Configuring Identity Authentication and Fraud Detection

Identity Authentication and Fraud Detection (IAFD) enables you to authenticate customers, verify employees, and detect persons of interest speaking on Interactions.

Voice biometrics

Identity Authentication and Fraud Detection utilizes voice biometrics. Voice biometrics uses the acoustic features of speech that have been found to differ between individuals and which create a recognizable pattern. Anatomy, such as the size and shape of the throat and mouth, determines the acoustic patterns produced.

Two key processes define the usage of Identity Authentication and Fraud Detection:

• Enrollment: Enrollment is the process of creating a biometric voiceprint model that is unique to each individual. Important points to understand about enrollment are:

  • Enrollment must occur before voice biometric analysis can occur.

  • The voiceprint model is a digital file. The generated file is a mathematical computation based on the speech of a person.

  • After a person is enrolled, the voiceprint model is said to be trained.

• Comparison: After voiceprint models have been created, the system compares the model to voice audio signals. The results enable you to determine how closely the voice biometric properties of a person speaking match the voiceprint model.

Metadata Detection

Metadata Detection detects the presence of a particular metadata attribute value in the metadata of a recorded interaction.

Two key processes define the usage of Metadata Detection:

• Metadata definition: Metadata Detection requires configuration of the metadata attributes and values you want to detect on interactions. Define the metadata attribute type and value. Creation of the metadata definition must occur before detection can occur.

• Watch lists: After creating metadata definitions, you organize them into watch lists, which are used for detection. When the system detects that a metadata attribute value is present, the system takes an action.

Topics