Dragon speech recognition for PC or iListen for Mac are able to work with recorded audio files.
The problem with this is that both programs only work well if they are trained for a specific speaker - the speech idiosyncrasies of the individual can be learned and enormously help in recognition, especially since the program gets better and better over time.
Also, ambient noise will probably make large portions of the audio recording useless to the software, as it cannot "zoom in" on the speech and filter out the voice from the surrounding noise the way the human ear/brain can (akin to the way your eye can discard extraneous information and focus on what is of interest to you at the moment).
If you are looking for a solution to record transcribe speech in varying locations, with varying levels of background noise and different or multiple speakers, I'm afraid that a personal secretary or a transcription service are, to my knowledge, your only useful options at the moment.