Really? Either you are terrifically bright and experienced for a college student, or you have bitten off more than you can chew...
Start with Google:
make a speech recognition engine[
^] and then start reading. Lots. This is not a trivial project regardless of whether you implement it as Web or windows based.
Think about it: Google speech works (just about). Siri works (just about). How many man-years do you think the two companies have put into researching this? You - on your own - are unlikely to finish this before the end of term. Or even the school year...