This demonstration presents a real-time, interactive system for wide-area six-degrees-of-freedom (6DoF) Ambisonic rendering from a sparse set of first-order Ambisonic microphone arrays. The system implements a graph neural network (GNN) that predicts scene-dependent interpolation weights for Directional Audio Coding (DirAC) descriptors, while preserving an analytical spatial-audio rendering stage. Attendees can move and rotate a listener on a metric map, reposition the microphone arrays, switch between learned GNN interpolation and a Delaunay baseline, inspect the graph features, and immediately hear the resulting binaural rendering over headphones.