Passive authentication in the contact center

Gatekeeper for passive authentication is a text-independent passive customer authentication platform used in a contact center. It can:

  • Verify a caller’s identity during the first few seconds of a natural conversation with a contact center agent and compares to the voiceprint that is stored in the database in seconds without interrupting the call.
  • Do this in any language, with any accent, and call quality regardless of the content of the call.
  • Eliminate completely the need for PINs, passwords, and security questions when a caller is interacting with an agent in a contact center.

This solution ensures that fraudsters cannot spoof the system and voiceprints cannot be used if stolen, keeping customers secure from fraud. Gatekeeper incorporates scores of unique features to ensure that a fraudster cannot spoof the system. This system looks for anomalies in pitch, liveness, and caller consistency and generates an alert if it finds synthetic speech or a recorded voice. Even professional voice imitators cannot fool our system.

Gatekeeper continually updates voiceprints by using new audio to update existing voice Templates and automated enhancement of voiceprints based on the analysis of failed authentication attempts by legitimate users. Gatekeeper maintains an accurate voiceprint even with changing background noises and voice tones that shift with age. A typical flow for a customer contact center involves the following:

  • Securing enrollment by ensuring that the agent is speaking to the true customer.
  • The customer simply needs to speak to the agent like they normally would, while the system captures enough audio to enroll the voiceprint behind the scenes. The system also collects more data about the call to increase the layers of security for the customer.
  • Depending on the organization, consent may need to be collected before enrolling a print profile.
  • Once the voiceprint is created for the customer, the next time they call in to the contact center, they are authenticated in seconds based on the characteristics of their voice.