Recognition of Piano Pedalling Techniques Using Gesture Data

This paper presents a study of piano pedalling technique recognition on the sustain pedal utilising gesture data that is collected using a novel measurement system. The recognition is comprised of two separate tasks: onset/offset detection and classification. The onset and offset time of each pedalling technique was computed through signal processing algorithms. Based on features extracted from every segment when the pedal is pressed, the task of classifying the segments by pedalling technique was undertaken using machine learning methods. We exploited and compared a Support Vector Machine (SVM) and a hidden Markov model (HMM) for classification. Recognition results can be represented by customised pedalling notations and visualised in a score following system.