Skip to main content
Every attacker speaks with a voice. You can start right away with a ready-made voice, or clone a real voice so the call sounds like a colleague. A cloned voice can only be used once its consent is recorded.
Before you start: cloning a voice needs the person’s agreement. For a director or executive, use Clone an executive’s voice with consent so they record it themselves.

​
Before you start

  • A workspace with the Voices segment. Open it from the Red Team tab and click Open voices.
  • For a clone: the person in the room, or a clean recording of them, and a quiet place.
  • A browser with a microphone, if you want to record rather than upload.

​
The six ready-made voices

You pick a ready-made voice on the attacker form, not on the Voices page. Choose Use a ready-made voice and pick one. No consent is needed; nobody real is copied. Sarah and George are shown first. Click More voices (4) for the rest. See Create an attacker.

​
Steps: clone a voice

1

Open Voices and click Clone a voice

The page is titled Voices. Click Clone a voice. A window opens, titled Clone a voice, with the line “Record a short sample, or upload one.”
2

Say whose voice it is

Type the person’s full name under Whose voice is this?. If your workspace has more than one voice engine, a Voice engine list appears; ElevenLabs is the default. Some engines also ask for a Voice language, so the accent is kept.
3

Record, or upload a file

The window shows a short passage under Read this out loud. Click Record, read it, then click Stop. The counter aims for 30 seconds. A sample must be at least 12 seconds; recording stops by itself at 90 seconds. Click Record again to redo it.To use a recording you already have, click Upload a file instead. Any audio file is accepted.
4

Confirm consent and save

Tick the box: “I confirm [name] is here and agrees to a copy of their voice being made and used in authorised DeepBlocker security awareness demonstrations.” Click Save voice clone. When it finishes you see Voice cloned, with Clone another and Done.

​
Voice languages

When an engine asks for a language, the list is: Each cloned voice shows one status. A voice with No consent shows Attest consent. Type the person’s full name, tick “I confirm [name] has given informed consent for their voice to be cloned and used in authorised security-awareness simulations run by this workspace”, and click Record consent. An executive voice cannot be attested this way; the person must record it themselves.

​
What happens next

A usable voice shows Create an attacker for this voice. Once an attacker uses it, the card lists the attacker with New campaign next to it, and Create another attacker for this voice underneath. The line near the top of the page shows how many executive voices the workspace holds, for example “1 of 3 executive voices used”. To remove a voice, click the bin icon labelled Delete this voice, then Delete voice. This frees a slot. Any campaign pinned to that voice will no longer be able to call with it.

​
Problems

The window says “The microphone is blocked. Allow microphone access for this site and try again, or upload a file instead.” In Chrome, click the lock icon next to the address, set Microphone to Allow, and click Record again.
The message reads “Please record at least 12 seconds for a good copy.” Click Record again and read the whole passage.
The workspace has reached its executive voice limit, counting invitations not yet answered. The message says so. Withdraw one, or write to contact@deepblocker.ai to raise the limit.
Open Voices and check the status. No consent needs Attest consent. Withdrawn cannot be fixed; clone the voice again with fresh consent, or pick another voice on the attacker.