Top 10 Epic Phail PC Technologies: Voice Recognition
To apply the vernacular of the modern online PC enthusiast: Voice Recognition is an Epic Phail. Why?
It was 1966 and for the first time we saw people speaking to computer screens and the computers replying in clear, perfect English. This was a time when in order to communicate with a computer (which usually took up an entire floor of an office building) you needed to punch silly little holes in silly little cards and feed them through in stacks like a Las Vegas shuffling machine at the blackjack table. Therefore the paradigm of accurate computer voice recognition was an idea whose time had come.
Unfortunately that time was the 23rd century as the only voice recognition that actually worked was on Federation Starships, and usually addressed by green-blooded, pointy-eared natives of the planet Vulcan. There was no effective, accurate, everyday usable voice recognition in 1966. And guess what? It's now 43 years later and there still isn't.
Voice recognition requires an inordinate amount of processing power and it also helps to have more RAM than a metro Chrysler dealership. When the Nehalem line of processors was first announced by Intel, it was widely touted to be the first platform to provide the performance necessary to bring effective, accurate, everyday blah blah blah to the desktop. Well, guess what again? I'm typing this on a Nehalem CPU, a Core i7 920 which even by modern standards has an outrageously large amount of RAM, a full 12 GB... and voice recognition on this PC sucks.
Hey, it sucks just a tiny bit less than on my previous main platform, a Core 2 Quad with 4 GB of RAM, but it still applies more suction than the entire Dyson vacuum cleaner factory.
I type 90 words per minute. Therefore, in order for voice recognition technology to be able to give my long-suffering fingers their well deserved rest after many decades of pounding poor innocent keyboards into submission and early retirement, I would have to be able to get accurate text on the screen at pretty well the same speed that I speak. After applying dang near every major voice recognition software on the market onto my Core i7 920 the best equivalent result I can get is eight words per minute.
Before you start accusing me of confusing the poor computer by having a thick Italian spaghetti bender accent and talkinga toa maia computera lika dissa, I have flawless "network news anchor" diction: even with that, whenever I try to use voice recognition software I still get mostly gobbledigook which I have to extensively edit and correct by using that very resource that I was trying to save... my worn out fingers!
So I have to go through all this to get 8 wpm? Voice recognition can pucker up and approach my malodorous rectum.
My hatred of voice recognition software has gotten so profound that I can't even stand the ubiquitous telephone systems any longer. Whenever I hear:
If you are calling about your telephone system, please say telephone.
If you are calling about your internet system, please say internet.
If you are calling about your television system, please say television.
I usually scream a stream of expletives into the handset and hang up. And that's what I've done to voice recognition on my state of the art system. Hung up and uninstalled it all as the expensive useless time wasting technology it is, and is likely going to continue to be for the foreseeable future.