Voice Biometrics: Building Its Business Base as New Uses Emerge
Biometric solutions built on voice recognition technologies are projected to create an incremental opportunity of $4.5 billion between 2017 and 2025, according to Persistence Market Research. By the end of 2025, the firm expects the global market for face and voice biometrics—which are increasingly being blended into one solution—will have reached a value of $24 billion, following seven years of 18.1 percent compound annual growth.
Still, despite those very upbeat projections, voice biometrics technology is still primarily used in government and financial services for security and fraud prevention applications and will be for some time.
Financial services companies and their customers are among the most interested in voice biometric solutions. A Visa survey released in December showed that 86 percent of Americans preferred some form of biometric identification over PINs and passwords for authentication. Nearly one-third (32 percent) said they were interested in voice biometrics as an authentication solution.
Ease of use is among the greatest reasons for its wide appeal, according to Visa, which also found that nearly half of consumers (49 percent) have abandoned online purchases because they forgot their passwords, and another 16 percent abandoned purchases because entering passwords was too much of a hassle.
Voice biometrics has been around since the 1990s, but it has really started to gain prominence only in the past few years due to continuing hack attempts by fraudsters and the need to quickly, securely identify customers.
Indeed, “2017 was a record year for hacks of personal customer details,” noted Robert Weideman, executive vice president and general manager of the Enterprise Division at Nuance Communications, in a recent industry report. “These breaches give fraudsters access to our identities, including the answers to those annoying security questions. One thing the fraudsters can’t do much with? Voice data. And that is why banks and telcos are increasingly replacing security questions with biometrics. With a few words of speech, voice biometrics can confirm you are who you say you are at accuracy and security levels better than PINs, passwords, and security questions. And it knows how to detect recordings from real, live speech, rendering the data useless to fraudsters in the case of a breach.”
Contact centers that want to reduce the risk of exposing sensitive customer information to fraudsters while easing the cumbersome customer identification and verification process should invest in this technology, experts contend.
It’s better than before, and more vendors are delivering these capabilities, says Donna Fluss, president of DMG Consulting.
At the same time, though, the biggest impediment to adoption is the lack of knowledge about and understanding of this tool by the very consumers it’s meant to protect, Fluss says.
“The verification process has turned into a torture session,” she explains. “It can take anywhere from 15 to 45 seconds to verify a customer [with traditional methods]; the best answer is voice biometrics. It’s a technology whose time has come; there’s been increased adoption of it in the last year, more than ever before.”
Visa and other financial service providers recognize that voice biometrics is a leading technology for positive customer identification, particularly with the growth of mobile transactions, which are quickly surpassing online transactions.
“Mobile-initiated payments are ideal for incorporating a wide range of biometrics (e.g. voice, face, fingerprint, iris), given the ubiquity of smartphones and the ease of implementation for both in-store and in-app payments,” Visa said in its report. “The promise of eliminating passwords in exchange for a more convenient biometric solution addresses a universal problem shared by nearly all consumers.”
In fact, mobile transactions present perhaps the biggest market opportunity for voice biometrics. Nuance has integrated its VocalPassword into a number of mobile apps, and it has also taken part in multifactor approaches, such as the one offered by USAA for members using its mobile app.
Going Beyond Security
On other fronts, voice biometrics as a technology has advanced to the point that companies are deploying it in applications outside of security and fraud detection.
For example, Nuance recently added voice biometrics to its Dragon Drive automotive system and enhanced the product’s audio processing technology to accommodate the technology.
Dragon Drive’s new “Just Talk” feature allows vehicle occupants to engage with the automotive assistant without having to press buttons or speak specific words to activate it; an enhanced multipassenger capability allows multiple people in the car, not just the driver, to interact with the automotive assistant.
Voice biometrics can also be used to identify multiple drivers, allowing the automobile to automatically move the driver’s seat into the preferred position, set the air conditioning or heat to the desired level, adjust mirrors, and set the radio to the preferred station, depending on which driver enters the vehicle and says “Hello, car,” or some other designated phrase, according to Brett Beranek, director of product strategy for voice biometrics at Nuance.
“Voice biometrics is the easiest, safest way to do this,” he says. “You wouldn’t want just anyone to be able to change the settings.”
As we move the calendar forward, look for voice biometrics to move into other areas as well.
The technology itself continues to evolve, with improvements in recognition rates and ease of use, according to Fluss.
Indeed, according to Beranek, only a few years ago, 15 seconds of speech was needed for a positive ID (or rejection); now only three seconds are needed. The improvement comes from the incorporation of neural networks, which Nuance started to use in 2015. Its deep neural network evaluates 1,000 different speech elements.
The neural network also helps overcome voice recognition challenges as people age, Beranek adds.
A recent report from voice recognition company Pindrop pointed out that the human voice’s speed and pitch can change over months and years. The changes are minute and wouldn’t be so obvious to untrained ears, but voice detection systems could be thrown off by them.
Pindrop’s researchers claimed that error rates in voice biometrics can double over a two-year period. As part of the research, the team studied Barack Obama’s daily addresses between 2009 and 2017 for changes in his speech patterns and found that the voice accuracy rating for Obama declined 23 percent over the eight years.
Incorporating neural networks into voice biometrics allows the technology to adjust so that it can still positively identify a speaker despite these changes, Beranek says. Voice biometrics has been shown to positively identify some speakers even over the span of 30 years, he points out.
Another recent augmentation of voice biometrics, according to Beranek, is the inclusion of speech analytics. With this addition, the technology can now look for patterns in speech, keywords, tone, pitch, and other factors to help prevent fraudsters from using synthetic speech or recordings to try to circumvent the system.
But for all of these gains, justifying the cost of voice biometrics technology can be difficult, Fluss cautions. “The challenge has been benefits, cost, and return on investment. It’s tough to achieve a positive ROI.”
Because of the high price tag still associated with voice biometrics, the technology is neither practical nor feasible for all companies, experts warn. Those that will benefit the most are companies with a high percentage of repeat callers, typically financial services firms, healthcare providers, retail companies, and companies with a sizable number of business-to-business accounts.
Though ROI challenges persist, the voice biometrics market continues to have huge potential, according to Fluss. “Companies that haven’t considered it in the past are considering it more.”
The Voice Health Summit explores all the ways speech technology and AI are transforming healthcare.
If you own a smart speaker, you know that it can be fun trying to trick Alexa, Siri, or Google into doing or saying something it shouldn't—like obeying your friend who imitates your voice commands. While such ruses are fun and harmless, the truth is that bad actors are undoubtedly attempting trickery of a more nefarious nature and voice-controlled systems (VCSs) and speech recognition systems (SRSs) can be easily fooled via clever techniques.