DEVELOPMENT OF A DOOR LOCK SECURITY SYSTEM BASED ON AUTOMATIC SPEECH RECOGNITION

Résumé

The security of homes and properties are of utmost importance and various methods have been employed by researchers to achieve this aim. This work developed a door lock security system based on automatic speech recognition. The developed system was programmed to recognize certain users and give a particular user access to a door. In training the system, MFCC feature extraction technique was used to extract appropriate features from five different individual’s speech signal and Vector Quantization using LBG algorithm (VQLBG) was done to recognize a speaker. A prototype door was designed and implemented to test the developed system.
Euclidean distance was used in calculating the parametric representation of individual speech signal to be recognized and an accuracy of 75% was obtained by calculating the Word Error Rate (WER). Results show that the developed system is more reliable in securing door lock in homes than traditional method.