Voice analysis aims to spot phone scams

Nagoya University researchers and Fujitsu Ltd. have developed technology they say can analyze suspicious phone conversations and detect bank transfer scams with a high degree of accuracy.

The technology detects what the researchers call the "overtrust" condition, in which a person believes the information given by a scammer based on the victim's voice pitch and tone. If this technology is installed on phones and successfully warns targets, it is expected to help prevent bank transfer scams.

Kazuya Takeda, a professor, and other researchers at the university focused on the fact that a person's pitch and tone becomes flat in the high-frequency range when he or she is overwhelmed with distressing information. This often results in a diminished capacity to evaluate information - a situation called overtrust, the researchers said.

Perpetrators of phone scams often pretend to be victims' relatives, saying distressing things like "I'm in trouble," or "I was involved in a crime."

The researchers detected overtrust with 90 percent accuracy in tests in which they examined 50 simulated phone-scam conversations using the technology.

The research group said it would be possible to detect phone-scam conversations by simultaneously analyzing keywords often used by criminals, such as "debt" or "compensation."