Voice-Assistant Liveness Analysis

Abstract

Deep learning is becoming a mainstream technology for speech recognition as well as face recognition at an industrial scale. The ability of devices to respond to spoken commands is basically speech recognition. The main objective of building a voice assistant is using semantic data sources available on the web providing knowledge to the users from the knowledge database. For the security purpose of the voice-triggered device, liveness analysis is required. The objective of this paper is to prevent spoofing attacks on voice assistant devices by introducing a liveness analysis of genuine faces. Different classification algorithms are used for face recognition purposes. Finally, the performance analysis of different classification models is made.