In the actual sound environment system, the observed data are inevitably contaminated by the background noise of arbitrary distribution type. Furthermore, the observations often contain fuzziness due to several causes and exhibit level saturation owing to the existence of a finite dynamic range. In this study, a signal processing method for estimating a specific signal based on fuzzy observations with the existence of background noise of non-Gaussian distribution forms is proposed in a form appropriate for the finite level range of the measured data. More specifically, after introducing a Jacobi polynomial defined within a finite fluctuation range of sound level, by considering the probability measure of fuzzy events, a method to estimate the fluctuation wave form of a specific signal under existence of background noise is theoretically proposed. The effectiveness of the proposed theory is confirmed by applying it to actual traffic noise data.
View full abstract