This work shows the results of research to classify odontocetes from their echolocation clicks. Current classifiers perform less effectively when presented with clicks from an unknown source. This study examines two bioacoustics algorithms with respect to rejecting data from an unknown species. It presents a comparison of the advantages and drawbacks of the two algorithms: Gaussian mixture models (GMMs) and universal background model (UBM). The algorithms are utilized to form a species detection experiment to evaluate the performance testing against unknown species. Additionally, a hybrid classifier is developed which combines GMM and UBM to form a third species detection experiment to overcome the drawbacks from the two algorithms. Furthermore, this study focuses on comparing score normalization techniques. Score normalization is a method used to account for variability of score distributions. Normal distributions are estimated from the sample statistics generated by either varying the models against which a set of echolocation clicks are classified (t-norm) or from scores produced by an alternative set of echolocation clicks classified by a single model (z-norm). In either case, the statistics are used to normalize test scores to a z-score that is used in the final classification decision. These normalizations along with tz-norm which combines the two are examined for the aforementioned classifiers. Multiple experiments are prepared for a species detection scenario to compare the variations of the three classifiers and the different types of normalization techniques.