Using voice biometrics for authentication
Voice biometrics take the aggravation out of authentication and help organizations verify the actual person behind the interaction based on the sound of their voice.
How does voice biometric authentication work?
Authentication is often the worst part of the customer experience. Do you remember your 13-digit account number? That password you created a year ago? Your childhood pet’s nickname? Traditional security factors like these are frustrating, unreliable, and vulnerable to fraud. Voice biometrics verify the actual person behind the interaction, rather than something they know or something they have.
Nuance Gatekeeper, our cloud‑native biometric security solution, makes it easy for organizations of any size to streamline authentication and prevent fraud. A person just speaks naturally, and Gatekeeper analyzes their voice to determine if it’s the real customer or a fraudster in a matter of seconds. This can be used in the call center with live agents, in your IVR, or in web, mobile, and messaging channels.
What is a voiceprint?
A voiceprint is not a recording. It’s an intricate, mathematical representation of the anatomical and acoustic factors that make each person’s voice one‑of‑a‑kind.
Nuance uses a type of AI called a deep neural network to analyze each customer’s voice against millions of parameters and create a single sophisticated and encrypted model—their voiceprint.
How secure is a voiceprint?
Account credentials can be stolen and purchased on the dark web, phone numbers can be spoofed, and SMS texts can be intercepted. But no one else sounds like you. There are millions of factors that make your voice unique, which means it’s one of the best ways to verify your identity and shut down fraudsters as soon as they speak.
But what if a fraudster recreates or “deepfakes” a customer’s voice? The growth of widely accessible tools for generating synthetic speech has caused concerns around voice deepfakes and voice playback attacks. Even in these cases, Nuance Gatekeeper can detect small, telltale artifacts left on the audio signal and will distinguish between the live human voice and the counterfeit. Our speech scientists are constantly improving our voice algorithms to combat these modern threats.