Bayesian integration of audiovisual speech by DNN models is similar to human observersCOSYNE 2025SaveMar 27, 2025Neuroscience