Speech recognition using frequency based analysis

Science fiction often presents technologies outside the scope of current reality. Mission Impossible III includes a scene where one man is able to steal another man’s voice purely on a recording containing all of the phonemes in the language. Although this scene is fictional, elements of it can be recreated on a computer. This thesis explores the scope of implementing such a system. It attempts to recreate the functionality of the ears, brain, and voice with respect to communicating Canadian English. Phoneme isolation and identification will be examined. Feature selection will be considered to improve the recognition process. An assessment of the effectiveness of the project is included, along with possible applications.