Bayesian inference in architectural acoustics: Estimation of decay times in coupled spaces

Sound energy in coupled spaces decays at multiple rates under certain conditions. Identifying decay times in these coupled spaces often demands considerable effort. Generally, different portions of multi-rate decay functions obtained from measured data cannot always be distinctly recognized. A model-based parameter estimation approach within the Bayesian framework facilitates the evaluation of multiple decay times through an extension of the decay model established in a previous work [N. Xiang, J. Acoust. Soc. Am. 98, 2112–2121 (1995)]. Unlike gradient-based approaches, no careful estimation of initial values is required. Therefore, a robust algorithmic estimation of multiple decay times from experimentally measured decay functions shows advantages over the existing nonlinear regression approach.