Synthetic-speech patterns
Reviews usable speech for characteristics associated with AI-generated audio.
Loading…
Save the original voice note, upload the audio file and review the synthetic-speech signal. If the message asks for money, codes or secrecy, verify the sender through a number or channel you already trust.
Reviews usable speech for characteristics associated with AI-generated audio.
Shows the strongest available audio segment instead of hiding everything behind one label.
Keeps identity verification separate from the model score.
A high synthetic-speech score is a reason to stop and verify. A low score does not prove who sent the recording; stolen real audio and human scammers remain possible.
Important limitation: Forwarded, re-recorded, noisy or very short voice notes can lose useful evidence. Use the earliest original file available and never delay contacting a bank when money has already been sent.
Read the full methodologyPause before sending money after an unexpected urgent message.
Confirm new bank details or executive instructions through the normal approval process.
Never share a one-time code because a familiar voice asks for it.