“In the field of multimedia forensics, experts have proposed the VSI model, which accurately identifies individual speech source devices through multi-scale feature fusion and dynamic enhancement, providing strong technical support for judicial practice and information security.”