Fake or real? Exploring brain signals for deepfake voice detection

Deepfake voice technology enables deceptively realistic voice clones that are used for fraud and manipulation. Neither humans nor machines can reliably detect deepfakes. The project investigates whether brain signals can capture unconscious distinctions between real and artificial voices. Neural measurements are combined with AI methods to develop more robust detection systems that also work with new generators and low-resource languages such as Swiss German. Together with the practice partner aurigin.ai, the project team explores transferring these findings to real-world applications.