Source

Warning: Humans cannot reliably detect speech deepfakes

Peer-reviewed experimental study Mai, K. T., Bray, S., Davies, T., and Griffin, L. D. (2023). PLOS ONE

Claim this source supports

“Listeners identified deepfake speech correctly about 73 percent of the time, and prior familiarization training improved accuracy by under four percentage points, with comparable results in English and Mandarin.”

View original source

Cited in