SGM (Spectral Graphical Models): SGM is an R package for learning a latent graph from stationary signals. It currently offers code to fit a first-order Graphical Autoregressive Model (GAR(1)), and a first-order Time Varying Autoregressive Graphical Model (TARGAR) as proposed by Peng, Paul, and Harwood (2024). It is available for download at https://github.com/jed-harwood/SGM. These models are useful for graph inference when the data can be viewed as a graph diffusion process. For more information, please see our preprint.