Free AI Voice Cloning Test
Can AI steal your voice in 5 seconds? Say a few words and hear a vishing call in your own voice.
Works on your computer or your phone. No account, no install, nothing stored.
How the AI voice cloning test works
Four steps, about a minute end to end. The whole thing happens inside one browser session.
You start the conversation
Press start and allow the microphone. The AI answers in a stock voice and asks you an ordinary question.
It collects 5 seconds of you
A meter fills while you talk. Only your own speech counts, and one answer usually fills it.
It clones your voice
The model is built in a few seconds from that short sample, and it lives only inside this session.
It calls you back as you
The AI chats with you in your own voice for half a minute, then switches back to its own voice and explains how you would have caught it.
How voice cloning powers voice phishing (vishing)
Voice cloning stopped being a research demo some time ago. A current model needs around 5 seconds of clear speech to produce a voice that holds up over a phone line, where the audio is already compressed and nobody is listening for artefacts.
Attackers do not need to record you. Five seconds of you is usually public already: a conference talk, a podcast appearance, a webinar, a product video, a social clip, or the greeting on your voicemail.
Then they call an employee, normally in finance or on the IT helpdesk, and ask for a payment, a password reset, or an MFA code. The voice is doing the authentication, and it is the one credential nobody was asked to protect.
Lengths are typical, on a log scale so the short clips stay readable. Anything past the dashed line is already more audio than a clone needs.
How to spot a vishing call
You will not hear the clone. Three habits catch the call anyway, and none of them depend on noticing anything about the audio.
Call back on a known number
Hang up and dial the number you already hold for that person. Never the number you were just given, and never one read out to you on the call.
Agree a code phrase
A short phrase shared with your finance team, and separately with your family, ends a cloned voice call in one question. Rotate it the way you rotate a password.
Treat urgency as the signal
Pressure plus secrecy is the attack pattern itself. A genuine request survives a five minute delay and a check on a second channel.
Try our other free tools
Voice Phishing Simulator
Hear your own voice turned into a boss, a colleague, or an unknown caller, live in the browser.Voice Phishing Accent Simulator
Hear the same scam script in a different accent, and see how much accent alone changes who your employees trust.Frequently asked questions
Yes. There is no account, no card, and nothing to install. The test runs in your browser on a live microphone and ends when the demonstration finishes or when you end it.
No. The voice model exists only for the length of the session and is deleted when the session ends. There is no file to download and no recording kept afterwards.
Around 5 seconds of clear speech, which is why the meter stops at 5. Attackers rarely need to record anyone: a conference talk, a podcast, a webinar, or a voicemail greeting already holds more than enough.
It would not get you anything. You confirm that you are the person speaking before the session starts, the test only listens to a live microphone and never accepts an uploaded file, the cloned voice will not repeat what it is told to say or speak any number, name, payment or password, nothing can be downloaded, and the voice is deleted when the session ends.
Once every 30 minutes, so that everyone gets a turn on the demo lines.
Chrome or Edge on a computer, Safari or Chrome on an iPhone or iPad, and Chrome on an Android phone. Other browsers cannot stream a live microphone to the demo line reliably enough to clone from it.
Train your employees against vishing
One test shows you what a clone sounds like. Callstrike runs controlled AI voice phishing simulations across your whole workforce, then trains the employees who fall for them. No GPUs, no installs, no per seat pricing.