Talking to Your Computer

Remember HAL, the talking computer that began to feel human emotions in “2001: A Space Odyssey”? Well, even though the year is 2001, we haven’t quite met up with HAL yet. Filmmaker Stanley Kubrick was a little ahead of himself.

But we’re not that far away from conversing with our computers — a scary thought, given some of the things I’ve said to my computer on occasion. I would hate to hear its retort.

Research in the field of speech technology has made great strides, and a variety of new and improved voice applications are beginning to hit the market. These include old standby applications that now allow you to use your voice for commands, voice portals that let you use any telephone to access your e-mail and a multitude of information on the Web, and call centers where you have a conversation with a computer that helps you check on your order or get your flight status.

Here in Southwest Florida, Sprint customers may soon have access to a voice portal. Sprint recently closed a deal to provide voice portal service to its wireless customers and may extend that service to its landline customers by year’s end.

All of these applications permit new, ubiquitous ways to access the World Wide Web. You no longer have to be tied to your keyboard to surf the Internet — and for e-commerce, this means opening up a much wider market. That’s the type of news companies want to hear these days.

Technology analysts are estimating huge global expenditures on speech-related services: $41 billion by 2005, says The Kelsey Group, a Princeton, N.J.-based market research company. Datacomm Research foresees 2 billion people using various speech technology systems by 2005.

Up to this point, however, voice applications have had only isolated success. The key to mass consumer acceptance will come “when the computer can engage the user in a dialog,” says Gartner analyst Jackie Fenn. “The computer must know what it is looking for and ask you for what it needs. We need to get beyond voice just as a text-entry tool. We are four or five years away from that. But we will see little pieces of it before that.”

A short history: About the time HAL was taking shape in Kubrick’s mind, researchers at IBM were hard at work on speech recognition technology — that is, creating machines that could recognize human speech. (A bit of trivia: The letters that form the name of the “2001” computer, H-A-L, each precede by one the letters I-B-M in the alphabet.) IBM’s first consumer application of the technology was dictation software in which the computer transformed the spoken word into text. Other companies have brought similar software to the market in the past few years. Reactions to it have been mixed at best, as the software was not always easy to use and accuracy was hardly perfect.

That was about the extent of speech technology in the marketplace for a few years, though research continued behind the scenes by the likes of IBM, SpeechWorks, and last but not least, Microsoft. Now the technology has progressed from raw dictation software to what researchers call a “conversational interface” — using natural language with a computer that not only recognizes but also understands and responds to what you are saying. All of these companies, and many more, are now working on applications that employ this technology.

IBM has moved into the enterprise market offering the tools companies need to develop voice applications.

Boston-based SpeechWorks, whose products are based on spoken-language research conducted at MIT’s computer science lab, helps companies develop speech recognition systems for large call-center operations. For example, if you call to check on your flight status with United or Continental Airlines, you will be having a conversation with a SpeechWorks computer system.

These have come a long, long way from those Touch-Tone response systems we have all come to hate. Continental claims that its new speech recognition system has seen a 50-percent reduction in the number of people who opt out for a live agent. On some of these speech recognition systems, people have forgotten they are talking to a computer and start chatting about the weather. The systems even have interrupt capabilities so you can cut the computer voice off mid-sentence and it will still respond to you.

Microsoft, meanwhile, has announced the release of Office XP, which adds voice capabilities to its suite of products. Office XP offers a dictation mode for dictating letters, memos, and e-mail messages, and a voice command mode that lets users access menus and commands using a voice interface. Microsoft is also putting its considerable resources to work on a prototype handheld device that combines a voice and stylus interface.

The mobile marketplace, with its wireless handheld devices, is what’s driving much of the movement toward voice technology. One of the biggest challenges to using handhelds and wireless phones to access the Internet is negotiating those tiny keypads. “Obviously, these devices are a natural for the speech industry,” says Fenn.

People have gotten used to having nearly unlimited information at their fingertips on the desktop. Now they want it no matter where they are — and not just at their fingertips but at the sound of their own voice.

One result of this need for Internet access away from the desktop has been the rise of the voice portal. A number of companies are now offering telephone access to the Web. Just pick up any phone — wireless or landline — and call into a voice portal for news, weather, sports scores, stocks, television schedules, horoscopes, e-mail, directions, movie times... You say your request, “Give me the news,” which is relayed through speech recognition to a computer that accesses the Web for the information. The Kelsey Group predicts an $11 billion voice portal industry by 2005.

Several voice portals have hit the market in the past few months. AOL and Yahoo both provide their customers with phone access. AudioPoint, BeVocal, and HeyAnita are among those launching such services.

Perhaps the best-known application of voice technology is OnStar, that comforting voice in General Motors cars that can help you out in an emergency. Now OnStar is adding applications that will let you access information on the Internet by pushing a button on the steering wheel. IBM and Motorola have teamed up to do the same.

Through OnStar and other widespread applications, more and more people will get used to hearing these speech recognition systems and will realize their potential. Voice applications have come a long way from the first stilted dictation programs and monotone computerized voices. But have we found HAL yet? No, he must still be floating in space somewhere. Thankfully, he has not yet crept onto my desktop or he might soon learn a couple of choice new words next time my computer tells me, “This program has performed an illegal operation and will shut down.”

Susan Holly is a freelance writer based on Sanibel.