Samenvatting

Music moves us, literally. We tend to move the body in synchrony with the beat. Individuals without any professional music training are capable of singing or humming along with an unfamiliar melody or indicating the melodic contour by means of hand gestures. Musicians who play by ear are able to do the same on an instrument. In this study an attempt was made to quantify the extent to which professional keyboard performers were able to play by ear, and whether improvising musicians were superior to non-improvising. During the experiment, subjects were asked to listen to short, unfamiliar music excerpts recorded on a MIDI controller. Subjects were asked either to play along, replicate the excerpt, transpose it to a different key, or to harmonize it. Subjects were recruited from two groups of classically-trained musicians: improvising and non-improvising pianists and church organists. The bass and treble parts extracted from each MIDI sequence were compared with the bass and treble from the aural model, yielding an alignment score for each task. The comparison was performed using content-based music retrieval software developed in the WITCHCRAFT project for the study of folksong melodies. Results showed that the top voice was replicated better than the bass. There were large differences between the musicians. As a group, improvising musicians scored better than non-improvising musicians, however this difference was not significant. Mixture analysis showed that top-scorers came from both groups. Subjects with perfect pitch did not perform better.