The State of Voice Biometrics

The need for voice biometrics to authenticate customers and patients and guard against fraud came to light even more as the COVID-19 pandemic moved an increasing amount of financial, retail, and other transactions online, says Brett Beranek, vice president and general manager of the security and biometrics line of business at Nuance Communications.

“In March and April, there was a shift in where companies were deploying voice biometrics to detect fraud,” Beranek says. The reason was twofold.

Though fraudulent phone calls looking to illegally obtain personal information or to have people send in payments has persisted for many years, phone fraud jumped as much as 400 percent when stay-at-home orders started in the spring, according to Beranek. “Fraudsters were trying to take advantage of the pandemic. They would pose as collectors and try to obtain someone’s personal information.”

Companies, particularly those in regulated industries like financial services and healthcare, also began deploying voice biometrics to authenticate not only their customers but also their contact center agents who had started working remotely.

“The requirement for most organizations is that agents will badge in to ensure they are complying with [personally identifiable information] rules,” Beranek explains. “With agents working remotely, that flew out the window.”

While many of the largest financial institutions had some measures in place to authenticate remote agents, fraudsters used the pandemic to drill down to try to scam smaller institutions, Beranek adds.

To show just how much voice biometric has grown in importance, Nuance had 600 million voiceprints and secured 8 billion transactions in 2020 compared to 400 million voiceprints and 5 billion transactions in 2019.

And as the technology demand has grown, so, too, has its capabilities.

“Ten years ago, we needed 10 seconds of audio to confirm an identity. Now we need only a half second of audio,” Beranek boasts.

Artificial intelligence and machine learning have been at the center of most technology advances, according to Beranek, who also cites a growing need for integration with video calling and conferencing systems like Zoom, Microsoft Teams, and Google Hangouts.

Year in Review

In mid-December, Nuance and Industrial Bank of Korea announced a biometrics-based fraud detection solution for video calls.

“IBK launched voice biometrics authentication formally to its customers in June 2020. Since then, the company has enrolled more than 100,000 customers,” Beranek says. “IBK is the first bank we’ve seen leverage biometrics for voice authentication during video calls—a trend we anticipate will only increase in the future. This is a perfect example of how to leverage AI to adequately protect consumers during a time of increased, advanced fraud attacks brought on by the reliance on digital experiences in recent months.”

Other technology vendors also stepped up their product innovation in the past year.

In December, Amazon unveiled a new voice recognition tool built into its suite of call center services. The Amazon Connect Voice ID suite uses machine learning and customizable confidence thresholds to authenticate customers’ identities when they contact call centers. If the threshold isn’t met, an agent or automated system can use additional authentication methods to confirm the callers’ identity.

Interactive Northwest (INI) partnered with LumenVox on a voice biometrics-powered Interactive Speech Attendant (ISA), which features a web-based administration interface that lets companies customize how inbound calls are handled and configure nicknames and aliases, department and location-based transfers, business hours logic, and alert messaging.

“CX-focused organizations understand there is only one chance to make a first impression,” said Danette Craig, INI’s president, in a statement at the time.

Earlier in the year, LumenVox released the newest version of its speech identification solution, featuring media server support for MRCPv2 over the transport layer security (TLS) protocol. TLS provides communication security between client-server applications for privacy, integrity, and protection.

NICE in November unveiled Enlighten, a voice-based fraud prevention solution for call centers. The solution does not use voice biometrics for authentication, but instead uses the technology to identify suspicious activity and build profiles of known fraudsters. The solution combines NICE’s customer engagement AI platform with its voice biometrics technology to scan millions of calls and then apply AI and predictive models to spot certain behaviors that have been associated with fraud. For example, address change requests might be suspect, especially if the caller does not seem too familiar with the original information.

NICE Enlighten also remembers the voices of suspected fraudsters and creates a ranked list of voice profiles that have been caught doing suspicious things.

Also in the fall, Aculab updated its VoiSentry voice biometrics system with enhancements in verification/identification algorithms; improved encryption for digital voiceprints; an extension to its presentation attack-detection function to mitigate against spoofing attempts; and other system upgrades that have cut its error rates in half.

A Look Ahead

The market for voice biometrics is expected to increase by $2.6 billion, at a compound annual growth rate of 15.9 percent, from 2020 to 2024, according to a new report from Technavio.

The research firm expects much of the growth to come from the banking and financial services sector, which has dominated the market for the past few years.

Financial services firms have long touted voice biometrics as the most secure, intuitive, and user-friendly of all modalities for verifying customer identities.

“In recent years, speaking to voice interfaces has become a normal part of our lives. We interact with voice-enabled assistants in our cars, smartphones, and smart devices at home,” says Raúl Navarrete, head of the mobile channel and smart assistants at Spanish banking firm BBVA.

“In the future,” he says, “voice biometric technologies will solve present user authentication issues when interacting with our bank or paying our utility bills.”

Expect to see other use cases emerge as customers become more familiar and comfortable with voice interfaces. Already voice-activated services are being used for speaker identity verification in customer service channels, for answering queries made through mobile banking apps, or by digital assistants such as Amazon’s Alexa, Google Assistant, or Samsung’s Bixby.

Deployment in the healthcare sector is expected to be a major driver of the technology’s growth in the next five years, according to Technavio.

“Recent years witnessed a rise in the use of mobile devices, such as smartphones, tablets, and personal digital assistants, in the healthcare sector, leading to the development of mobile apps for various applications, such as electronic prescribing, diagnosis and treatment, coding, and billing,” Technavio said in the report. “To prevent unauthorized access to confidential data, vendors will start integrating these apps with voice biometrics, which, in turn, will fuel the growth of the market in the coming years.”

Moving forward, voice biometric technology will need to continue to improve to stay ahead of fraudsters, who are constantly developing new technologies and techniques to defeat voice biometrics and other fraud prevention technologies.

One of the latest fraud technologies on that front is the deep fake, which is capable of re-creating a person’s voice. While the leading voice biometrics providers say they can identify many of the deep fakes in use today, they will need to continue to improve their solutions as fraudsters continue to evolve, experts agree. 

Phillip Britt is a freelance writer based in the Chicago area. He can be reached at spenterprises@wowway.