Efficiency of sound power level determination by engineering method using bootstrap algorithm and Bayesian inference

Sound power level is one of the basic parameters describing a sound source. It is used to model the distribution of sound levels in the environment, to assess noise exposure in workplaces and to compare machines and devices of a given type. Therefore, knowing the exact value of this parameter is very important. This paper investigates the usefulness and effectiveness of using non-parametric statistical methods to determine the sound power level of sound sources using the engineering method. This was done using a bootstrap method and Bayesian inference. Inference was based on non-parametric statistical tests at a significance level of a = 0.05. The tests were performed on actual measurement data recorded for a B&K 4205 reference sound source. The statistical analysis performed showed that the minimum measurement sample size for determining the sound power level using the engineering method should be 4 elements for Bayesian inference and 6 elements for the bootstrap method.