| Picture this...you're setting in front of your | | | | software uses the text from several popular |
| computer, staring at the desktop trying to decide | | | | novels in these reading/learning exercises. As you |
| if you want to surf the Net or do a little word | | | | read the text, the computer compares the way |
| processing. You finally decide to write to your | | | | you pronounce certain words with patterns |
| in-laws and see if they got the pictures you sent | | | | previously stored in its memory. As you read into |
| them. | | | | the microphone, the computer builds a voice file |
| Instead of reaching over and moving the mouse | | | | that describes in phonetics, exactly how "you" |
| pointer to the "Word" icon, you simply speak, | | | | speak. After a while, the computer will have |
| (Assuming you've already put on your microphone | | | | discovered patterns in your speech and will be |
| headset) "[computer load Microsoft Word.]" And | | | | able to extrapolate these patterns to new words |
| while you watch, the desktop disappears and | | | | that it has never heard you speak before. |
| Microsoft Word appears. So far so good. Next | | | | Working with "voice recognition" takes a little |
| you say, "[new file]" and immediately, a new white | | | | extra thought. Because the computer doesn't |
| screen appears in front of you. (This could be | | | | know when you've reached the end of a |
| habit forming) Now, it's time to write, but you | | | | sentence or a paragraph, you have to tell it. |
| never touch your keyboard. | | | | There are dozens of commands you can use |
| Again, you speak to your computer, "[new | | | | already programmed into the software. Words |
| paragraph] Hi Aunt Ruth [comma new paragraph] | | | | like, "new paragraph, space, new line, backspace, |
| I was just thinking about you and wondered what | | | | erase word, correct that, etc." These words tell |
| you thought of the kid's pictures I sent you last | | | | the computer that you want it to perform a |
| week [question mark new paragraph]." | | | | command of some kind. |
| I'm able to operate in this "No hands" mode | | | | In the actual document, these commands never |
| because of a voice recognition program called | | | | appear on the screen. I enclosed the commands |
| Naturallyspeaking. Not only can it transcribe your | | | | in brackets above so you could see how, when, |
| speech, as you speak, it can also be used instead | | | | and where I was using them. |
| of the keyboard and mouse to control your | | | | The first time I spoke into my microphone and |
| computer. This is only one of several such | | | | watched the words magically appear on the |
| programs on the market. I chose | | | | screen gave me an eerie feeling. It was like some |
| Naturallyspeaking because of its excellent reviews. | | | | unknown hand was typing my thoughts. |
| How well does it work? Most of the reviews say | | | | Naturallyspeaking comes with a microphone |
| that with sufficient training the computer will | | | | included in the package. All of these types of |
| achieve an accuracy of over 95%. I've not | | | | programs are very particular about the quality of |
| reached anywhere near that yet, but my | | | | the microphone being used. The voice engine |
| computer is still in its infancy. Fortunately, it's a | | | | requires high quality sound files so the microphone |
| quick learner! | | | | plays a crucial role in determining the final |
| Before I could actually start controlling my | | | | accuracy that can be achieved. It's also very |
| computer by voice, I had to train the program to | | | | important to speak into the microphone exactly |
| recognize me, and the pattern of my speech. | | | | the same way every time. Otherwise, you'll have |
| Each person will be different. Humans have little | | | | to reset the sound levels every time you want to |
| trouble in deciphering what another person is | | | | use voice control. |
| saying, even if they stumble over a few words | | | | Dragon Software is not the only producer of |
| or leave one out entirely. We group words into | | | | "voice recognition" software. IBM has its own |
| sentences and our brains tend to fill in any missing | | | | product as well as several others. If you think you |
| parts. But computers don't have this luxury. So it's | | | | might be interested in finding out more about |
| very important that they know our voices, the | | | | voice recognition, check out these links. |
| way we enunciate, and most of all, the patterns | | | | One thing I haven't yet mentioned is that you can |
| in which we string words together. | | | | turn the tables and have the computer read |
| Training is easy though somewhat boring. In order | | | | printed text back to you. I'm using it to proof my |
| to understand your voice, the computer has you | | | | new book by having the computer read aloud my |
| to read something it has already analyzed. Dragon | | | | text files. Welcome to the 21st century! |