Small Scale Voice Recognition
Abstract
This report describes a small-scale, isolated word voice recognition module. I explain my approach to the problem of voice recognition and provide performance figures and sample runs to give a picture of how this module operates. In general the module is successful in matching isolated words. However, my project goal of speaker independence was not met.
Voice recognition is not a new idea. Successful voice recognition was achieved in the 1950's using analog speech storage and comparison methods. However, most research focuses on recognition of large vocabularies and continuous speech. For my recognition module, I use a simple matching algorithm that matches waveform data processed by a Fast Fourier Transform to previously stored patterns to recognize human speech.
Description
Program year: 1994/1995Digitized from print original stored in HDR
Subject
voice recognitionisolated words
performance figures
Fast Fourier Transform
recognition module
Citation
Armstrong, Bryan (1995). Small Scale Voice Recognition. University Undergraduate Research Fellow. Available electronically from https : / /hdl .handle .net /1969 .1 /CAPSTONE -CharlesD _1979.