As many of the members of the committee know, one of our main projects
is to develop a modern screen reader which will be even more useful than
the Speaqualizer was in its day. In pursuit of that goal, I have written a
non-technical description of the device a few of us envisioned during the
1995 National Convention in Chicago. Here is what we came up with. Please
comment, debate, point out errors, omissions, etc. as freely as you
desire.
-Brian
On the second Generation NFB Speaqualizer.
By
Brian Buhrow
ABSTRACT
As the number of "standard" computer data busses prolypherate, it becomes
harder and harder to design access hardware which is cheap, portable
accross various hardware architectures, and which provides enough access to
meet today's sophisticated access requirements. Because it has taken main
stream software developers years to develop the current generation of
hardware/firmware and software, and because designers of access
technology have always taken the tack of trying to design for the
"already-emerged" technology, access technology has always been more than a
few steps behind the currently available technology. This paper will
attempt to describe an alternate approach to the computer access problem
for the blind, as well as to describe a prototype piece of access
technology which follows that approach.
Although computer technology has changed drastically over the last ten
years, there is one piece of technology which has remained virtually the
same for more years than the computer, as we currently know it, has been
around. This technology is, of course, the video monitor, or, the CRT.
This technology, which is based on analog components, relies on analog
signals to tell the monitor where to traw the pixels on the video display.
There are a few well documented standards for interfacing monitors to
their display controllers, allowing multiple vendors to sell monitors and
display controllers which will interoperate.
With the rollout of the graphical user interface, (gui), the task of
getting the information displayed on the screen and converting it into a
format suitable for use by the blind has become increasingly complex.
Because many of the techniques used in getting information out of current
graphical environments rely heaveily on an understanding of the internals
of the operating systems on which those graphical environments exist, and
because operating systems are changing so quickly, it seems imparative tha
in order for the blind to achieve a reasonably secure level of access, a
technology must be developed which can work independent of the operating
system in use, the computer in use, or even the location in which the
system is used.
With the knowledge that, for the time being at least, computer systems
will continue to have as their primary interface, a visual display, and
with the knowledge that those displays will, for the most part, be operated
by a standard display controller, we can conceive of a device which, when
connected to the videeo monitor plug, would convert the analog signals
presented there into a usable representation of the data which is being
transmitted to the visual display. Using optical character recognition
techniques, as well as artificial intelligence techniques, one could turn
the digital pixels extracted from the analog signals coming off the video
port of the host computer into an intelligible collection of data which
could be presented to the blind computer user in much the same way as he or
she currently accesses ascii-based material. Depending on the
sophistication of the device, it could interpret visual windows, toolbars,
radio buttons, variable fonts, and all the other trappings of the gui
interface. In short, it would be a "screen reader" in a very real sense.
One could argue that this is wonderful science fiction, but consider
that we already have many of the pieces of this technology available today.
It is currently possible to purchase, from the mainstream market, video
capture boards which will digitize an analog NTSC television signal and
store it in the memory of a computer for later manipulation. Another
piece, optical character recognition technology, OCR, as it is often
called, is currently widely available and is quite capable of recognizing a
number of different type faces, fonts, and sizes. With the inexorable
unwinding of Moore's law of computers, which says that the speed of
computers doubles every 18 months, it will not be long before OCR
technology can be used in real time applications, such as the one we're
describing here.
This archive was generated by hypermail 2b29 : Sun Dec 02 2012 - 01:30:03 PST