(Almost) Word-for-Word

As voice recognition programs improve, students reap the benefits.

Fans of science fiction author Isaac Asimov might remember his description of a futuristic voice recognition machine (for the true geeks out there, it was in his 1953 novel, Second Foundation). Asimov wrote that the machine, which was designed for students, turned out copy "in a charming and entirely feminine handwriting, with the most beautifully graceful capitals anyone ever saw."

Asimov may have been wrong — and perhaps a bit sexist — in assuming that future students would want a voice recognition system to churn out words in flowing cursive text. But his imagination proved to be not far off from the technology available to students now.

Voice recognition software is hardly new — attempts at capturing spoken words and turning them into written text have been available to consumers for about two decades. But what was once an expensive and highly unreliable tool has made great strides in recent years, perhaps most recognized in programs such as Nuance's Dragon NaturallySpeaking (which is now in version 9).

Adam Krass, assistive technology specialist for the Bergen County Special Services School District in New Jersey, uses voice recognition software (specifically, Dragon NaturallySpeaking) when working with some of his students. Having extensive experience with using these systems in the classroom, Krass says voice recognition is becoming an increasingly viable tool.

"It's unbelievable — it's so much better than it used to be," says Krass, who started working with voice recognition about 15 years ago. "Anyone can just pick it up and use it. Before, it wasn't an off-the-shelf product."

Other products in the market include SRI International's EduSpeak and Microsoft's Vista operating system, which will have a built-in dictation feature. (A live demo of Vista's voice recognition feature in July, however, went terribly awry; the program confused "aunt" with "mom" and then went haywire as the demonstrator attempted to fix it. To be fair, Vista is still in beta.)

Krass says voice recognition software can serve a vital purpose for students with physical limitations that prevent them from typing effectively and for students with dyslexia, for example. In Bergen County, voice recognition software has been paired with talking word processors such as Write: OutLoud and IntelliTalk 3.

Although he doesn't see voice recognition programs expanding much beyond a core group of users-in education, at any rate-Krass welcomes the continuing improvement of the systems. For example, these programs used to require extensive training of the software itself by the user, which meant hours of reading preselected text to the program to "teach" it to recognize the timbre and phrasing of the user. Newer programs require much less prep time.

"Typically anyone that can type even a little bit prefers to type rather than dictate," Krass says. "But the people who do well with this program really need it."

Mark Smith is managing editor of Technology & Learning.