Example: Identification Rules and automatic enrollment

In automatic enrollment, the system uses an Identification Rule to associate a voiceprint model to an interaction attribute value that is unique to a specific user. This association enables the system to find the voiceprint model for that user when the user participates in an interaction.

Example: Customer Identification Rule for automatic enrollment of customers.

To support automatic enrollment of customers, you create a Customer Identification Rule.

In this example, we assume you create a Customer Identification Rule that includes a custom attribute named AccountNumber (where the value of AccountNumber is a unique account number for a specific user).
You assign this Customer Identification Rule to a Recorder Analytics Rule when you configure a Recorder Analytics Rule to perform automatic enrollment on customers.
When the Recorder Analytics Rule is invoked on an interaction, it uses the Customer Identification Rule attribute when analyzing the audio channel that carries the customer’s voice. In this example, the system looks for the value of the AccountNumber attribute in the metadata on this channel.
The system determines the value of the AccountNumber attribute in the customer audio channel metadata.
- If the system has never seen this AccountNumber attribute value on an interaction before, the system creates an initial voiceprint model using the audio from this interaction. The system associates this voiceprint model with the AccountNumber attribute value and stores the voiceprint model in the Biometrics database.
- If the system has seen this AccountNumber attribute value on an interaction before, the system retrieves the existing voiceprint model associated with this attribute value from the Biometrics database.
  
  The system enhances this voiceprint model using the audio from the interaction. This enhancement process continues with each subsequent interaction that contains this AccountNumber attribute value until a configured number of interactions containing this attribute value are handled by the Voice Enrollment Engine.
  - Before the minimum number of interactions have been found, the user’s voiceprint model is not valid for identity verification operations.
  - After the maximum number of interactions have been found, the voiceprint model enhancement process is complete and no further enrollment occurs.

The automatic enrollment process for employees is basically the same as for customers. The differences are:

In step 1, you create an Employee Identification Rule (rather than a Customer Identification Rule).
In step 3, the system uses the Employee Identification Rule attribute to analyze the audio channel that carries the employee’s voice (rather than using the Customer Identification Rule attribute to analyze the audio channel that carries the customer’s voice).
In step 4, the system uses the the value of the Employee Name from the Employee Identification Rule (rather than the value of the attributes specified in the Customer Identification Rule).