Date: Thu, 07 Mar 2002 12:24:16 -0500 From: "Samuel W. Heywood" <[EMAIL PROTECTED]> Subject: Re: Do screen readers stumble badly over misspelled
On Sat, 02 Mar 2002 08:41:38 +0500, Jude DaShiell wrote: > Sam, this is very interesting. If I'd been working at the U.S.G.S. I'm > pretty sure they'd still be using spech recognition today. The reason is > there's a solution for the similar sounding words problems, actually a > couple of them. First and simplest is a type of command to the system > through another computer program that translates numbers a person speaks to > letters. Each number is always two digits long. The second solution would > be a numbered menu of similar sounding words the system couldn't learn in > software. Run the program by voice command that it manages and speak the > number and the program writes the similar sounding word where you need it > and exits. Of course, you'd have to have a cheat sheet around so people > could look up the numbers, but operators would get accustomed to the list in > time and know the numbers. You are right about the ways we could go about solving the problems. We were well on our way in the process of discovering solutions along the lines you have mentioned. As we moved on to more advanced digitizing technology we found it no longer necessary to continue our experimentation with speech processors for the purpose of attaching attribute tags to lines. In the early days of the Altek digitizers we did what was called in the trade as "blind" digitizing. The operator was a sighted person of very sharp visual acuity but he could not see the results of any of his work until after all the data which he had collected onto magnetic tape was later fed into a line plotter for producing a hard copy version of the work. There were no CRT screens in those early days to view what we were doing in real time. Later on we started to use systems with real-time CRT displays and we tagged the line attributes by use of a mouse-like device using menu tablets and color codes. If we were to have continued doing "blind" digitizing I think we would have perfected the speech processor technology which was being developed for this purpose. The main disadvantage of the so-called "blind" digitizing is that the operator is not able to see his mistakes and correct them immediately along every step of the way. In order to do that he needs a real time CRT display interfaced directly to his digitizing table. Sam Heywood -- This mail was written by user of The Arachne Browser - http://arachne.cz/ ******************************************************** To unsubscribe from this list, send a message to [EMAIL PROTECTED] with the single word Unsubscribe as the subject. You MUST use the same address with which you subscribed! ********************************************************
