Using FIR Coefficients to Form a Voiceprint

Abstract: Voice recognition is a method of security that can be accessed almost by everyone. It provides those with visual impairments the same level of security as those without. Similarly, technologies that require physical interaction may prove difficult barrier that other methods of technology may carry. Voice recognition systems are typically uti- lised in contact centres where there is no other method for physical authentication. However, it can be used across a wide range of industries such as: Hospitality, Retail and Logistics, and Banking and Finance. In this research paper we will introduce a method of creating human voiceprints based on using finite impulse response filter coefficients, this linear prediction coding method will be tested, implemented using various wave signal files to prove the act of how to identify the person and how to identify the spoken word or phrase, allowing the recognition system to use a voiceprint as a password.

Keywords: Digital Signal, Voice Signal, Voiceprint, FIR, LPC, Filter Order Recognition System, Correlation, MSE