User Identification System Using Biometrics Speaker Recognition by MFCC and DTW Along with Signal Processing Package

Abstract

User identification proof framework is essential for securing data from illicit access. To build a robust user identification system using voice, a new system is proposed to identify users using Mel-Scale Frequency Cepstral Coefficients (MFCC) and Dynamic Time Warping (DTW) along with a package of digital signal processing. Human voice is a sign of boundless data. Precise voice recognition requires computerized processing. Proposed method extracts unique features from a voice signal by MFCC and DTW to compare the components between two signals with the aid of some efficient signal processing such as filtering, signal alignment, removing unvoiced part, amplitude normalization and zero-part removal. All these steps work perfectly for accurate voice signal recognition. Based on the similarity between voice signals, it distinguishes different users and grant access to the secured area for multiple users which could be substantial for internal security for any classified organization or nation.