Intelligent Voice-based Authentication System

Abstract

This paper presents an intelligible voice based authentication system for IoT devices. As voice changes over time due to different characteristics and other environmental effects, limited features based training approaches may not provide the desire level of performance. The accuracy of the system is also dependent on several other factors like- the control over recording equipment, recording conditions, cooperation from speaker, etc. Therefore, this work analyzes the voice variation of the speech over varied conditions to address the long term speaker variability to enhance the performance of the voice based authentication system. The integration of IoT technology with voice based communication technology may help users to make their systems work based on voice commands instead of typed commands. A number of tests were also conducted to show the effectiveness of the presented model under varied conditions.