Beijing, March 16 (SANA) A study by researchers at Tianjin University and the Chinese University of Hong Kong found that most people struggle to distinguish real human voices from those generated by artificial intelligence, even after specialized training.
In the experiment, 30 participants listened to audio recordings while undergoing brain scans as they attempted to determine whether the voices were human or AI-generated. Results showed that most participants could not reliably tell the difference, with training providing only limited improvement.
Lead researcher Xiang Bin Ting said training enhanced the brain’s response to human speech, suggesting the auditory system can detect subtle differences between the two types of voices even if participants cannot consciously identify them.
The findings follow earlier research by Queen Mary University of London warning that deepfake voice technology can now produce audio almost indistinguishable from real human speech.
K.A.A
