Proper

Abstract

This paper proposes a system architecture for convenient and accurate voice authentication by incorporating mobile phone voice input and a One Time Password (OTP) to the voice authentication algorithm. Voice authentication has the advantage of enabling remote applications using existing infrastructure, but one of the hurdles in using voice authentication method has been its low accuracy. The users may prefer using whisper when providing their personal information for voice authentication over the mobile system in public circumstances. The experimental results show that the speech recognition performance over whispered speech for voice authentication system was improved by employing high-pass filter and acoustic model adaptation algorithm.