This article explains how to manage and record custom voice clones for use in realistic voice phishing (vishing) simulations. By creating clones of key internal personnel, you can build highly persuasive multi-channel simulations to test your organization's defenses.
🧭 1. Navigating to Voice Clones
To begin managing your organization's voice library, follow these steps:
In the revel8 platform, go to Settings → Context Collector.
Select the Voice Clones tab at the top of the dashboard.
➕ 2. Adding a Voice Clone Request
Click the Add button in the top-right corner to start a new clone request
Enter the personal information, which determines how the AI identifies as the caller and what role that person has:
Field | Impact on Simulation |
First & Last Name | Used to identify the person calling or used for email signatures. |
The address where the recording invitation link will be sent to. | |
Phone Number | Used for reference in Signatures within Emails potentially (Optional). |
Job Title | Used for reference in Signatures within Emails potentially. |
Role | Defines the individuals role (e.g., CEO, Finance, HR) used by the simulation engine to match the voice to the right simulations. |
Language | Determines the language the AI assistant will use during the recording process - this should be the language used by the individual.
Note: If you record for example in German, we can still schedule Voice Calls in other languages such as English automatically. |
After completing the form, click Submit to trigger the invitation email.
🎙 3. The User Recording Flow
Once a request is submitted, the individual being cloned will receive an invitation to record their voice sample. The process takes approximately 2–3 minutes and is guided by an AI assistant.
Email Invitation: The user clicks Start Your Voice Recording in the invitation email.
Note: The email will be sent automatically from the domain info.revel8.ai with the Subject Voice Clone Recording AssignmentEnvironment Check: The user is prompted to choose a quiet space and use a good-quality microphone.
(If required) Microphone Access: The browser will request permission to use available microphones.
Active Recording: The user clicks Get started and speaks naturally, answering prompts provided by the AI.
Review & Submit: After the recording is processed, the user can Listen to their sample, Re-record if necessary, or click Submit to finalize.
Upon a successful submission, the user will see a confirmation screen and can safely close their browser.
📊 4. Understanding Voice Clone Status
You can monitor the progress of each voice clone request directly from the Voice Clones dashboard.
Pending: The request has been created in the system but the invitation has not yet been sent.
Sent: The invitation email has been delivered to the user.
In Progress: The user has clicked the link and started the process but did not yet submit the recording.
Completed: The voice sample has been successfully recorded and processed. The clone is now ready for use in simulations.
Failed: An error occurred during the recording or processing phase.
Expired: The recording link has timed out and is no longer valid.
💡 Tip: If a request remains in "Sent" status for too long, you can use the Send again button to resend the invitation email to the user.
🧪 5. Testing a Completed Voice Clone
Once a voice clone reaches Completed status, you can verify the quality by running a test execution within the Simulations section.
Navigate to Simulations
Locate a simulation template that supports voice (e.g., "CEO Fraud") and click Test Simulation
Select Employee: Choose the target for the test (typically your own record will be auto-filled if you are an existing employee within the platform)
Voice Clone Selection: In the Voice Clone dropdown, select the specific person and role you wish to test (e.g., "John Smith - CEO")
Run Test: Select your preferred locale (language) and click Run test execution
⚠️ If you have missing fields or properties to run the simulation, these will be flagged and need to be filled first or added manually. This could be due to missing context or missing employee attributes for the selected employee.