Artificial intelligence sounds are identical to humans ... and warning

A research team from Queen Mary University in London revealed that the techniques of generating sounds with artificial intelligence, known as “deep counterfeiting”, have reached a level with which most people are difficult to distinguish between them and the real human voices..

According to the study published in the magazine PLOS ONEParticipants listened to real and artificial samples, some of which are reproduced from records of real people, while others were born from public audio models. The results showed that the artificial sounds seemed convincing and realistic to almost the extent of congruence with human voices.

The lecture in university psychology, Nadine Lavan, made it clear that the production of such records “no longer requires only limited experience, a few minutes of recordings, and commercially available programs at a semi -non -cost”, warning against the need to study how people realize this “artificial truth” in light of the accelerating progress of artificial intelligence.