection," said Roseann Sanji, director of marketing at Arial. "We dynamically route communications based on location, and we're using voice recognition technology to streamline the process. All of this is done to enhance the collaborative effort within an organisation."
ArialVoice enhancements will include enabling employees to update their location and availability via voice. Sanji would not say when this would be available.
Before the end of the year, Arial will also switch from an infrared badge to a smaller, lighter and less-expensive radio frequency badge that is even easier to pinpoint an employee's exact location, Sanji said.
L&H's big bet
One of the appeals of speech recognition software is that it raises the level of comfort for a business' customers without adding the cost of hiring a live person.
Providing an easy-to-use and approachable system was important to Eckerd Health Services, which last month began offering an automated voice-enabled prescription refill ordering system that it expects to be used by seniors and others facing serious health problems.
But just having the speech recognition in place is no guarantee that customers will use it. At Eckerd, an affiliate of Eckerd, 65 percent of customers calling into its call centre still use the touch-tone method to refill prescriptions.
L&H wants to expand speech recognition beyond customer call centres. This month it will announce a suite of telephony-based, Web-hosted services called iChart for the health care industry. Using iChart, doctors will use the phone to dictate patient medical reports and send them to a speech recognition server at a Web site, where they will be transcribed, edited or corrected, and coded so that other systems can extract data from the text files.
Future iterations of iChart will introduce natural language processing to extract and structure clinical data from narrative text, according to L&H officials.
For embedded devices, L&H has produced Project Nak, a prototype handheld computing device that uses voice as the primary user interface. The device, due in the first half of 2001, contains a large-vocabulary, continuous-speech dictation engine and a text-to-speech engine, which enables playback of text in a natural-sounding human voice. These engines will enable users to send and receive e-mail, search the Web and conduct e-commerce transactions.
Look ma, no hands
While the enterprise is looking to speech recognition to provide the friendliness of communicating with a live person any time at an affordable price, telephone carriers are betting that it will produce a new territory for providing billable services, say industry insiders.
Nuance Communications plans to sell its Voyager voice-enabled Web browser to wireless carriers but give away development tools and speech objects to the enterprise to create applications.
Voyager, which will enable consumers to navigate the Web using spoken hyperlinks, is now in beta testing and expected to be available in the fourth quarter.
Officials at the company envision a day when Voyager-like speech technology will let a caller pick up the phone, simply say the name of a company and get routed to that party -- all without touching the keypad.
Some companies are already headed in that direction. Tellme Networks, a voice portal, announced in late July the availability of its toll-free service, which allows callers to use simple spoken commands to find and connect to people, businesses and information across the United States.
"The direction in which this will evolve is integrating services together with one interface so that it flows seamlessly between voice dialing and [caller requests ]," said Bill Meisel, president of TMA Associates.
