dannye recently asked me about my experiences with voice recognition, and at the same time, I noticed that Dragon NaturallySpeaking had been cooled.
This is a different viewpoint regarding the software.
The write-up about Dragon NaturallySpeaking presents a fairly optimistic viewpoint of the only readily available voice recognition software at the
moment. Granted, that article was written in 2002. Having been a user since that time, I would like to point out after nine years of use that in
reality voice recognition software (for the purposes of this article, I'm using this as a synonym for Dragon, since it really is the only easily
obtained, mainstream solution) is far more cumbersome and unwieldy than this article might lead you to believe. Here are some specific examples:
In the past couple of days, the server migration caused the user list to not empty correctly. Using voice recognition, I had begun an article detailing
this, explaining why it had happened (to the best of my ability), listing the total number and names of users that appeared, and documenting how it
encouraged me to create some e2 propaganda. Now for a little background, at home I use version 6.0 of the software because the new versions require
online authentication, and all of the machines that I have put it on did not have Internet access. For this reason I cannot justify spending an extra
$200 to support different versions of software on different machines at my house. At work, I use version 10.0 because that is the earliest version
compatible with the newer flavors of Windows. The result is that my functionality is effectively limited to what was available nine years ago on my
Windows XP box at home. (For those of you who cringed when I mentioned Windows, I'm sad to report it is my only option. There is no such thing as a
*nix version of the software (to the best of my knowledge anyway)).
Typing up paragraphs is much slower than the previously referenced article indicates. So it was taking a little while to get the article together. In
the meantime, another user had already created a similar write-up. This flushed quite a bit of work down the can for me. (Please note, this is no
condemnation of that particular user, it is really great that they wrote what they wrote. I'm just explaining why voice recognition can be so
To give you some idea, the previous three paragraphs that I just dictated have taken about 20 minutes. And that is a record speed for me. Think to
yourself, how long would it take to actually type out those three paragraphs? < 1 minute? At most maybe two minutes? Additionally, every single
sentence in those three paragraphs has had to have a correction. And that is after nine years of using the software. Nine fucking years! (Just
writing those last three words has taken a couple of minutes because the software refuses to use profanity of any kind.)
Do you get colds often? Or have allergies? Guess what, the software will no longer recognize your voice. Every time you're congested, and having a
2-year-old daughter this is fairly frequently for me, the software struggles to recognize your speech, no matter how thoroughly you have trained it.
And this is just dictation. What is never mentioned when discussing voice recognition is the onerous burden that comes with it. The quality of your
microphone is far, far more important than the software distributor generally indicates. The quality of the sound card in your computer, even more
important. At work, I actually have a microphone that costs over $200. The quality of my dictation is noticeably better. At home, I have a 30 or $40
microphone because I just do not use the computer as much here. And all of these microphones have cables, which continuously tangle with my office
chair, my pen tablet (another requirement for using a computer because of my condition), and the keyboard. This microphone has to be worn at all times
when using a computer. Many, many times people have come to my desk at work and left because they assume I am on the phone. Even people who know that
I use voice recognition are hesitant to approach me when I'm using it because they know that I have to take a moment to turn it off.
Voice recognition software does not play well with others. Anything Web-based (including our little corner of the Internet here) is almost completely
off-limits if it uses text fields or text areas. Yahoo mail? Not going to happen. I do all of my dictation into Notepad, then cut and paste it into
text areas in whatever web site I am working. It is related to some kind of memory bug (or maybe buffer is a better explanation). And that is on the
newest version, if you're a poor sucker like me using older versions things are even worse.
At work, I am forced to deal with a lot of PDFs. This is a gargantuan pain in the ass, because another bug occurs whenever voice recognition is running
and Adobe Acrobat is initialized in any way (including through a Web browser). The instant that Adobe loads, voice recognition fails and, no
exaggeration here, the only way to recover is to completely shut the computer down. I have repeatedly asked our system administrators about fixing this
and to date there has been no success.
While the article this was written in response to might be overly optimistic, it would be fair to say that this response is overly pessimistic. This
grows from years of frustration. Because of my personal history, I am trapped in this odd Catch-22. I am the first generation of personal computer
users, having started with Commodores and Apples in the early 1980s and working through early BBS and telnet access to the Internet up to the Tech
explosion of the 1990s. Much of my entertainment and interest in the first two decades of my life was centered around personal computers. Then I
developed Thoracic Outlet Syndrome, and was no longer able to use them or type or click a mouse or any other mundane computer task, until I began using
voice recognition. But computers are so prevalent, it is almost impossible to get a decent job without using computers in some way (even cashiers at
gas stations use a computer all day). To do tasks I need to do for work, I must use voice recognition. But using voice recognition is monumentally
frustrating. So I apologize for the overly negative tone, but there are definitely drawbacks to using the software.
Anyway, now I am hating on computers again. Enough of this, I'm going camping. I will check in later in the weekend using an Ipad, which means that I
will be able to read messages but not really reply to any length.
Total time to submit this: 1 hour 30 minutes. Ridiculous.