Designing an Intelligent Learning System For Practicing the Oboe Embouchure

Abstract

The oboe is considered one of the most difficult woodwind instruments to learn due to the labor involved in mastering the embouchure required to produce a steady, long tone through the instrument's reeds. Previous research in music education has suggested that using visualization or imagery as a training method, so-called audiation, helps novice learners gain a better understanding of the relationship between sound and embouchure. Inspired by this concept, we designed an interactive learning system using real-time acoustic analysis (MFCCs) and support vector machines (SVM). Our interactive visualization aims to support the user's learning process by rendering the quality of good and bad sounds as pitch indications through particle swarm animations. A pilot user study showed that this makes the learning experience more reflective and self-directed, as the learner can understand the relationship between their breathing technique and the visualization of the generated sound quality. In this paper, we intend to address issues potentially involved in designing an intelligent learning system using sensing technologies in the context of music education.