Talking to your computer

I spent the last few days helping one of my friends set up voice recognition software. Theoretically, this is the "holy grail" of computing if you believe in Star Trek. We would all like to speak to our computer and have it do all the keyboarding for us automatically but since its introduction at the Seattle Worlds Fair in 1962, voice recognition has never quite lived up to its promise.

Just to give you an idea of what voice recognition can or cannot do, I an going to dictate the rest of this Blog post. So that you will get an idea of how it works (or doesn't) I will NOT MAKE ANY CHANGES or edits to the text of the post:

The dictation starts right here. I am using DragonDictate on a Macintosh computer. I have a Plantronics  digital headset with a USB connection to that computer. Over the years, I have used Dragon naturally speaking and Dragon dictate and various other programs because I am always looking for a way to input text or rapidly than typing from the keyboard. Unfortunately, whenever I have tried using the programs, I always find that certain words are not recognize properly. I realize that this occurs primarily because as I continue to speak, I have a tendency to slur my words together.

As I dictate this portion of the post, I am refraining from making any error corrections purposely to show what you have to do to go back through the text and correct the mistakes. The main thing that I have found, however, is that I am used to rewriting as I go along. This is quite hard to do using a voice-recognition program.

Another challenge, comes from using voice recognition in a database program. Unless the database program adapts well to the limitations of voice recognition it is no easier to use voice recognition than it is to use the keyboard. This is especially true, if you have to continue to use the mouse every time you move to a new window or location for entering data into the program. I have noticed, that the voice recognition programs do a credibly good job in recognizing names. You can probably guess, that the VR programs do not do a very good job with name variations. Such as spelling Stevens within the or APH.

As I continue to dictate if you read carefully, you will see that some of the sentences are garbled. Technically, you're supposed to go back and make the corrections using voice recognition which teaches the program how to understand your oral dictation. It is just the nature of the products,  even when they have a provision for learning your speech patterns, to miss some types of words. In addition, I find it quite difficult to (I might note, that the program crashed at this point in the dictation). In addition, I find it quite difficult to proofread the text, especially when there is a missing word. It is also difficult to detect the wrong word inserted in the text.

All in all, using voice recognition, like speeding along on the freeway, does not always get you to your destination. Because you travel so fast on a freeway you have a tendency to drive out of your way to travel along the freeway, when it would've taken less time on the regular city streets. Because of the novelty, the ease-of-use, and the apparent ability that voice recognition has recognized most of your text, you have an illusion of productivity but I truly question whether you lose any advantage in the need to do careful proofreading.

Not that my usual level of writing is so exemplary, but likely you can detect some difference in the way that the dictated text reads from my normal way of writing. As I have said many times before, voice recognition offers great benefits to anyone who has difficulty using a mouse or keyboard. I suggest trying for strict edition for some time on documents that you are not inordinately worried about to learn to accommodate the shortcomings of the programs.

I guess that's all I have to say today.