Speaker Identification And Verification Over Short Distance Telephone Lines Using Artificial Neural Networks

SPEAKER IDENTIFICATION AND VERIFICATION OVER SHORT

DISTANCE TELEPHONE LINES USING ARTIFICIAL NEURAL

NETWORKS

Ganesh K Venayagamoorthy, Narend Sunderpersadh, and Theophilus N Andrew

gkumar@ieee.org sundern@telkom.co.za theo@wpo.mlsultan.ac.za

Electronic Engineering Department,

M L Sultan Technikon,

P O Box 1334, Durban, South Africa.

ABSTRACT

Crime and corruption have become rampant today

in our society and countless money is lost each year

due to white collar crime, fraud, and embezzlement.

This paper presents a technique of an ongoing work

to combat white-collar crime in telephone

transactions by identifying and verifying speakers

using Artificial Neural Networks (ANNs). Results

are presented to show the potential of this technique.

1. INTRODUCTION

Several countries today are facing rampant crime and

corruption. Countless money is lost each year due to

white collar crime, fraud, and embezzlement. In today's

complex economic times, businesses and individuals

are both falling victims to these devastating crimes.

Employees embezzle funds or steal goods from their

employers, then disappear or hide behind legal issues.

Individuals can easily become helpless victims of

identity theft, stock schemes and other scams that rob

them of their money

White collar crime occurs in the gray area where the

criminal law ends and civil law begins. Victims of

white collar crimes are faced with navigating a daunting

legal maze in order to effect some sort of resolution or

recovery. Law enforcement is often too focused on

combating "street crime" or does not have the expertise

to investigate and prosecute sophisticated fraudulent

acts. Even if criminal prosecution is pursued, a criminal

conviction does not mean that the victims of fraud are

able to recover their losses. They have to rely on th

criminal courts awarding restitution after the conviction

and by then the perpetrator has disposed of or hidde

most of the assets available for recovery. From the civil

law perspective, resolution and recovery can just be a

difficult as pursuing criminal prosecution. Perpetrators

of white collar crime are often difficult to locate and

served with civil process. Once the perpetrators have

been located and served, proof must be provided that

the fraudulent act occurred and recovery/damages are

needed. This usually takes a lengthy legal fight, which

often can cost the victim more money than the fraud

itself. If a judgement is awarded, then the task of

collecting is made difficult by the span of time passed

and the perpetrator's efforts to hide the assets. Often

after a long legal battle, the victims are left with a

worthless judgement and no recovery.

One solution to avoid white collar crimes and shorten

the lengthy time in locating and serving perpetrators

with a judgement is by the use of biometrics techniques

for identifying and verifying individuals. Biometrics are

methods for recognizing a user based on his/her unique

physiological and/or behavioural characteristics. These

characteristics include fingerprints, speech, face, retina,

iris, hand-written signature, hand geometry, wrist veins,

etc. Biometric systems are being commercially

developed for a number of financial and securit

applications.

Many people today have access to their company's

information systems by logging in from home. Also,

internet services and telephone banking are widely used

by the corporate and private sectors. Therefore to

protect one's resources or information with a simple

password is not reliable and secure in the world of

today. The conventional methods of using keys, access

passwords and access cards are being easily overcome

by people with criminal intention.

Voice signals as a unique behavioral characteristics is

proposed in this paper for speaker identification and

verification over short distance telephone lines using

artificial neural networks. This will address the white

collar crimes over the telephone lines. Speaker

identification [1] and verification [2] over telephone

lines