Voice Biometrics Engine settings

Use Voice Biometrics Engine settings to configure the system to perform watch list List that contains one of more voiceprints of people who are of particular interest to an enterprise. detection Function of the IAFD product that compares a caller’s voice to one or more voiceprints in a watch list to detect if a target is participating in a call. and identity verification Feature in voice biometrics where the speaker’s voice is compared to a collection of employee or customer voiceprints that should match the call. If a match occurs, the speaker is said to be verified..

The IAFD > Fraud Detection or IAFD > Identity Authentication license must be enabled on the System Management > Licensing Management > System Licensing screen for the Voice Biometrics Engine configuration screen to be available in the user interface.

Voiceprint Detection

Setting

Description

Available Watch Lists

The Recorder Analytics Rule does not use the watch lists in this box.

If you want the system to use a watch list in this box to perform watch list detection, click on the watch list, and then click the right-pointing arrow Add to Selected icon to move the rule into the Selected Watch List box.

In a Multi-tenant SaaS environment, only watch lists associated with the same tenant as the Recorder Analytics Rule are available.

Note: To disable watch list detection for this Recorder Analytics Rule, move all watch lists into this box.

Selected Watch List

The Recorder Analytics Rule uses the watch lists in this box to perform watch list detection.

When this Recorder Analytics Rule is invoked on a selected interaction In Speech Analytics, an interaction represents a single part of the contact between one employee and the same customer. In Text Analytics, an interaction is the communication session between one or more employees and the same customer with a unifying contextual element., the system compares the voice on the interaction to the voiceprint File containing a mathematical summary of the vocal patterns of a person’s voice used in the IAFD product to assist with identity verification and detection. models that are included in these watch lists to perform watch list detection.

If you do not want the Voice Biometrics Engine to use a watch list in this box to perform watch list detection, click on the watch list, and then click the left-pointing arrow Remove from Selected icon to move the rule into the Available Watch List box.

In a Multi-tenant SaaS environment, only watch lists associated to the same tenant as the Recorder Analytics Rule are available to be selected.

Note: If no watch lists display in this box, this Recorder Analytics Rule does not perform watch list detection.

Minimum Score

A watch list detection score must exceed this value for the system to conclude that a watch list detection has occurred.

If the detection score exceeds this value, the Analytics system returns a result of “Detected” for the watch list detection operation.

The default setting is 20. (The valid range of settings is -100 to 100.)

Note: The system ignores this setting if the rule also performs verification. For a complete explanation of the term One to five words that are meaningful to a specific type of business, or phrases that stand out in interactions in Speech and Text Analytics. “detection score” and how minimum scores work, see the Real-Time Analytics (RTA) Framework Configuration and Administration Guide.

Minimum Score If Verified

This setting specifies the detection score that must be exceeded for the system to conclude that a watch list detection has occurred when a successful identity verification occurs on the same interaction as the watch list detection.

If the detection score exceeds this value, the system returns a result of “Detected” for the watch list detection.

The default setting is 20. (The valid range of settings is -100 to 100.)

Note: For a complete explanation of the term “detection score” and how minimum scores work, see the Real-Time Analytics (RTA) Framework Configuration and Administration Guide.

Minimum Score If Not Verified

This setting specifies the detection score that must be exceeded for the system to conclude that a watch list detection has occurred when an unsuccessful identity verification occurs on the same interaction as the watch list detection. If the detection score exceeds this value, the Analytics system returns a result of “Detected” for the watch list detection.

The default setting is 20. (The valid range of settings is -100 to 100.)

Note: For a complete explanation of the term “detection score” and how minimum scores work, see the Real-Time Analytics (RTA) Framework Configuration and Administration Guide.

Analyze Audio From:

Specify the audio streams that you want the Recorder Analytics Rule to analyze when performing watch list detection.

The options include:

Employee - Select this option to analyze only the audio stream outbound from the enterprise device (for example, the audio stream on which the employee in the enterprise speaks).

Other Party (Default) - Select this option to analyze only the audio stream inbound to the enterprise device (for example, the audio stream of a customer contacting the enterprise or an employee who contacts another employee in the enterprise).

Both Parties - Select this option to analyze the audio streams of both the Employee and the Other Party.

Usage Notes:

Accept the default setting of Other Party if you are interested in performing watch list detection only on people who contact The entire communication experience for a customer, from beginning to end. employees in your enterprise.

Many enterprises have no reason to perform watch list detection on their employees.

If you select Both Parties when you are interested in performing watch list detection only on people who contact employees in your enterprise, system resources are wasted analyzing the employee audio stream.

Voiceprint Verification

Setting

Description

Customer Identification Rules

Select one of these Identification Rules to enable the system to perform identity verification on customers (people who contact, or are contacted by, employees in the enterprise).

An Identification Rule selected for this setting:

  • Is applied to the audio channel of the interaction that is inbound to the enterprise device (the channel on which the person who contacts, or is contacted by, the enterprise is speaking).

  • Specifies the interaction attribute(s) the system examines to retrieve the voiceprint model associated with the specific customer that is speaking on the interaction.

