By Summerfield, Clive
Journal of Banking and Financial Services , Vol. 117, No. 1
Speaker Verification (SV) is a biometric technology providing a unique voice identifier that can be captured over the telephone, verified reliably, and permanently attached to an individual consumer's identity credentials.
SV promises to play a powerful role in financial services and, indeed, some forward-looking financial institutions have already successfully adopted the technology.
The authentication problem
The Australian Institute of Criminology (AIC) estimates that identity fraud costs the consumer between $2 billion and $6 billion a year. By its nature, identity fraud is difficult to detect and nobody really knows just how much it costs.
Authenticating identity has always been problematic. Currently, methods rely on examining documents such as driving licences, passports and birth certificates. In most cases, these documents are simply not designed for the job. Many are readily available and with the advent of low cost reproduction technologies creating high quality forgeries is easier than ever.
Further, with the trend towards call centres and internet service, establishing identity is even more problematic, relying on customers to remember and keep secret passwords and PINs.
Passwords and PINs are notoriously weak forms of security. When a password or PIN is lost or forgotten, the customer needs to re-establish his/her identity, usually by calling the call centre and answering a sequence of personal questions.
Not only is this costly and time consuming, this process does not necessarily establish the authenticity of the caller. In addition, the call centre agent is also privy to identity information, providing yet another opportunity for fraudulent activity.
While call centres and the internet are a convenient and effective way of providing customer self-service, they are also the `weakest link' as far as security is concerned.
Speaker verification is a technology that enables callers' identity to be authenticated using the unique characteristics of their voice. Typically, a customer records his or her unique spoken password (which can be as simple as their name) when initially establishing identity. This is then used to confirm his or her identity on subsequent calls. The technology uses both the unique password word the customer has recorded and the unique way the customer says the password.
The deployment of SV technology benefits both the end user and the organisation. It is convenient for the customer, eliminating the need to remember PINs and passwords; it does not require the intervention of a call centre agent, and thus the process is highly secure.
What's more, as the process is performed automatically, SV offers significant cost benefits in terms of call centre efficiency.
While the deployment of SV has lagged behind that of speech recognition, the advantages of SV are starting to be recognised. In many ways, the technology has benefited from much of the work done in the area of speech recognition, where the technology is now starting to find wide ranging acceptance in the banking and finance sectors.
In many ways, the technologies go together and can be combined to provide both convenient and secure voice interfaces for self-service systems.
`For banking services where the user population is so widely dispersed, only speaker verification using the telephone system will work,' head of ANZ Bank fraud and risk group Lawrence Cox says.
In an example presented to the recent Biometrics Institute conference, Cox illustrates the benefits of SV by quoting the Home Shopping Network experience in the US.
Every Monday, Home Shopping would receive about 160,000 calls requesting PIN resets. Re-establishing identity involved caller centre staff asking the caller about 20 personal questions. This was not only insecure, but was very expensive in terms of call centre agent costs. …