Sound spectrograms have been a traditional method for comparing animal vocalizations and demonstrating their similarities or differences, but quantification of the comparative process has been difficult. This lack of quantification has hindered progress in documenting the timing and degree of song development in song birds. With this in mind, computer-based methods for sound comparison and sound averaging were developed. These methods utilize a numerical representation of a sound's frequency-time structure to compute a similarity value between any two sounds and an average sound spectrogram for a set of sounds. These analytical techniques were applied to an analysis of song notes of the swamp sparrow (Melospiza georgiana). The results demonstrate the usefulness of the methods and reveal new details in the description of the swamp sparrow's set of species-universal note types and in the decrease of note variability during the course of developmental transitions from subsong, through plastic song to song crystallization.