On the Noise of Crowds

Abstract

The behavior of crowds of people as acoustical-noise sources is a matter concerning which there is little definitive information, in spite of an increasing need for such knowledge. Although quantitative data are available for individual talkers and listeners, analysis of an aggregate of individuals—i.e., a crowd—is complicated by the fact that the total acoustic output is influenced by the effect upon communication of the resultant noise level. A particular problem, discussed in several papers a few years ago under the heading “cocktail-party acoustics,” is the spontaneous increase of crowd noise to a rather high level under certain conditions. A simple model is given that attempts to define these conditions. This model assumes that (1) both direct and reverberant sound must be considered, although either may be dominant in a given case, and (2) if, in a specific situation, the noise level due to normal conversation is such as to inhibit communication, the level will increase to a certain maximum value, at which point talkers are exerting maximum speech effort short of shouting. Examples analyzed according to the model are given.