Voice Biometrics


Secure and Convenient Authentication

Adding Voice Biometrics to our Intelligent Virtual Assistant allows you to verify a customer’s identity using the unique characteristics of the human voice — giving your customers convenient access to secure information, while enabling personalization in everyday transactions.

Secure Icon

Secure

Because voice biometrics enrollment occurs within the Intelligent Virtual Assistant, it reduces the risk of social engineering, making you less vulnerable to fraud.

Convenient Icon

Convenient

Verify your customers using their voice to make authentication easier and improve the customer experience.

Cost Icon

Cost Effective

Voice biometrics reduces Average Handle Time for live agents by 30+ seconds per call on average, by handling the authentication. The service is hosted and managed by Interactions, so no IT deployment and support is required.

Tuning Icon

Accurate

Interactions updates voiceprints with every call and the voiceprint is adapted for any changes in the caller’s voice, so accuracy remains high.

Authentication
Identification

How is voice biometrics used by IVAs?

Voice biometrics is a technology used to verify a person’s identity using their unique vocal attributes. Everyone has different physical and behavioral characteristics that influence the sound of their voice and in combination these are—just like a fingerprint—unique to every individual. This ease of use and added security makes voice biometrics the perfect replacement for or addition to traditional knowledge-based authentication processes.

Virtual assistants can use voice biometrics to confirm a caller’s identity as part of self-service interactions or before routing the caller to a customer service representative. Interactions Intelligent Virtual Assistant (IVA) voice biometrics can support both active and passive authentication. Active authentication using voice biometrics matches a caller’s unique voiceprint to a passphrase that is generated at the time of the call. Passive authentication using voice biometrics matches the voiceprint as the caller speaks naturally and no special phrase or password is required. Interactions voice biometrics is highly accurate, requiring just several seconds to enroll users and authenticate callers and continuously adapts to your customers’ biometric signature.

Does machine learning help with voice biometric authentication?

The science behind voice biometrics is not new, but effective business implementations in the area of customer authentication are made possible by advances in machine learning and computational power. Interactions continuously improves technologies including our voice biometrics authentication using machine learning and deep neural networks. Regularly improving voiceprints helps ensure we continue to provide accurate matches over time, significantly reducing the likelihood of fraud and improving the customer authentication experience.

How do companies use voice biometrics for customer authentication?

Many companies have turned to voice biometrics for customer authentication–specifically in the customer care space. Customer authentication via voice biometrics has significant advantages over legacy authentication methods such as knowledge-based questions which people often forget, and PINs and passwords that can be easily guessed, stolen, or extracted by social engineering. It also offers callers a secure, private, and convenient way to authenticate themselves either by itself or in conjunction with other forms of authentication. On subsequent contacts with customer service, voice biometric customer authentication is completed automatically during the interaction, significantly reducing customer frustration and effort, while increasing security in the face of increasingly sophisticated fraudsters.

Does voice biometric customer authentication always work?

While error rates of voice biometrics for customer authentication are significantly lower than facial or fingerprint biometrics, they are not zero. Therefore, while using voice biometrics is generally more than sufficient to complete low risk service transactions that represent the vast majority of calls today, it’s also effectively used as one factor in multi-factor authentication, or a layering approach to security. Interactions voice biometrics allows you to select how authentication is achieved, applying up to three authentication factors for a valid match. These factors include: 1) knowledge or something you know such as a password, PIN, or last four social security digits, 2) possessions or something you have, like phone ANI and 3) inherences, or something you are, such as your unique voiceprint. Interactions updates voiceprints with every call and the voiceprint is adapted for any changes in the caller’s voice, so accuracy remains high. Because voice biometrics enrollment occurs within the Intelligent Virtual Assistant, it reduces the risk of social engineering, making your customers less vulnerable to fraud.

How it works

As part of the Intelligent Virtual Assistant experience, Interactions collects customer consent and creates voiceprints.The voiceprint is securely encrypted and stored, and updated each time the customer calls customer service.

We support both Passive and Active authentication. Passive authentication requires no specific phrasing from the caller for voice biometric authentication. Active authentication requires the customer to speak a specific phrase — such as “My voice will be my password” — as agreed upon during enrollment. Liveness detection is also available for additional security.

Multi-Factor Authentication

Interactions Voice Biometrics provides a third factor for caller verification. The Interactions Virtual Assistant can apply up to three different authentication factors for enhanced security.

Knowledge

Something you know, including password, PIN, or the last four digits of a social security number, etc.

Possession

Something you have, including a phone ANI match or a device ID match

Inherence

Something you are, using the caller’s unique voiceprint for verification

Want to learn more? Let’s talk.