Sape is a popular traditional musical instrument from Sarawak, Malaysia. This instrument is handmade and differs in wood, dimension, design, etc. This impedes the preservation and promotion of the Sape tradition, as well as the development of new Sape instruments. Wood is the primary factor affecting Sape sound, and this study aimed to evaluate the quality of three commonly used types of wood in the instrument manufacturing. Rectangular wood samples were fashioned and tested for physical, vibroacoustic, and timbre characteristics using a flexural vibration test. The objective was to identify the top features influencing the sound quality and develop a reliable method for classifying wood quality. Results showed that Adau wood had the highest quality, followed by Merbau and Tapang wood. Four key features, including acoustic radiation damping coefficient, spectral flux, spectral centroid, and inharmonicity, were selected by the MRMR algorithm and used to train and test various classifiers in MATLAB. The decision tree classifier achieved 98.1% accuracy in predicting wood quality. This study demonstrates the potential of using machine learning to classify Sape wood quality and provides a useful guide for its production. The findings could contribute to the preservation and advancement of this cultural tradition.
Read full abstract