Speech Recognition is an interesting subject that comprises several stages of analysis, modelling, and processing. Basically, an Automatic Speech Recognition System (ASR) is capable of transcribing the words that a certain speaker says into text. The method essentially mimic the way how humans transcribe the codes from speech.
You can download a paper that describes an ASR-related system that I developed on 2011 for the Pattern Recognition class. This was based on Hidden Markov Models (HMM) and is capable of recognising some preset utterances. The motivation was to use the recogniser for automation of musical instruments in live performances.