I asked one of my coworkers about this, and he said that voice recognition requires training for the computer to recognize the words an individual speaks, so while it seems like 'your rain yum' must have been done by a computer it was almost certainly done by a human... my co-worker sent me some links about how captioning is done...

http://www.robson.org/capfaq/look.html#Misstrokes

http://www.odc.state.or.us/tadoc/techcp8.htm

http://www.odc.state.or.us/tadoc/techcp22.htm

http://www.odc.state.or.us/dhhaptac.htm#Captioning

actually quite interesting to read....

from seeing the captions it is clear that the deaf are being screwed... on sitcoms and other shows it is clear that the captions are written fron a 'beta-version' script so, as tom pointed out, the text delivers a different joke than the one spoken. also I have noticed that words that are hard to spell or enunciate are skipped altogether...

The deaf deserve better

I do believe that impairment in one area can lead to enhancement in other areas, I was dyslexic (at least that is what my school labled me) as a kid... I still read at an early age, and it hasnt been a handicap at all- quite the contrary, very often when I see the 'word jumbles' game in the newspaper I read the scrambled words immediately.... this also comes in handy in 'Scrabble'...:)