Voice Recognition System for Home Security Keys with Mel-Frequency Cepstral Coefficient Method and Backpropagation Artificial Neural Network

In this paper, we present the design of a home door lock control system that is activated by automatic speaker recognition (biometrics). Access to a house or building with various conventional keys, PINs, or smartcards is not reliable enough to increase security because it cannot detect the real key owner. Furthermore, the introduction of the speaker as the key to the house door is applied to overcome this problem. Speaker recognition is the process of automatically recognizing someone who is speaking based on the sound characteristics of the input speech. This technique allows the use of the speaker's voice to verify identity and control access to their homes. It is proposed mainly since votes cannot be stolen, copied, forgotten, lost, or accurately guessed. The proposed system uses Mel-frequency Cepstral Coefficient for feature extraction and Artificial Neural Network Backpropagation for speech recognition. The results of this study for voice recognition show that the success rate in distinguishing homeowners reaches 97% with optimal conditions, namely in quiet environmental conditions (34 dB) with a sound collection distance of about 10 cm.