January 17, 2020
Pindrop for Amazon Connect | A Balancing Act
Defending the phone channel presents various challenges, especially due to…
Voice biometric security is designed to recognize your customers and provide a frictionless experience, but what happens if your customers only call your business once a year or less? And how can your call center’s anti-fraud and authentication systems adapt to new kinds of attacks, like synthetic speech and voice conversion? Pindrop is hosting a three-part webinar series to answer these questions and more.
In the first session, we sat down with Pindrop Principal Research Scientist, Elie Khoury, to discuss the problem of voice aging in voice biometric security.
Khoury explained how every instance of speech involves 70 to 100 muscle interactions, including, by proximity, the hearing and nervous systems, as well as vocal cords and other muscles. Aging impacts all of these systems, thus changing different elements of your voice over time. For example, the degeneration of body tissues affects muscle mass and strength, which changes the pitch and tone of the voice. Similarly, hearing loss causes the speaker to talk louder, and aging may impact the nervous system, resulting in shakiness in the voice. As we move into the era of a conversational economy, voice aging becomes challenging for biometric security. Are your security and authentication systems prepared?
Pindrop® Labs found only 48% of customers called their banks during an eight-month period, and only 25-50% of the calls exited to the agent (range dependent upon IVR quality). Voiceprints taken at the first call to the bank would be significantly different after an eight-month duration of no calls. Aging decreases accuracy decreases due to alterations in pitch and voice speed, with noticeable degradation to voiceprints beginning after only four months.
For state-of-the-art voice biometric systems, voice aging is a real concern, as error rates may double in a three-year span. Unlike other sophisticated solutions, our Deep Voice™ biometric engine compensates for voice aging, as a text-independent system, passively identifying legitimate and fraudulent callers by voice alone. Passport, Pindrop’s multi-factor authentication system incorporates this technology to provide a friction-free customer experience without compromising call center security.
To learn more, view the on-demand webinar here.