Abstract: Spatial orientation and memory within architectural environments play a crucial role in environmental design and cognitive science; however, the mechanisms by which music, as a controllable auditory stimulus, influences spatial memory remain unclear. This study, based on dynamic virtual environments, investigates the effects of musical mode and tempo on the efficiency of spatial information memory, layout memory, and information-layout binding memory, while examining the mediating roles of emotional valence and arousal. Results indicate that major mode and slow-to-moderate tempo music significantly enhance memory performance under moderate to high cognitive load. Quantitative analyses of emotional responses reveal that the impact of music-induced emotional changes on spatial memory is moderated by task difficulty, and that valence and arousal exert independent effects. Mediation analysis further shows a significant interaction between musical mode and task difficulty, with their opposing effects forming a suppression effect on memory performance. In contrast, tempo exerts a complete mediating effect on layout memory through arousal, independent of task complexity.