NWA-PCUG Newsletter Article, April 2008
Look Ma No Hands
By Bruce Jacobs, Phoenix PCUG newsletter editor
http://www.phoenixpcug.org/
editor(at)pcug.org (click to email author)


Obtained from APCUG with the author's permission for publication by APCUG member groups.

This article was not typed. I dictated it using the speech recognition feature in Windows Vista Ultimate.

I have been using this feature on and off for certain applications for a while and it has its good points and its bad points.

The intent of the software is to allow the user to speak words into the computer and for the computer to recognize those words, understand that some of them are commands, and also understand that some of them are input to programs. In other words, when I say the words “please save me”, the computer must decide whether to add the words to the document I am editing as if I had typed them, or add the word “please” to the document, and then bring up the save dialog box to save the document with the name “me”.

The program that is attempting to do this has two major hurdles to conquer. It has to recognize the spoken words correctly. This is helped by the training exercises. But it has its limitations.

It must also understand "in context" which of those words are commands and which of those words are to be part of my document. As time goes on, you learn to pause before speaking a command. So in the above example of “please save me”, I would say the words altogether relatively quickly in order to enter them in my document. If I wanted to add the word "please" to my document then save the document, I would speak the word "please", then pause for a few seconds, then save the word " save." This would bring up the save dialog box and I could continue as if it was a command.

As far as the good points are concerned: If I'm working off printed list or when I know for sure what I want to say clearly, it can be a much faster way of inputting text than for me to type it. Some of that is due to my typing speed being somwhat slow. I am a faster typist than most hunt and peck folks, but I would never be able to keep a job as a secretary. It is also much better at spelling than I am.

If I don't know what I'm going to say in advance the program does not facilitate me speaking what I want to say and then recomposing it later. Some of this may be my fault. I find that I can type up an article by hand and maintain enough control so that in the end I have something coherent. When I just speak out what I want it does not always make sense. I spend way too much time revising it.

One of the disadvantages is that you need to spend a little bit of time training the voice recognition system. The system works best when you are running a very simple application such as WordPad or Notepad. It can even be used to some extent in Microsoft Word. However when using an application such as PowerPoint, the program spends so much time trying to figure out what possible commands you might be wanting to perform and not enough time actually realizing that you are trying to enter words.

Another problem is homonyms. These are words that sound the same but they have different meanings in English except for the possibility of some contextual clues. There is no way for the speech recognition software to know whether I want "2", "two", "to", or "too." During setup the program asks you if they can look at the contents of documents on your hard drive to help you determine what word you used most often and in what context. This does help it guess which homonym you want.

The final problem I wish to discuss is the editing features. Whenever I create a text document I will want to fix mistakes. There are some commands in the software that will help fix mistakes, but they are limited and clumsy. Some of them will even (in theory) help prevent mistakes by the voice recognition software in future. All that being said, I find that editing documents using the voice recognition software is very frustrating. I always resort to using the arrow keys and the mouse to navigate through the document.

Vista speech recognition is a helpful tool when transcribing printed text or when words have been spoken into a tape recorder. However for initial composition of documents or editing I do not find it useful. I do not believe that I could ever learn to use it as a complete 100% substitute for the keyboard and mouse. I would not consider it a replacement if I was disabled.

This article has been provided to APCUG by the author solely for publication by APCUG member groups. All other uses require the permission of the author (see e-mail address above).

Click here to return to top



==================================================================