Voice’s promise for replacing passwords with biometrics unique, Pindrop execs argue
Many passwordless efforts involve applying biometrics to replace vulnerable credentials within the enterprise and for accounts accessed through the online or mobile channel.
For Pindrop, it is the unique ability of voice to be used to communicate or exchange data and perform authentication at the same time that gives it a decisive advantage over other biometrics.
What is most exciting to Pindrop CTO Collin Davis and CMO Mark Horne is the range of applications that voice biometrics can now improve by providing passwordless authentication or personalization, as they explain to Biometric Update in an interview.
The rise in multi-factor authentication, particularly in regulated industries, is also an opportunity for voice, which typically is delivered over the phone, which provides a possession factor. Combined with increased demand for personalization in digital products and services, and increasing use of voice, Davis and Horne see passwords as increasingly redundant.
Contact center core
The most common deployment of Pindrop’s voice recognition is by contact centers, which use different signals to narrow their authentication down to a 1:1 biometric match.
“Having said that we also do one to many searches, because we look for known fraudster voices,” Davis clarifies. “So we have a consortium of known fraudsters voices that we will search. And that’s a long list.”
Fraud rates experienced by banks and fintechs, already high and growing, have increased during the pandemic, a situation explored by Davis in a guest post for Biometric Update earlier this year.
Pindrop’s voice biometrics can authenticate a known speaker in less than a second, and perform enrollment sufficient for high-confidence matching in between five and ten seconds. Combined with other factors like the device the speaker is calling from, the authentication can easily be performed without involving a password, during the normal flow of a call to the contact center of a bank or other service provider, Davis suggests.
Davis also notes the company’s algorithms’ performance in presentation attack detection, noting that the company was able to pick out all of the deepfaked passages in a recent Anthony Bourdain documentary.
The company plans to continue to focus on the contact center market, but is also expanding its core to support developers working on a range of and offline use cases.
Pindrop has a partnership with TIVO to provide personalization technology, but has also spoken to large auto manufacturers, and sees a wide and ever-growing range of potential applications.
Long a Pindrop customer group for the contact center channel, banks are struggling with password overflow, according to Horne.
Rethinking a process which is now more often carried out remotely, Horne says, banks are even considering the possibility of using voice as a baseline identifier when the account is set up.
“When the new account is provisioned, you use voice to identify is it the right person,” he says. “Use voice to deal with synthetic fraud on day one. Build a richer, more detailed profile of the individual with age and those things.”
Horne also has a background in ATM technology, so he knows that there is already a microphone present in most of them, as well.
The possibilities, Davis says, are expanding all the time.
“That’s why we’re excited about the API,” he explains. “One of the beauties of a developer-based API is it sort of unlocks the creativity of the developers that are out there, and I’ve already been surprised by the use cases that have crossed my inbox.”
He gives an example of a company in India using Pindrop’s biometrics to detect online bullies.
Horne says that while the company does have to educate customers about what can be done with voice, when Pindrop talks to many of new businesses, they are attempting to address a major problem, such as the weakness of the credentials they are currently using. “Nobody wants to admit it but they’ve all been stolen,” he states.
The company sees significant promise in IoT applications as well. For those IoT applications, Davis explains that they are probably 1:N biometric matches “but N’s pretty small. Like how many people live in your house.”
Like any biometric, voice data is sensitive and security around it must meet a high bar, Horne says, but he also notes that consumer understanding about how to safely use new technologies inevitable changes over time.
“Comfort with this technology has increased,” Horne asserts. “That’s due to the likes of Siri and Google.”
Consumer comfort with voice biometrics may have even increased enough to finally replace passwords, from the contact center to the IoT.