If you do not select a Customer Identification Rule, the system does not perform identity verification on customers.

If you do not select either a Customer Identification Rule or an Employee Identification Rule, this Recorder Analytics Rule does not perform identity verification.

In most cases In Risk Management, use cases to group interactionss according to the needs of the enterprise. Interactions can reside in multiple cases simultaneously., if you select a Customer Identification Rule, you would not also select an Employee Identification Rule. Generally, a Recorder Analytics Rule that performs identity verification of customers should execute a different action than a Recorder Analytics Rule that performs identity verification of employees.

Examples of different actions for customer and employee verification results:

  • A customer passes customer identity verification. The Recorder Analytics Rule might notify the employee on the interaction using desktop integration. The employee can then use the notification information to determine if additional customer authentication is required.

  • An employee fails employee identity verification. The Recorder Analytics Rule might notify a security manager in email. The security manager can follow up at his or her convenience.

Employee Identification Rule

Select one of these Identification Rules to enable the system to perform identity verification on employees (people who make or answer interactions for the enterprise that is recording interactions).

An Identification Rule selected for this setting:

  • Is applied to the audio channel that is outbound from the enterprise device (the audio channel on which an enterprise employee speaks).

  • Specifies the interaction attribute(s) the system examines to retrieve the voiceprint model associated with the employee speaking on the interaction.

If you do not select a specific Employee Identification Rule, the system does not perform identity verification on employees.

If you do not select either a Customer Identification Rule or an Employee Identification Rule, this Recorder Analytics Rule does not perform identity verification.

In most cases, if you select an Employee Identification Rule, you would not also select a Customer Identification Rule. Generally, a Recorder Analytics Rule that performs identity verification of employees should execute a different action than a Recorder Analytics Rule that performs identity verification of customers.

Examples of different actions for customer and employee verification results:

  • A customer passes customer identity verification. The Recorder Analytics Rule might notify the employee on the interaction using desktop integration. The employee can then use the notification information to determine if additional customer authentication is required.

  • An employee fails employee identity verification. The Recorder Analytics Rule might notify a security manager in email. The security manager can follow up at his or her convenience.

Minimum Duration required for Verification Failed

Enter the minimum amount of time (in seconds) an interaction must last before a “Not Verified” or a "No Attributes" result can occur. This setting controls how long an interaction must last before a verification failure will be reported.

  • If a positive verification result occurs at any point in the interaction then the result is “Verified” (configured duration does not apply in this case).

  • If a positive verification result does not occur and the configured duration elapses then the result is "Not Verified". In this case, the required metadata and voiceprint to identify the person are found, but the minimum score for a match is not met.

  • If a voiceprint is not found then the result is "No Voiceprint" (configured duration does not apply in this case).

  • If the metadata needed to identify the person to be verified is not present and the configured duration elapses then the result is "No Attributes".

It is possible for there to be no result generated at all if a positive verification does not occur and the interaction ends before the configured duration. No result is returned under the following conditions.

  • The interaction (or INum for real time) is shorter than the configured duration, and one of the following is true:

    • The metadata is not sufficient to identify the customer

    • The metadata is sufficient to identify the customer and locate a voiceprint, but a good match does not occur with the available audio.

The default and recommended setting is 30 seconds. The maximum setting is 3600 seconds.

Adjust this setting if verification does not yield satisfactory results, such as an excess number of false “Not Verified” results due to short interactions with insufficient audio.

The Recorder Analytics Rule condition configuration determines what Action is triggered. Conditions are configured on the Conditions screen.

Minimum Score

An identity verification score must exceed this value for the system to conclude that an identity verification has occurred (that is, for the system to conclude that the voice on an interaction matches a voiceprint model).

If the verification score exceeds this value, the Analytics system returns a result of “Verified” for the identity verification operation.

The default setting is 5. (The valid range of settings is -100 to 100.)

Advanced

The Advanced Settings can affect the Recorder Analytics system performance and should be changed only under the direction of an authorized support representative.

Setting

Description

Process Whole Call

Select this check box to analyze all of the audio from each interaction when performing identity verification or watch list detection operations.

Analyzing all of the audio from each interaction increases the chances that the system can successfully perform an identity verification or watch list detection analysis. However, this configuration can consume more system resources and slow server performance.

This check box is not selected by default.

If you select this check box, the Max Audio to Process (Secs): setting is unavailable and cannot be used.

Max Audio to Process (Secs):

Specify the number of seconds of the interaction you want the system to analyze when performing identity verification or watch list detection.

Note: If you select the Process Whole Call setting above, this setting becomes unavailable. In this scenario, this setting is not used and all of the audio data for each interaction is analyzed.

Analyzing only a portion of the interaction reduces the chances that the system can successfully perform an identity verification or watch list detection analysis. However, this configuration consumes fewer system resources and results in faster server performance.

The minimum value for this setting is 4 seconds and the maximum value is 600 seconds.

Engine settings

Configuring Identification Rules for IAFD

Example: Identification Rules and automatic enrollment

Example: Identification Rules and identity verification