Open sourcerers tweak Linux for access
The recent commercialization of Linux has brought with it mass appeal, with its open-source status allowing those masses to more easily share tools and solutions. But ease of use is a different issue for the nation's 54 million disabled citizens, and accessibility is a somewhat complex proposition to define.
Determining what people want and reasonably expect from "accessibility" is something that IBM's T.V. Raman thinks a great deal about. Raman, who lost his sight as a teenager, developed what has become a standard text reader while a graduate student in the computer science department at Cornell University. That program, Emacspeak, permits blind users to write and send e-mail, surf the Web and do most things a sighted user would do.
Raman, an ardent open sourcerer, wrote the system to work over a Linux platform running on any low-end PC. Indeed, Linux is an excellent, if largely untapped, format for the disabled, particularly the blind, he said.
"Something like Emacpseak would not work in any other form. I could not have written it if I couldn't see other people's code," said Raman, an engineer who focuses on speech technology at IBM Research (San Jose, Calif.).
Perhaps it is his faithfulness to the open-source ethos, but Raman, who left Adobe to work on speech technology at Big Blue, needs prodding to accept that Linux appears headed toward the land of shrink-wrapping. "The commercialization of Linux is happening," he conceded. No longer an ugly duckling for computer cognoscenti, the platform is now "slick, stable and getting easy for Grandma and Grandpa to use."
Which raises some challenging issues for what it should allow Grandma and Grandpa, or anyone else with special needs, to do. "What does it mean to make it easy to use for people with low vision, or hand problems that make it hard for them to control the mouse?," Raman asked. "These are very different questions from, say, what will it take to get that Windows software to run on Linux. Even if you got them running on Linux, what then? The screen reader throws one more wrinkle into the whole mess."
Fortunately, Linux's greatest asset is its flexibility. "It's not a one-size-fits-all world, and you don't have to build all of the solutions on top of the same thing. Linux is a different size for everybody," Raman said.
Despite its egalitarian roots, even Emacspeak wasn't a tool for every man (or woman) at first. It required a voice synthesizer priced at the time in the neighborhood of $1,000. Last year, however, IBM released its ViaVoice software, which eliminated the need for synthesizing hardware. Within a week, Raman released, over the Internet, a free version of support for the new program.
In June, IBM issued ViaVoice for Linux, making Big Blue the only speech vendor in the language. Tom Houy, manager of client systems marketing for IBM's voice systems, said the program differs a bit from its Windows version, which has been out for five years and has some 12 million users, including not only the blind but people with carpal tunnel and other conditions that make mice and keyboards inhospitable.
ViaVoice might have reached Linux earlier, Houy said, but for the unavailability of audio drivers. "A year ago they were lacking," he said, but in the last year the open-source community has solved that problem. Once it had a driver, it took the company just two months to turn out a Linux-based edition of the screen reader.
Windows and Linux users don't exactly have parity when it comes to using ViaVoice, however. "ViaVoice for Windows can command and control the complete desktop. But [in Linux] there really isn't any solid desktop," said Houy.
While blind users are obvious beneficiaries of the speech technology, Houy believes there's a far larger group that will need some form of screen readers during a portion of their normal day. Those with fair sight may use it while their minds are supposed to be on more important matters, like avoiding accidents. "You probably have about 100 million visually impaired people coming into the world: drivers. They'll want to be read to, not to have to read," he said. "I'm not going to be wanting to be looking at the data coming onto my dash" while hurtling down the highway.
Voice software offers "situational transparency" across user interfaces technospeak that basically means your desktop PC, personal digital assistant, car and cell phone talk to you in the same way.
Raman believes PDAs are a perfect venue for Linux. "You can mix and match applications instead of having a big white elephant simply because the person who is selling it is selling a big white elephant," Raman said. "The P in PDA is a much bigger P [than in PC]. There is more room for building a PDA optimized for different methods of use."
Screen readers and other accessibility tools for the blind haven't been, at least until lately, on the top of any company's to-do list. Raman points to Microsoft Corp. as an example. The first application that read back to blind users ran on DOS, at a time when "the vision interface was poor for everybody, so in that context reading the screen made sense."
As screens became graphically richer in the early '90s, the DOS program became less and less useful. Smaller companies approached Microsoft about developing better screen readers, but were "told to take a dive," Raman said. Only in 1995, he said, in the face of mounting political and consumer pressure, did Microsoft do an about-face.
Windows users "have been playing catch-up ever since," he said. "And in my opinion it's a losing battle."
Feels like Unix
Curtis Chong, director of technology at the National Federation of the Blind, said Linux has another asset: It feels a lot like Unix, and "in days before, a lot of us would use programs to log into a Unix shell."
In Chong's view, the advances in Linux accessibility are a positive development, but variety in the operating system industry isn't necessarily a good thing. "We usually do not greet new platforms with optimism, because the designers haven't given accessibility a thought, and you have to wait for a third-party industry to spring up to generate [tools]." Indeed, the proposed breakup of Microsoft could spell trouble, Chong said. "I would not want Microsoft producing more than two or three platforms. It would be a disaster for the blind community."
As Chong fishes out phone numbers and URLs from his PC, another man's voice can be heard in the room. It is the sound of Jaws, the speech program that allows Chong access to the Windows basics: Word, Word Pad, Outlook Express and the calculator. The metallic words tumble out of the keystrokes like an anthropomorphic pipe organ.
"Every time a new computer comes out, we mostly know we're the last in line [to be able to use it]," Chong said. "It was a huge hurdle for us to be able to get into Windows. We had our first real access to Windows 3.1 in 1993, but most of us didn't migrate energetically into it until 1996 or 1997."
The blind community is paying particular attention to the explosion in PDAs, Web phones and Internet appliances, none of which is user-friendly for the sight impaired. "There is no technology that will allow me to use a Palm Pilot," Chong said. "The blind are not on anybody's radar screen until after this whole edifice is built."