Objective Comparisons of Song Syllables: A Dynamic Programming Approach

It is difficult to compare the sounds made by birds objectively. Visual comparisons are open to question on the basis of interobserver variability. Previous computational methods based on sonagrams have not solved the problem that continuous variation occurs on both the time and frequency axes at the same time. A Dynamic Programming Algorithm is presented and its use demonstrated with elements digitized from chaffinch (Fringilla coelebs) song. Statistical analysis of the distances calculated suggest that consistent discrimination between the sounds compared is possible.