My research interests primarily revolve around physically informed models of audio processes, with a particular emphasis on the preservation and restoration of historic audio. The emergence of data-driven, gradient-based approaches to signal processing has opened new possibilities for recovering degraded recordings beyond what is achievable with classical DSP alone. My work explores how knowledge of the physical processes underlying historical audio technologies can inform these emerging methods, drawing on techniques from system identification, differentiable DSP, and analog modeling. Beyond presenting challenging and exciting engineering problems, this research offers opportunities to collaborate with colleagues in the humanities and at audio archives around the world and to provide students with hands-on experience working with historic media and audio systems.