Automatic detection of odontocete whistles yields numerous useful applications in the bioacoustical study of these marine mammals. Silbido is a software system capable of performing such automatic whistle detection and extraction. An integral component of silbido is its spectrum analyzer, which is responsible for the spectral analysis of audio signals containing these whistles. Silbido uses a conventional spectrum analyzer which is characterized by a filter bank whose filters exhibit equally spaced center frequencies and equal bandwidths. In contrast, constant-Q spectrum analyzers use filter banks whose filters exhibit bandwidths that are proportional to their center frequencies. An implementation of a discrete Fourier transform (DFT) based constant-Q analyzer is presented. The input signal is preprocessed with a number of half-band and band-pass filters, as well as 2-to-1 and 4-to-1 downsampling stages. Such preprocessing enables variability in time-frequency resolution, reduces workload, and prepares each frequency octave for DFT processing. A sliding frequency window of variable bandwidth applied to the DFT spectrum allows estimation of arbitrary center frequencies. Spectral estimations with center frequencies and bandwidths coincident with the center frequencies and bandwidths of a desired proportional filter bank are then taken. This thesis replaces the conventional analyzer in silbido with this constant-Q analyzer and evaluates its performance on a corpus containing whistles from various odontocete species. Results show an unfavorable loss in silbido’s recall ability once the constant-Q analyzer is integrated.