1999 Conference Proceedings

Go to previous article 
Go to next article 
Return to 1999 Conference Table of Contents


CHOOSING AND USING VOICE RECOGNITION

Claudia Atmore, MA, OTR
Santa Monica College

For the purpose of this presentation, I am going to address three different voice recognition systems: DragonDictate, Dragon Naturally Speaking Professional 3.0, and IBM ViaVoice 98 Executive Edition. At the conference presentation, a comparison chart of the different features will be discussed and shown, and a demonstration of each program will be performed.

There are a couple of factors to consider when choosing a voice recognition system:

If voice recognition is being looked at as a preventative measure (preventing a repetitive strain injury, cumulative trauma disorder, over-use syndrome), we further break down the category. Is the person trying to prevent getting a repetitive strain injury (RSI)? OR Is the person experiencing warning symptoms of a cumulative trauma disorder (numbness, tingling, aching in hands and/or arms)? For those who are trying to prevent an RSI or who are at the beginning stages of a repetitive strain injury, it is a good idea to utilize a combination of voice, keyboard, and mouse or mouse alternative.

If people use a combination of input methods, they are spreading around the use to different areas so that one set of joints, ligaments, and/or muscles does not get fatigued and over-used. There is a possibility of getting a cumulative trauma disorder in the vocal cords, so voice recognition should not be seen as the panacea of solutions. However, it is a viable solution to add to a repertoire of input methods. Either ViaVoice or NaturallySpeaking would be good options.

If voice recognition is going to be used by someone with a disability, the type of system chosen depends on the disability. Is it a physical or a learning disability? If it is a physical disability, does the person have use of his/her hands, or does he/she have no use of his/her hands? This presentation will focus on physical disability factors.

If people still have some use of their hands, meaning they are able to use keyboard commands and/or mouse or mouse alternatives (although somewhat limited), then there is not necessarily a need for a mouse grid option, so ViaVoice could make a good choice. It has the ability for one to dictate directly into a Microsoft application. NaturallySpeaking would work well too, even if you weren't going to need a mouse grid.

If people should not use keyboard commands, or a mouse or mouse alternative extensively due to their disabilities, then NaturallySpeaking with its Mouse Grid feature might be a good choice. It gives the individual an opportunity to use his/her voice to do most commands if it is used in its own word processor, and then the text can be copied into Microsoft Word, if necessary.

If people have no use of their hands, but they are able to use a non-hand-held mouse alternative, like a head pointing device or foot mouse, then either NaturallySpeaking or ViaVoice would make a good choice.

If the person is not able to use a mouse or mouse alternative at all, then Dragon Dictate or NaturallySpeaking used in conjunction with Dragon Dictate would be in order.

So far, we have looked at the type of disability, and determined choices for word processing. However, other types of applications may be needed. If the person needs to access the Internet, chat rooms, and instant messages, then at the presentation we will explore which would be the best solution. In addition, for spreadsheet and database programs, we will discuss at the presentation what might be a viable option.

No matter which voice recognition system is chosen, the strategies to consider when teaching are the same. I have done a task analysis on training voice recognition. That is to say, I have broken down the steps of training into a logical sequence. Obviously, the first thing to do is train the program to learn the user's voice by going through the program's voice training. Next, have the person dictate a pre-written paragraph and the trainer teaches how to correct words that were not understood by the program.

I usually have the person read a sentence, verify its accuracy, and make any necessary corrections, before dictating the next sentence. The reason why I have them read a paragraph, rather than compose their own work, is that the person needs to first concentrate on learning the commands of the program. Reading text is a different experience, and much easier for people if they do not have the pressure of putting their own thoughts into words.

The next step is to teach navigation commands, and selecting text. Teach them how to use their voice to move the cursor through the text in the most efficient manner: by paragraph, by line, to beginning of document, to end of document, etc. Have them select text using their voice: a word, a line, a paragraph, an entire document, etc. Teach basic file management - open, save, new, close, exit - using their voice. You can do this by having the students save the document they just created, close that file, and create a new file. Have the student read a second paragraph to hone in on the basic skills you have just taught him/her. Next, have the person dictate an original paragraph of his/her own thoughts. Dictating one's own ideas is a much different thought process than reading pre-written material. Emphasize that it is important to think of what you want to say before saying it. If people stumble over their words because they are trying to compose while they dictate, then recognition will not be as accurate. After the person has gained comfort with the basics of creating a file, editing it, etc., then go on to text formatting features. Show the student how to change the alignment of text - center, right align, left align - by using his/her voice. Have some text for the individual to read so that he/she is focusing on learning the commands. Next, teach the student how to bold, underline,


Go to previous article 
Go to next article 
Return to 1999 Conference Table of Contents 
Return to Table of Proceedings


Reprinted with author(s) permission. Author(s) retain copyright.