Parameter values for a signal processing methodology with constant maximum sample kurtosis across fractional-octave-bands

摘要

Noise control engineers may be asked to select a power tool or machine that will complete a defined task with the minimum cumulative noise exposure. For impulsive noise, the time weighted average sound pressure level yields an inadequate description of the cumulative noise exposure; furthermore, there is not a generally accepted metric for rating the hazard of impulsive noise. A signal processing methodology is needed that considers more of the information contained in the sound pressure waveform and makes it easily interpreted. This paper does not define a metric for impulsive noise; however, it describes the parameter values for a signal processing methodology that considers the kurtosis of the sound pressure. By making each time-frequency wavelet frame have equal area and have a constant number of samples, the kurtosis has a systematic scaling. This signal processing methodology is described and results are presented. Simulations show that the mean, sound pressure level, skewness, and kurtosis level are sufficient statistics for describing the first ten moments of the sound pressure in each time-frequency frame. This signal processing methodology is a necessary step toward the future goal of developing a metric for rating the risk of hearing impairment due to impulsive noise exposure.