Fig 1. Relationships between judges’ ratings and musical criteria.

Boxplots of Spearman correlations between either pitch interval deviation (top), number of contour errors (middle) or number of modulations (bottom), and the average scores of expert judges (left), non-expert judges at the test (middle) and non-expert judges at the retest (right). Each boxplot refers to a particular number of judges and all possible combinations of judges are considered. Horizontal lines indicate the threshold for significant correlation coefficients (based on samples of 166 performances at significance level 5%).