AMMRI: a computational assessment tool for music novices’ replication and improvisation tasks

Abstract

We present computational analyses of musical performances during 12-months study by novice participants aged 65–80. They learned two instruments (an electronic piano keyboard; the iPad app ThumbJam) each with two distinct approaches: replication by ear of melodies, and improvisation using specified methods. Here we present computational simulations and analyses of such processes and the corresponding R script. Using MIDI recordings from one participant group, we reveal diverse performance levels. Our tools are apt to analyse of our full dataset and potentially other assessments of early musical learning. The code can readily be developed for more advanced learners.