Optimum reverberation time for speech intelligibility in classrooms

Abstract

This work addresses the topical question of the optimum reverberation time (RT) for speech intelligibility in classrooms. Existing literature is reviewed and is found to be inconsistent. Experimental methods generally predict zero optimum RT, while theoretical methods predict nonzero values. These differences are discussed; they appear to stem from different treatments of the background noise. Assuming diffuse-field theory, a theoretical model to predict optimum RT was developed from a previously proposed model [S. R. Bistafa and J. S. Bradley, J. Acoust. Soc. Am. 107, 861–875 (2000)]. It incorporates noise in a physically realistic way and predicts nonzero optimum RTs. Subsequently, to reduce reliance on diffuse-field theory, the model was modified using empirical methods to predict speech level and reverberation times in classrooms [M. R. Hodgson, J. Build. Acoust. (in press, 2001)]. Again, nonzero optimum RTs were predicted. The prediction models and results are presented, practical conclusions are drawn, and suggestions for further work are made.