Abstract

Access to an area or environment can be controlled by conventional and electronic keys, identity cards, personal identification numbers (PINs) pads and smartcards. Due to certain limitations of existing door access schemes deployed for security in buildings, this paper presents speaker recognition for building security as a better means of admission into important places. This is proposed due mainly to the fact that speech cannot be stolen, copied, forgotten, lost or guessed with accuracy. This paper, therefore presents design of an affordable voice activated door control system for building security. The proposed system uses the Mel Frequency Cepstrum and the Gaussian Mixture Model for feature extraction and template pattern matching respectively. The analysis of the result which is based on the false acceptance and rejection rates indicate a system accuracy of more than 80%.