02/23/19 FOR THE DISABLED, A NECESSITY

Return to main story

FOR THE DISABLED, A NECESSITY
Cheaper speech-recognition technology is a godsend for these users
Donna Davis was a vice-president at a New York shirt manufacturer in 1992 when tragedy struck. At 47, she suffered a stroke that paralyzed the right side of her body and ended her career. Worse, the stroke caused a terrifying condition called aphasia, which erases a person's language skills. She couldn't speak or write. She couldn't even construct sentences in her head.
Last year, Davis found hope. She started using a language-therapy system that shows pictures on a computer screen and asks patients to describe the events depicted by speaking into a microphone. The core of the Unisys Corp. system is its ability to give patients feedback on whether they are putting sentences together correctly, a capability known as natural-language processing. After less than a year of the therapy, Davis has regained some of the basic language skills she lost. She can slowly pronounce single words, construct simple sentences, and someday, she may be able to converse again. The therapy ''is the high point of her life right now,'' says her husband, Art Davis.
For tens of millions of people with disabilities, speech-recognition technology is more than a new toy or even a productivity tool--it's a way to profoundly change their lives. Those who suffer from blindness, dyslexia, and paralyzing conditions such as aphasia are starting to use speech recognition as the passkey to better communication and independence. Dyslexics, for example, can have computers read to them so they don't have to ask others for help. The deaf are using computer simulations of exactly how a mouth moves when it pronounces words so they can speak more clearly. ''It's a way of recapturing lost-world citizens,'' says Thomas A. Furness, director of the Human Interface Technology Labs at the University of Washington.
The potential impact, in economic terms, is huge. Approximately 26 million Americans receive some form of disability assistance, at a total cost of $200 billion a year to the federal government. Better speech-recognition technology could make it possible for some of these people to be more productive, saving billions. ''If we could provide access to just 1% so they could make their own way and return to the workforce, the (savings) would be big,'' says Furness.
But numbers tell only a fraction of the story. The example of Stephen Hawking, the British theoretical physicist and best-selling author of A Brief History of Time, hints at the untold contributions that disabled persons can make. Hawking, who suffers from a degenerative neurological disorder called amyotrophic lateral sclerosis (ALS) or Lou Gehrig's disease, has used a computer-synthesized voice developed by Digital Equipment Corp. to communicate as his muscles have deteriorated. With this, his writings and lectures have continued to further the understanding of the dynamics of the universe and the nature of black holes.
Disabled people have been some of the earliest adopters of speech technology. That's partly because their disabilities gave them a powerful reason to put up with the quirks and inconveniences of early products, such as the requirement that users insert a pause between each word. Still, the expense and customized engineering required to develop sophisticated systems have long kept them beyond the reach of most--until now. As PCs have become more powerful, speech-recognition systems have found their way to desktops, slashing costs. Hawking's DECtalk system, for example, cost $4,500 in the early 1980s but now goes for $200.
''A PILL.'' Consider the case of Dave McCrossen, 54, of Canandaigua, N.Y. McCrossen had tried to go to college six times since he graduated from high school 33 years ago, but was held back by severe dyslexia. ''Each time I tried, I couldn't keep up with the workload,'' he says. Last September, McCrossen enrolled as an undergraduate at Empire State College, using a $2,000 reading machine developed by Kurzweil Educational Systems Inc. (KESI) for people with dyslexia and other disabilities. The system uses text-to-speech software that highlights words as it reads them. By having the computer read to him, McCrossen says he is able to retain enough to keep pace with his assignments. It's ''a pill for my dyslexia,'' he says.
Not surprisingly, the newest frontier for voice technology is the Internet. Chieko Asakawa, a blind researcher at IBM Japan Ltd., is part of a team working on a browser that opens up the vast World Wide Web to the blind. Released six months ago in Japan, IBM's home-page reader uses a computer-synthesized male voice to read normal text in Japanese and switches to a female voice when it encounters highlighted hyperlinks to other sites on the Web. Programming the numerical keypad on a computer keyboard can enable blind users to move from line to line on a Web page and navigate backward and forward the way Web surfers do with ordinary browsers. And a scroll feature lets blind users speed up the voice. Some 2,000 Japanese are now using the browser. IBM is working on an English version it hopes to release later this year.
The technology also is helping the deaf. At the Tucker-Maxon School for deaf children in Portland, Ore., students are engaged in a novel experiment to learn how to pronounce words from a computer-generated talking head called Baldy. The product of Fluent Technologies in Beaverton, Ore., Baldy is a simulation of a wire-frame model that is covered with more than 900 tiny triangles composing his skin, eyes, nose, and--most important--his mouth. Despite his lack of hair and his resemblance to a mannequin, Baldy is one complex guy, representing 20,000 lines of software code. His mouth accurately enunciates words, a critical aid for deaf people learning to speak.
Baldy's most arresting feature is his ability to replicate facial expressions appropriate to the words he uses. The importance of visual cues in understanding speech was brought home to the developers at Fluent four years ago when they were negotiating to license the technology to Lucent Technologies, then part of AT&T. An AT&T executive vice-president watching a demonstration of Baldy praised the team for making great strides in improving the synthesized voice. But the voice was unchanged from earlier demos--the only addition was Baldy's repertoire of facial expressions.
HISTORIC MOTIVE. Since December, the deaf students at Tucker-Maxon have been working with Baldy to extend their classroom training in speech. Chelsea Angelina Crump, a fifth grader, sits in front of a topographical map on the computer screen that traces a river from its source to the ocean. An arrow on the map points to different features, and in the upper left-hand corner of the screen, Baldy asks: ''What land form is this?'' His voice sounds robotic, but the children look at Baldy and read his lips. Chelsea nails nearly every question, speaking her answers into a microphone. If the pronunciation is off, Baldy repeats the answer slowly and breaks it down into syllables. Chelsea then repeats the word until Baldy tells her it's right. ''The big challenge for deaf kids learning to talk is getting lots of good practice with feedback,'' says Patrick S. Stone, executive director of Tucker-Maxon. ''With this computer program, we've added another discriminating ear.''
It's often forgotten that Alexander Graham Bell was laboring to translate words into images for his deaf wife when he discovered the telephone. Does the future hold any other revolutionary technologies devised initially to help disabled people? Certainly some of the technology under development for them could be adapted for wider use. For instance, devices capable of converting speech to text, and vice versa, in real time, provide deaf people with a way of speaking over the telephone. But such a device might also help with simultaneous translation of foreign languages. Computer-screen cursors that follow the path the eyes take over a document started out as a project to help physically disabled people use computers but may someday provide an alternative to the mouse.
In the meantime, small revolutions are occurring every day. Deaf children are learning to talk. Blind people are starting to surf the Web. And some day, Donna Davis may be able to tell us all about herself.
By Paul C. Judge in Boston, with Seanna Browder in Portland, Ore.

Return to main story