The effect of occupancy in the speech intelligibility in churches

Abstract

Measurements were done in four churches with and without occupancy (mean volume of 6314 m3 and mean occupancy of 155 persons) to test the effect of occupancy in the STI values. The results show that occupancy induces a mean increase for the speech intelligibility of DSTI (STIoccup - STIunoccup) of 0.05 with the use of a public address system (PA) and 0.027 without a PA. This increase is caused mainly by the reduction of the room reverberation time (RT), due to the additional sound absorption by the presence of persons. The reduction in the RT values in a church depends mainly on its volume and on the RT of the unoccupied room as well as on the number and dispersion of the persons in it. The mean sound absorption by person (Apers) calculated for the average RT calculated for each church, equals to 0.57 ± 0.09 m2. The decrease in the S/N ratio (due to the background sound level augmentation by the presence of people and to the sound level decrease with the distance in the congregation area) could reduce the STI gain in particular cases. The variations of humidity and temperature by heating in churches induce a weak variation of the RT values (less than 0.2 s). These effects are generally negligible for the STI values. The variations in the STI values with occupancy can be predicted in churches (mean error DSTI < 0.005) by empirical formulations based on RT in occupied and unoccupied churches: DSTI = 0.276*ln(RTunocc (2kHz) / RToccup (2kHz)) - 0.052 with a PA (R2 = 0.96), and DSTI = 0.045*ln(RTunocc (2kHz) / RToccup (2kHz)) + 0.012 without a PA (R2 = 0.34).