Intel's device for the blind or visually impaired combines a digital camera, OCR, and speech synthesis, with impressive results
Great advances in digital technology shower all kinds of benefits on society—among them, life-altering opportunities for people with disabilities. Innovations of this order range from digital hearing aids and cochlear implants to computerized prosthetic limbs. Unfortunately, such products are often relegated to a ghetto known as "assistive technology," where high prices and limited distribution place them beyond the reach of many people who might benefit.
Intel (INTC) is trying to change that as part of a major corporate push into health care. Its new Digital Health Group has just come out with a remarkable product, the Intel Reader, designed to make printed matter accessible to millions of people who find reading difficult or impossible, either because of vision problems or learning disabilities such as dyslexia.
The Intel Reader succeeds because it does such a good job of integrating three well-established technologies. It uses a digital camera to capture printed material, optical character recognition to convert print to computer-readable text, and speech synthesis to turn the text into comprehensible spoken language.
The Reader is a 1.4-pound unit about the size and shape of a paperback. It is designed from the ground up for use by the visually impaired. All menus are presented in both spoken and visual form—displayed on a 4.3-inch black-and-white screen that can magnify type up to about an inch tall. The buttons that control functions are large and easily distinguished by their shape and placement. At $1,500, the Reader is expensive by consumer-electronics standards, but relatively cheap for the assistive technology market, where low volumes keep prices high.
The primary way to get content into the Reader is by photographing pages using the built-in 5-megapixel camera and flash. A copy stand called the Portable Capture Station ($400) makes it relatively fast and easy to scan large volumes of text, especially from books. As soon as the words are captured, the Reader begins the text conversion, and you can have it start reading aloud while the conversion is still in progress. A USB port lets you import any text file—for example, books from Project Gutenberg, a community-based project that makes texts in the public domain available for free download.
The text-to-speech synthesis is surprisingly good. The voice is robotic, but with generally correct pronunciation and good inflection. You can set the reading speed as high as 500 words per minute. I found anything faster than 250 incomprehensible, but a listener's ability to understand very rapid speech improves with experience.
Scanning in printed material also requires some practice. Although the software is supposed to compensate for the curve of pages at the binding, you must learn how to position the camera properly, even if you are using the copy stand. The simpler the book's formatting, the more likely the process is to succeed. Footnotes, for example, can confuse the Reader, and technical material, such as mathematics notation, tends to produce garble rather than formulas. (Intel says this is a failing of the OCR software.)
Despite the limitations and its steep price, the Reader is an important step toward making adaptive technology more accessible. Let's hope it is the first of many.
This column marks my farewell to Tech & You after 15 years. I left BusinessWeek on Dec. 1 upon the magazine's acquisition by Bloomberg. It has been a wonderful ride, but it is time for me to move on. I will especially miss the many contacts I have had with readers over the years, and I would love to hear from any of you at firstname.lastname@example.org. For news of my next move, please follow swildstrom at Twitter.
For past columns and online-only reviews, go to businessweek.com/go/techmaven/