Speaker Indentification using Labview

Abstract

Now days, Biometrics is being used extensively for the purpose of security. Biometrics deals with identifying individuals with their physiological such as fingerprint DNA, ECG etc or behavioral traits i.e. rhythm, gait, voice etc. Voice is a most natural way of communication and non-intrusive as a biometric, Voice biometric has characteristic of acceptability, cost, easy to implement as no special equipment is required. Also Voice based biometric system can be easily combined with other biometric systems to enhance the reliability and security of the system. In the present work a speaker identification system has been developed. The developed system uses the LabVIEW (Laboratory Virtual Instrument Engineering Workbench) 8.5 platform. Speaker Identification involves features extraction, preprocessing, pattern matching, decision-making. Silence removing of voice signal is key factor to improve the identification. In feature extraction stage, Mel frequency cepstrum coefficients (MFCC) have been calculated which provides a better measure of Speaker Identification than the other features. Speaker identification can be done by various methods but in this thesis vector quantization based recognition system using LabVIEW has been developed and tested. The developed system is user friendly and provides the results in real time. A database of 20 person having 5 samples per person including male and female has been created. The experiments conducted on the above database suggest that an accuracy of 90% has been achieved with the developed system.