Determining Bayesian evidence and decay time estimates in acoustically coupled volumes

Abstract

Sound energy decay analysis is an important element in understanding acoustical properties of coupled volumes in architectural acoustics. A Bayesian model formulation using Monte Carlo algorithms [Xiang and Goggans, J. Acoust. Soc. Am. 110 1415–1424 (2001); Jasa and Xiang, ibid. 117, 3707(A) (2005)] has been applied to estimating both the decay times and decay order required in sound energy decay analysis. The need to focus on model selection in a Bayesian model formulation of sound energy decay analysis was presented by Jasa and Xiang [J. Acoust. Soc. Am. 119, 3343(A) (2006)] along with a discussion of the limitations of the existing Monte Carlo algorithms used for this purpose. This paper will present some recent algorithms that can overcome the limitations of a Monte Carlo approach to model selection and show how these algorithms can be applied to determining both the proper model and decay time estimates for acoustically coupled rooms.