Bayesian room-acoustic modal analysis

An important element of architectural acoustic design is the prediction and mitigation of modal effects. Modal effects can create undesirable aural phenomena, especially in smaller rooms. The use of Bayesian methods for spectrum analysis was pioneered originally in the field of nuclear magnetic resonance spectroscopy. A time-domain model may be used with methods of Bayesian inference to analogously analyze room modes, accounting for frequency, amplitude, damping, and phase. Bayesian model selection can be used to determine the number of modes present in the signal, while simultaneously estimating the associated modal parameters. Accuracy of the method is demonstrated using simulated data, addressing scenarios like well-spaced modes, overlapping modes, and unevenly damped degenerate modes. Then, the method’s usefulness is demonstrated by running it against real, measured impulse responses. Finally, the Bayesian model-based approach is compared to more standard frequency-domain analyses.