since people have gotten a chance to be confused by the first, i will now post the original that i was reading from. ive actually been very impressed with it so far. the above example reads like engrish because i deliberately spoke too fast for it to recognize what i said accurately. when speaking a little slower id say that it has about a 95% accuracy rate on recognizing the words that i say.
original :

ive always been interested in voice recognition software, but have never really taken the time to use iti bought unreal tournament 2004 the other day, and it installs the microsoft speech sdk along with it. so I decided to play around with that and see what would happen.
i took a couple minutes to type this post, and then copied it to notepad. i then dictated it to the webform. this is the end product
i tried to follow the rules of dictation. here are those rules as put forth by microsoft

When speaking to other people, we are usually understood whether we speak slowly or quickly, loudly or softly. When speaking to a computer, however, it is better to speak in a more consistent tone. When you talk to your computer:

Speak in a consistent, level tone. Speaking too loudly or too softly makes it difficult for the computer to interpret what was said.
Use a consistent rate of speech; do not speak too slowly or too quickly.
Speak without pausing between words; a phrase is easier for the computer to interpret than a single word.
Because we do not usually need to think about how we speak, it might seem a little awkward at first when you talk to your

computer. With a little practice, however, speaking to your computer will become more natural.

i also spent some time training my profile before playing with the dictation. that was boring. i ended up falling asleep mid-sentence and taking a 1.5 hour nap.