Matthew Sneddon
BBN Syst. and Technol., 21120 Van Owen St., Canoga Park, CA 91304
A set of algorithms for real time classification of noise events as low altitude military aircraft overflights was developed for use in a personal noise monitoring system. The classification algorithms, which are applied to polynomial curve fits to time series of sound pressure level measurements rather than to raw data, operate primarily on the temporal symmetry and relative rates of increase and decrease of sound levels in different one-third octave bands. Correct classification performance and misclassification errors vary with disjunctive decision rules applied to the outputs of the algorithms. [Work supported in part by the U.S. Air Force NSBIT program.]