Speaker Verification System using LabVIEW

The need for simple, ubiquitous and cheap security procedures to verify the authenticity of the user are increasingly gaining importance to enhance the reliability and integrity of the transactions conducted electronically. The conventional means of user authenticity such as password, personal identification number (PIN) are no longer reliable enough to satisfy the security requirements because of their inability to differentiate between an authorized person and an imposter who fraudulently acquires the access privilege of the authorized person. Speaker verification is a biometric which provides positive verification of identity from an individuals voice characteristics. In this paper a text dependent speaker verification system using Mel-scale filter bank for feature extraction and minimum distance classifier for hypothesis testing is developed using LabVIEW. The availability of datalog files in LabVIEW makes it one of the most promising candidate for its usage as a database . Datalog files can access and manipulate data and complex data structures quickly and easily. It makes writing and reading much faster. The system developed in LabVIEW detects the user in almost real time with an accuracy of nearly 94 and is highly repeatable.