There have been a few threads about it, an mroogle search will find them.
That said, speech recognition requires you to train the computer to recognize your voice. It takes time. Also you will still have to proof whatever you dictate and expect errors.