Overview
Deepfake Profiles are reusable AI voice profiles for authorized security-awareness simulations. A profile is created from an approved voice sample and can then be used to generate spoken messages for AI Voicemail landing-page templates.
Please contact the Support Team at support@phishingbox.com to inquire about this feature.
Open this page in the Portal by navigating to Templates > Deepfake Profiles.
Use the links below to move directly to a section:
- Page layout
- Profiles table
- Sandbox
- Create a voice profile
- Record with guidance
- Upload existing audio
- Profile readiness and availability
- Recording best practices
- Troubleshooting
Page layout
The Deepfake Profiles page contains three work areas:
- Profiles lists the voice profiles available to your account and shows each profile's current state.
- Sandbox lets you enter a short message and hear it with a ready profile before using that profile in a template.
- New Profile opens the guided workflow for recording or uploading a voice sample.
Profiles table
The table at the top of the page provides a concise view of the profiles owned by your account. It contains the following columns:
- Name identifies the voice profile. Use a clear name that helps template authors recognize the approved speaker or intended use.
- Status shows whether the profile is still being prepared, is ready for use, or needs attention.
- Sample provides sample playback when audio is available, allowing you to check the voice before using it elsewhere.
If the account has not created any profiles, the table displays No voice profiles found. Create a profile from the New Profile panel to populate the table.
Sandbox
The Sandbox is a test area for ready profiles. It does not add a voice to a template or campaign.
- Open the Voice Profile menu and select a ready profile.
- Enter the text that you want the profile to speak in the Message field.
- Select Play to generate and listen to the sample.
When no profile is ready, the menu displays No ready voice profiles, the Play button is unavailable, and the Portal prompts you to create an AI voice and wait for it to become ready.
Create a voice profile
In the New Profile panel, select Get started. The Create a Voice Profile window opens.
- Enter a descriptive name in Voice Profile Name.
- Choose Record with guidance or Upload existing audio.
- Complete the instructions for the selected method (jump to Record with guidance or Upload existing audio).
- Save the profile and wait for voice preparation to finish. Only a ready profile can be selected in the Sandbox or in an AI Voicemail landing-page template.
Record with guidance
Record with guidance is selected when the creation window first opens. The workflow contains seven prompts and shows both the current prompt number and completion progress.
- Read the displayed READ THIS ALOUD text before beginning so you can speak naturally.
- Select Record Prompt and read the complete prompt at a normal pace with clear pronunciation, steady volume, and natural pauses.
- Listen to the result. Redo the prompt if the recording contains a mistake, interruption, or poor-quality audio.
- Select Continue after the current prompt is complete. Use Back to return to the previous prompt.
- Repeat these steps until all seven prompts are complete, then finish the workflow to save the profile.
Start Over clears the guided recording progress. Cancel closes the creation window without finishing the profile.
Upload existing audio
Select Upload existing audio to create the profile from one existing recording instead of the seven guided prompts.
- Enter a name in Voice Profile Name.
- Select Upload existing audio.
- In Audio File, select Choose File and choose a clean recording.
- Finish the workflow to save the profile.
The uploaded file must be in webm, wav, mp3, ogg, or m4a format and must not exceed 25 MB.
Profile readiness and availability
Saving a recording does not necessarily make the profile immediately available. The Portal prepares the voice, and processing can take additional time.
- Ready profiles can be selected in the Sandbox and in an AI Voicemail landing-page template.
- A profile that is still training or being verified cannot yet be selected for message generation.
- If preparation fails, use the guidance displayed in the Portal. Depending on the profile's state, a status check or retry option may be available.
- A profile cannot be deleted while a template uses it or while audio generation for it is pending.
AI Voicemail availability is controlled for each account. If provider setup is required, an account administrator must connect and activate ElevenLabs in the Integration Store before the profile can be prepared or used to generate speech.
Recording best practices
Use only voice samples that your organization has approved for authorized security-awareness simulations. For the clearest and most consistent result:
- Record one speaker in a quiet room.
- Keep the microphone in a steady position.
- Speak at a natural pace and volume.
- Avoid music, echo, background voices, long periods of silence, and heavy noise reduction.
- Review each guided prompt before continuing, or listen to the table's sample when one is available.
Troubleshooting
No profiles appear in the Sandbox
Review the Status column in Templates > Deepfake Profiles. The Sandbox lists only ready profiles. If no profile is ready, finish creating one and allow preparation or provider verification to complete.
The Portal says ElevenLabs setup is required
Ask an account administrator to connect and activate ElevenLabs in the Integration Store, unless the account is expected to use platform-provided AI Voicemail access.
Profile preparation failed
Follow the message displayed in the Portal, then check the provider API key, subscription or credits, rate limits, and recording format. Use the available retry control after the issue has been corrected.
A profile cannot be deleted
Confirm that no landing-page template references the profile and that the profile is not involved in pending audio processing. Remove the dependency or wait for processing to finish before trying again.