Opening new portals for the blind

Free-range surfing
WebAnywhere, an Internet-based service released last month, boasts an even better price tag: free. The program’s innovation isn’t so much about what it does — no more than existing Web readers that convert written text to digital speech — as it is about its availability on almost any computer.

Screen readers normally require users to download software that connects to the Web server and converts text into speech from an electronically generated voice, compressing it into an MP3 file and playing it back to the person.

WebAnywhere instead processes the text on an external server before sending the audio file back to whichever Web browser is in use, thus allowing blind people to access unfamiliar computers in libraries, schools and other public places.

“All the individual parts had been designed, but hadn’t been linked up together before to come up with this,” said Jeffrey Bigham, a Ph.D. candidate in computer science and engineering at the University of Washington and WebAnywhere’s creator.

Once a user connects to the service, WebAnywhere can act as a search engine or as a conduit for a specific URL. By default, the system processes all of the text on a Web page. Alternatively, the software can anticipate what the user may want to read, retrieve it ahead of time, and cache it on the computer browser to minimize the time needed for the text-to-speech conversion.

This algorithm-driven shortcut, known as pre-fetching, can waste computer time if it retrieves the wrong text block. But by taking account of user preferences during the session, the program can get better at anticipating upcoming text blocks to minimize the latency, or the delay between a user’s request and the actual playback.

“You hit a button and expect it to play text to you. Any delay is really noticeable and really annoying,” Bigham said.

Reducing the lag time, in fact, is key to making the system fully functional, and Richard Ladner, Bigham’s thesis advisor, said other computer scientists were initially skeptical that a Web-based system could get around the problem. By incorporating pre-fetching, caching and other known concepts into the system’s design, however, he said Bigham was able to neatly sidestep the issue. “No one had put them together in this unique way,” Ladner said.

WebAnywhere’s current server can support 100 to 200 people simultaneously (based on current Web server limitations), though Bigham said he’s commandeered other computers in the department to help support the system. So far, early reviews have been generally positive, and speed hasn’t been an issue.

“People are really positive about this, they really see a need for this,” Bigham said.

In initial tests, blind Web surfers have successfully used the program for reading e-mail, retrieving bus schedule information and performing Internet searches. Ladner said he dreams of the program eventually being added to Google’s roster of free products as its first accessibility tool, an arrangement that would greatly expand WebAnywhere’s reach.

Already, inquiries have poured in from China, Portugal, India and other countries. With the release of the program’s HTML and Javascript source code, the researchers are hopeful that programmers can add some of the many extras requested by users as well as translate the software into multiple languages — further helping to break down many of the long-standing barriers for the blind community.

MORE FROM FRONTIERS

Tourist or terrorist? Computer knows

Tourist or terrorist? Computer knows Climbing no obstacle for snakebots The next 'American Idol'? Ask your computer How a deadly avalanche can be triggered Green alternative to plastics: liquid wood Getting lost for better architecture Turning air into drinking water Revving up the race for fuel efficiency How to be a disagreeable (but likable) robot Can concept of clean coal be salvaged?   Add Frontiers headlines to your news reader: