When performing F0 stylizations and parametrizations in order to establish the essential parts of an F0 contour, it is necessary to assess the similarity between original and stylized contours. The most reliable method is a perceptual assessment, which means in its most severe form that no differences are perceptible between original and resynthesized versions. Such tests have a number of possible influences (resynthesis quality, auditory storage, whether or not perceptible diffences are included), and they are tedious and time-consuming, as will be exemplified by results obtained in the evaluation of this maximum-based description. Computational comparisons, on the other hand, are unable to mirror human perception, because it is well known that small differences in the acoustic domain may be perceived while larger differences may pass unnoticed. The mean difference, as measured by some researchers, is problematic, because the goal of stylization is to abstract away from unimportant features. A successful stylization is perceptually equal to the original but has a large mean difference caused by successful abstraction. The current limited knowledge of pitch perception and interpretation requires perceptual tests to evaluate stylization quality.