Speech Technology Magazine

 

Voice Biometrics Grows to Meet Global Challenges

Today's solutions are more accurate and easier to use and implement.
By Dan Miller - Posted Oct 1, 2007
Page1 of 1
Bookmark and Share

Global plans, global exposure. Those two forces are accelerating deployment of biometric solutions, including voice biometrics, for security worldwide. That’s why "Gaining Global Acceptance" is the theme of this fall’s Voice Biometrics Conference in London Nov. 28-29.

One example of the global plan is a voice biometric-based system from VoicePay that enables secure card-based payments from any phone. This greatly expands prospects for friction-free use of spoken words to authorize phone-based payments. Other examples include customer-facing deployments in banks (such as ABN AMRO in the Netherlands and Israel’s Bank Leumi) and large telephone companies (such as Bell Canada).

Global exposure is a corollary to expanded deployments. Security breaches across all manner of commerce and payment systems have greatly stepped up the average person’s exposure to identity theft. Thanks to careless handling of customer records, coupled with an increase in organized efforts among fraudsters to obtain information required to carry out online transactions, one out of every three Americans has become the potential victim of identity theft, according to research by Intelligent Biometrics Solutions.

A higher level of exposure to identity theft alone has driven security experts to take a closer look at biometric solutions. Simply, fraudsters can dumpster dive behind hospitals for personal information. They can intercept communications that are supposed to be encrypted or protected. Yet they are incapable of stealing the physical attributes of an individual, such as her fingerprint, iris pattern, facial characteristics, or voice. When it comes to commerce carried out over the telephone, the use of voice-based biometric characteristics to
identify or authenticate customers only makes sense.

The Help Desk Phenomenon
Before long, password reset became a priority IT function due to security-mandated, stringent protocols that call for frequently changing passwords. To preserve convenience and control costs, voice biometrics stepped in to help the help desk streamline the process of resetting an individual’s password with high levels of confidence. This proved to be a very popular point solution, and many technology providers built up hope that it portended great things for enterprisewide deployments. That expectation proved inaccurate because the underlying business rules and policies surrounding enterprisewide security are much more complex than those governing password reset.

The ensuing years witnessed rapid growth of the Internet, e-commerce, and online banking. Closed networks to support branch banking and ATM functions have given way to wired and wireless links between PCs, laptops, and mobile phones. Each new device and network configuration creates new challenges for security experts and new opportunities for potential fraudsters. Yet the well-worn decision variables of cost control and user convenience continue to dominate vendor selection in building new security schema.

Responding to the expressed demands of key customers in finance, government, and healthcare, IBM/VOICE.TRUST, VoiceVault, Nuance, PerSay, Diaphonics, and others have delivered products and services designed to shorten the time from demo to implementation while overcoming several long-standing impediments to full-scale deployments. They’ve placed equal amounts of attention on engine accuracy, ease of installation and implementation, quality of user experience, and adherence to security policies, practices, and existing infrastructures.

IBM’s introduction of Speaker Identity Verification (SIV) as feature packs for its WebSphere middleware exemplifies how voice biometric resources will be seamlessly integrated into enterprise security workflows, now and in the future. By embedding a voice biometric engine and rating logic in middleware, Big Blue creates opportunities for technology partners, like VOICE.TRUST, and integrators, including IBM Global Services, to closely mate existing business policies with customizable, off-the-shelf software solutions.

Today’s speaker authentication engines are more accurate than ever before, largely because vendors have benefited from experience in the field to tweak the underlying algorithms, and have taken advantage of servers with high-speed CPUs and low-cost memory. Just as important has been a certain level of standardization surrounding workflows for both online and telephone-based authentication. Speed, accuracy, and standardization make it more attractive for integrators and security specialists to include  a voice biometric factor in their solution sets. 


Dan Miller is the founder and senior analyst at Opus Research. He published Telemedia News & Views, a monthly newsletter covering developments in voice processing and intelligent network services. He can be reached at dmiller@opusresearch.net.

Page1 of 1