Modeling the effects of air currents on acoustic measurements in large spaces

Abstract

Listen

Translate article

Historically reverberation times were accurately measured with an explosive or an interrupted tone or noise. But in theory using convolution methods for acoustic measurement requires that an impulse response be precisely time stationary. Air in large spaces is never stationary, and the further a sound wave travels—as it must in a reverberant tail—the more it will be affected by moving air. Long stimuli and synchronous averaging increase the time over which the air must be still. We propose that air currents can be modeled through the fluctuations they create in the time base of the received stimulus. We find that noise-based convolution methods such as MLS are particularly sensitive to this problem, as they must use long stimuli to achieve an adequate signal-to-noise ratio (S/N). For these methods, when there is no net flow, the principle degradation is a rapid reduction of S/N at high frequencies. If there is a net flow, the high frequency reverberation time can also be significantly shortened. Logarithmic sine stimuli are shorter than MLS for an equivalent S/N, and are less sensitive to air currents. The relevance of these effects in actual spaces is under investigation.