Comparison of models to predict the effect of background speech on work performance in open-plan offices
Proceedings of the ICA congress
There are different models to estimate the relationships between measurable acoustical parameters and office workers' perception, well-being and performance. These models are based on speech transmission index (STI), fluctuation strength (FS) or percentile level statistics. Whilst STI requires a loudspeaker and is measured in unoccupied conditions, other metrics, such as FS or percentile level statistics, can be determined in situ during usual working hours. Nonetheless, the established models
... established models have some shortcomings. STI can estimate the effect of one simulated speech source on short-term memory performance but it cannot assess the effect of office noise exposure in occupied offices. FS correlates with the impact of temporal-spectral variability of a background sound on short-term memory performance but it depends on the sound pressure level which does not have an effect on the short-term memory performance. Percentile level statistics correlate with the speech-to-noise ratio: higher differences between the 10 th and 90 th percentile levels measured with fast time weighting lead to lower number recall performances. As part of this study 110 sound conditions under which subjects have to complete a number recall task are evaluated with respect to their relationships with these three acoustical parameters. A cross-validation reveals comparable prediction qualities of the models.