The evaluation of singing voice accuracy: How tolerant are we?

The evaluation of the singing voice accuracy is partly based on the precision of the intervals between the tones of a sung performance. This study aims to observe the listeners’ tolerance when judging melodies in order to properly evaluate the singing voice accuracy in a melodic context
For this purpose, an interval contained in familiar and unfamiliar sung performances was manipulated in two directions (compression and enlargement from 10 to 60 cents). This material was presented through the method of limits procedure to 30 non-musicians. They were asked to define whether the melodies proposed were “in tune” or “out of tune” in a test and a retest. The accuracy thresholds for compression and enlargement errors as well as the tolerance zone were computed.
The results showed that participants were consistent in their evaluation between the test and the retest. The tolerance zone (M = 42 cents, ES = 1.60) was not affected by the familiarity of the melody. Concerning the accuracy thresholds, participants were more tolerant when the interval was compressed (M = 25.68 cents, ES = .1.34) than when it was enlarged (M = 16.32 cents, ES = .91), for the familiar (t(30) = 4.94, p < .001) and non familiar melody (t(30) = 3.27, p = .003).
This study highlights the ability of non-musicians to evaluate familiar and unfamiliar melodies with a similar tolerance. Furthermore, the error direction is relevant in perceptive rating and should be taken into account when defining a sung performance as “in tune” or “out of tune”.