AI voices cannot (yet) deceive humans
It is almost impossible for listeners to tell whether a real person or an AI-generated voice is speaking. However, according to a research team from the University of Zurich, the brain reacts differently to deepfake voices than to natural ones. In a study published in the journal Communications Biology, the researchers led by Claudia Roswandowitz report that fake voices appear to lead to less pleasure when listening to them. They analysed how well human identity is preserved in voice clones and conclude that humans can be partly tricked by deepfakes, but the neurocognitive mechanisms identified during deepfake processing open windows for strengthening human resilience to fake information.
Read complete news item (in German) at Tagesschau - Wissen: Link
Read paper in Communications Biology: Roswandowitz, C., Kathiresan, T., Pellegrino, E. et al. Cortical-striatal brain network distinguishes deepfake from real speaker identity. Commun Biol 7, 711 (2024). https://doi.org/10.1038/s42003-024-06372-6