Prediction of subjective annoyance penalty of impulsive noise based on objective measurements

• Impulsive components in noise cause higher annoyance than steady-state noise having the same A-weighted sound level. • We investigated several alternative models to predict the annoyance penalty of impulsive sound. • The models were validated against psychoacoustic experimental data, which covered a large range of impulsive sounds. • The best prediction model was based on sound level variability, which is available in most sound level meters. • The best model outperformed the model of ISO PAS 1996-3, which is based on onset rate and level difference. Many regulations and standards penalize noise with an impulsive character by adding a constant correction to the measured equivalent A-weighted sound level, L Aeq . This is expected to correspond better with perceived annoyance. Annoyance with impulsive noise may not correlate with such a simple procedure. The ISO/PAS 1996-3 describes a method for the measurement of onset rate and level difference of impulses and offers an objective procedure to calculate the penalty. Purpose of our study is to compare the results of this and four other procedures with experimental psychoacoustics data. Based on these comparisons, a procedure that uses the difference between the 1 and 99 percentile levels is recommended for assessing outdoor impulsive sounds in the range 45–65 dB L Aeq because of its simplicity.