Abstract

Many open-plan offices adopt an electronic sound masking system in order to reduce distraction from background noises, primarily intruding speech. Sound masking systems should uniformly generate a masking sound over the entire office area to homogenize the speech privacy whilst minimising occupant perception of the masking sound. This study evaluates the spatial uniformity of the masking sound field in an example open-plan office, where the masking system was set up to represent supposedly optimal installation conditions; 1-speaker zones, individually-calibration of each zone to match the specified curve precisely, and smaller zones than typically specified. Sound level measurements performed as per ASTM E1573-18 were made at each workstation, as well as every 0.6 m across the office, for a total of 117 measurements. Measurement results show that tolerance of ± 0.5 dB for the overall A-weighted level is only achievable at 61% of measurement locations, whilst ± 1 dB is achievable at 99% of locations. For one-third-octave band sound pressure levels between 250 Hz to 4 kHz, ± 2 dB is achieved only 55% of the time, and tolerance of ± 3.5 dB is required to achieve 95% compliance with the specified curve. By using calibrated computer simulations, the study also examined parameters that can influence spatial uniformity in open-plan offices. It was found that the number of sound masking loudspeakers, partition height, and the scattering and absorption coefficient of the ceiling all affect the uniformity of the masking sound. Speech intelligibility was assessed by calculating the Articulation Index (AI) to determine an acceptable tolerance for masking sound variation. Increasing the number of loudspeakers was the most effective way to improve the uniformity of the masking sound. The AI results suggest ±2 dB, when including octave band sound pressure levels, is a minimum required tolerance for a sound masking sound field in an open office to provide AI values within ±0.1 of the targeted value across the office area.