Voice based access control system

Abstract

This report follows the implementation of Voice-based Access Control System from inception phase to its completion. The system recognizes the person basing on his/her voice in real-time for easy access. It also provides the information about the techniques used during the development of Voice-based Access Control System such as linear predictive coding, Mel Frequency Cepstral Coefficients, High pass filter for noise reduction or signal smoothing and Convolution Neural Network CNN for person recognition. With the increased need for data security the use of username and password has been broken down by shoulder surfing, pre-texting and other social engineering tricks. This has therefore rendered the use of what the agent (person/system) knows as security measure to be less likely used in some systems top-secret security archive, missile launching control systems, medical records access systems among others. This is because any other agent who gains access to what the agent knows (username, PIN, password), they can gain access to the sensitive data and/or system, the problems associated with the inability to know a person’s identity, the limitations with other biometrics for person identification and since a person’s voice cannot be stolen, lost, forgotten, guessed or impersonated with accuracy. The project produces Voice-based Access Control System to recognize a person based on his/her voice and be able to tell whose voice it was for access control, The system can also be attached in a door access system of a smart home, office, train, bus and aero planes to control access to only authorized personal, voice dialing and voice mail, tele-banking, telephone shopping, database access related services, information services, security control for confidential information areas and remote access to computers.