You already have good answers on text-to-speech on your previous question:
convert from text to speech[
^]
So either you didn't look at the answers to your last question or you are actually asking about speech-to-text.
Google: "speech to text" or "speech recognition"
Nuance has been producing Dragon Naturally Speaking for years and they are still perfecting it.
There appear to be some free web services that will do it, but I suspect their accuracy may be poor.
Or perhaps you should investigate the new Kinect SDK which includes a sophisticated microphone array and speech recongition software.
Depends on what you want to do with the speech you are recongizing.