by George Nott

ANZ trials voice biometrics in app and call centre

Apr 03, 2017
Financial Services IndustryInnovationIT Management

ANZ will be adding voice biometrics to its mobile app so customers can authorise large payments with their voice. The biometrics technology, by Nuance, will also be introduced into the bank’s call centres to replace security questions.

“Voice biometrics is the next step in making banking more convenient for our customers while also strengthening security,” said managing director customer experience and digital channels, Peter Dalton.

“A person’s voice has five to ten times as many security points than other methods such as fingerprints so we know this will improve security and be welcomed by our customers.”

Customers transferring more than $1000 through ANZ’s mobile apps will be able to use their voice to automatically authorise payments, instead of using internet banking or visiting a branch.

A pilot will begin with ANZ staff and select customers in May using the Grow by ANZ mobile app. The service will then be rolled out to ANZ goMoney and other digital services progressively.

ANZ claims to be the first Australian bank to use the technology to improve security specifically on mobile devices. Voice authentication has been used for calls to the National Australia Bank (NAB) contact centre for a number of years.

The Australian Taxation Office introduced Nuance’s voice authentication into its call centres in 2014 and added the biometrics to its mobile app in 2015. Other banking users of biometrics include Barclays Bank and HSBC in the UK; ING in Europe; Royal Bank of Canada; and Wells Fargo and Bank of America in the US.

Safe and sound

Speaking to CIO Australia last month, Robert Weideman, general manager of Nuance’s enterprise division, said the technology could also help banks detect fraudsters, and had been used as evidence in criminal trials.

“We have had two people go to jail now where they signed up for Barclays Bank,” he said.

To create a voiceprint, hundreds of characteristics of an individual’s voice are codified, and run through a proprietary algorithm.

“It’s actually not a recording or print of your voice. The captured voice is then compared with the relevant stored voiceprint and a verification result is provided. In parallel, the captured voice is also compared to a fraudster voiceprint database, a blacklist, to check if the voice belongs to a known malicious individual.”

Often a fraudster will have set up a legitimate account with their true identification. If their voice is used to commit frauds, it can be traced back.

“Often though, you never know the fraudsters real name because they don’t give it to you,” Weideman says. “But the bank can avoid that voice in the future. If they call in again: this is a bad voice, send it off to the fraud department. And if that voice is ever identified truly, now they’ve got a whole bunch of evidence to use to prosecute.”

The technology is able to distinguish between a real voice and a recording of one, and between identical twins.

“One of the key challenges today for banking as the world becomes more digital is making it easier for customers to do what they want to do in a safe and secure way,” added Dalton.

“We also know that people are becoming more comfortable with using their voice to do basic commands on their devices, so we see this is a natural extension of current technology and we are expecting this to be a popular enhancement of our mobile apps.”

As well as it’s voice print technology, Nuance also develops voice recognition tools (such as that used in Apple’s Siri), and intelligent virtual assistants, such as those used by Domino’s and IP Australia.