Identity authentication and verification are increasingly important to call centers in a number of industries. Financial services, government applications, health care - all have either regulations or compliance standards to adhere to.

As the technology grows more advanced, call centers are facing an ever-increasing number of options for ensuring security with their data and their caller base. On choice, voice biometrics, potentially offers multifactor authentication with an added convenience -- there are no scanners to buy, PIN numbers to remember or cards to lose -- you always carry your voice with you.

We recently had a transoceanic conversation with Israel Ronn, head of CellMax Systems, a company that provides voice biometric identification and verification technology for call center industry. We asked him to explain a little about the technology behind it, and the market for such tools.

Q: Where in the call center technical architecture does your tool fit?

There are two possibilities: As a part of various IVR sessions, whether as access control to the call center, or connected to the call center voice system, performing Conversational Biometrics (i.e. free speech based authentication) during various human or automated sessions.

Q: What's the value proposition for using it in transaction processing? I can understand the application in a physical access or presence scenario, but isn't there a huge cost differential between this and a similarly secure low-tech triple access system?

Voice biometrics provides a higher level of security relative to four-digit passwords, and return on investment is achieved within a short period of time.

Voice biometrics provides automated identification and verification, thus dramatically saving agents' time. It takes 20-40 seconds to answer challenge questions when done by an agent, while a voice biometrics system recognizes voiceprints automatically (with no human involvement) within 3-7 seconds. By getting rid of this process, a large-sized call center can save tens of thousands of dollars every month, because process automation is increased -- and that keeps costs low.

Using voice biometrics also improves service-oriented parameters such as "queue wait time" and "abandon rate". Voice biometrics can also be used to eliminate "password reset", which is a time-consuming process. So, information input is automatic, data control levels rise, quality of service increases, wait time is reduced, and all this ultimately results in rapid ROI.

There is cost differential between the use of voice biometrics and the methods currently used is not huge at all. In fact, it's quite the opposite -- voice biometrics is first and foremost a cost-saver, saving on agent time and 1-800 waiting time.

Also, the level of security provided by voice biometrics is higher. At present, most contact centers require a four-digit password and, if they want more security, ask "challenge questions." But PIN numbers can be forgotten or stolen, and many people -- aside from the individual in question -- can know the answer to a challenge question. My brother, for instance, can easily answer questions about my birth date, our father's birthplace, etc. But there's only one person in the world with my voiceprint.

At CellMax Systems, the voice biometrics solution we've developed is based on 15 physical parameters that it takes to create a personal voiceprint and uses algorithm with an unprecedented rate of accuracy -- I think we've stated that one-to-one verification can go beyond 99.8% and one-to-many identification, up to 98%.

Q: Your literature describes the standardizing of the Voice Data File Format, including work with the ISO. What's the timeframe for that project and how far along is it currently? Are you working with other vendors in that area?

We are an active participant in ISO, and also have relations and connections with other bodies, such as the BioAPI Consortium, the VoiceXML Forum and others. The idea is to create an international standard that will enable universal installation, communication and interface between all voice biometric formats, and the person charged with that task at CellMax Systems is Ziv Barzilay, our founder and CTO. Ziv is also the developer and patent-filer of our voice biometrics solutions, and represents the Standards Institution of Israel to ISO.

In January 2007, ISO approved a new work group for standard for Voice Data File Format and Ziv was chosen editor of the "Speech Data Interchange Format for Speaker Recognition" project. We believe it will take about a year and a half until it's completed.

Q: How many centers are using this now, and what's the potential market?

Our product has traditionally been used in the security arena and we have installations in the physical access control and unified messaging spaces. In the call center space, we have about 10 installations including a pilot at a major Israeli bank as part of a bid for tender. We also have agreements in place with a number of call and contact center integrators (the full list is posted on our website) and we are in the process of training four integrators in the US, Canada, Panama and Israel, with implementation at their call centers scheduled by the end of 2007. Plus, we are in the process of negotiating with several others.

According to Frost & Sullivan, this is a fast growing market that is due to take off from $80 million turnover this year to $510 million by 2011.

Q: Where does this fit into the kinds of work that other call center companies are doing, like speech analytics and emotion detection? Is this a niche product from the call center point of view that would be added as a module onto a voice processing system? Would you sell it into the call center market that way? That way you remain a security company rather than a call center company.

CellMax Systems is a speaker identification and speaker verification technology. What we do is increase security and reduce cost during the call with the agent. If a need arises for speech analytics then our capabilities can serve as an additional feature to the analytics solutions from providers such as Witness, Verint, Nice and others. We haven't seen strong demand for speech analytics from our customers, although we do have free-speech identification capabilities as part of our range of technologies.

As for emotion detection, this is a tool capable of pinpointing cases where a person's voice indicates tension or other signs of instability. It can serve as an adjunct to improve customer routing and treatment but it cannot serve as an authentication tool. As such, it can neither replace current manual authentication methods nor can it increase the authentication level. So that is a niche product.

Regarding voice biometrics, I think that, for call centers -- especially those handling financial services, healthcare and the like -- authentication is an absolutely fundamental need and not a niche. That it's of major concern is supported by current legislation and regulation as well. So there's a very strong push for change on the part of operators along with high motivation to save costs on the human-based authentication methods used today, which is where we come in. You have to bear in mind that voice biometrics is the only biometrics technology that can be used effectively over phone lines.

You're correct in noting our roots in security and surveillance -- many of our team served in the Israel Defense Forces -- and we continue to provide solutions to the defense and law enforcement sectors. Our specific focus within the call center space continues to be comprehensive security. We think that an implementation of speaker recognition needs its own integration infrastructure, inclusive of fine-tuning the VUI and security levels to make a given installation effective and strong. We don't view our product as a part of someone else's but as a stand-alone capability.