Jim Pitton         Principal Engineer         EIS Department         Applied Physics Laboratory         University of Washington	Together with the Naval Research Laboratory (NRL), DC, we are evaluating the ability of trained sonar operators to discriminate targets from clutter using appropriate datasets for mid-frequency active and impulsive source sonar systems. We train and evaluate naïve listeners in target/clutter discrimination, and compare their performance to that of trained operators. Auditory experiments can be very time consuming. Since it may not be possible to engage sonar operators in such time-consuming studies, we are thus training naïve listeners in echo classification for subsequent experiments. The performance of this group is compared to that of the sonar operators to ensure the validity of subsequent experiments while minimizing the impact on the operators' schedules. We are analyzing the active sonar data to identify hypothesized signal features for both machine classification and human listening experiments. Listening experiments are conducted to identify perceptually important features for aural classification. We will design and conduct auditory experiments using trained listeners to test the perceptual importance of various echo features for aural classification. By having listeners judge the similarity or dissimilarity of the various test signals to target and clutter echoes, we can identify those signal dimensions (i.e., features) that distinguish targets from clutter. We then design an automatic classifier using perceptually important features identified in the above studies, as well as standard features currently used in automatic classification. We quantify the performance of individual and combined features to determine the best scheme for echo classification. The new features are compared against standard features for target classification of active source systems.