One of the most worrisome aspects of using speech-recognition software, especially for those of us whose hand injuries make us dependent on the software, is the possibility of injuring our voices through overuse. While this is a real possibility which should not be ignored, it should not be a cause for panic. Just as with repetitive strain injuries of the hands and arms, there are number of things you can do to minimize your chances of damaging your vocal cords -- prevention is the key.
(Important Note: I am not a doctor or speech therapist. I am simply a user of speech recognition software who has had some problems with hoarseness and loss of voice, and has gotten relief by seeing a speech therapist. (I saw Kate DeVore at the Beth Israel Deaconess Medical Center in Boston. She has moved to Chicago, but the Voice and Speech Clinic is still there). The advice on this page should not be taken as a substitute for medical advice. If you are already experiencing hoarseness or strain from using speech recognition software, you should see a doctor.)
Some of the most important advice for preventing voice problems is the same is that for preventing repetitive strain injury in the hands and arms:
The most common advice is to drink plenty of fluids. Avoid caffeine, which causes dehydration. Cold water tends to make muscles tighten up, so some people recommend water at room temperature, or (non-caffeinated) tea, which musicians or people with colds often used to soothe their throats. I personally like to keep a bicycle water bottle at my desk, although alternatively having to go down the hall to get a drink would force me to take periodic breaks. However, water alone is not sufficient to prevent overuse injuries.
Now that large-vocabulary continuous speech recognition is available for PCs, I would strongly recommend that you use it for any tasks for which it is suitable. (I still use DragonDictate for command and control and programming, but dictate English text almost exclusively with NaturallySpeaking.)
Just as typing technique is important to preventing RSI, vocal technique is important to using your voice safely for long periods of time. You should not whisper to your computer, since this actually requires more effort than speaking normally. I heard this advice before I started using speech recognition software. However, I did not know that the same could be true of speaking softly. I thought that speaking softly would naturally be less effort than speaking loudly. As a result, I ended up using the minimal amount of breath I could, a bad habit which I already had in my everyday speech. If you breathe properly and use your mouth rather than your vocal cords to shape your sound, you should be able to project a reasonable volume without any more effort than speaking quietly.
If you have experienced symptoms of vocal strain (such as hoarseness or loss of voice) during or after extensive use of speech recognition software, I strongly recommend that you see a doctor. Even if your doctor finds no visible signs of vocal cord irritation, or if you have not had symptoms but are concerned, it is a good idea to see a speech therapist. Some of the techniques taught by speech therapists are also taught by voice coaches for actors, singers, and public speakers. Obviously, if you have symptoms, it is better to see someone with medical training, but if your insurance and budget do not allow that, a good voice coach would probably be better than nothing. Some of the tips I learned from speech therapy were:
Most of these suggestions should not interfere with the accuracy of the speech recognition software, and may even improve it through better projection and enunciation. The one possible exception is that speech recognition software often picks up the noise of breathing. This can be solved by correctly positioning the microphone off to the side, at the corner of your mouth, rather than directly in front, as well as by remembering to close your mouth when you finish speaking.
This document last modified on 1/3/2002
Questions? contact web@ataword.com