My ER1 can do that easily; it interfaces with Microsoft Speech Recognition and displays a string with the word(s) and/or the sentence it recognized. Unfortunately, the software is rather bulky and overkill for this one application (It's also no longer made/supported).
I know many programming languages have extensions that allow you to access the Windows API and COM (These allows you to directly control Windows, I think). PyWin32 for Python is one example. Microsoft Visual Basic (or any of the .NET languages) can also control certain aspects of Windows as well. If the speech recognition on Windows is actually part of the OS, you could access it by this method. This will require programming; exactly how much, I don't know.
Try researching Vista's Speech Recognition, perhaps it is possible to upgrade to Vista's on an XP.