Sound quality classification of wood used for Sarawak traditional musical instrument- Sape / Wong Tee Hao
Sape, a traditional musical instrument in Malaysia, is meticulously handcrafted through a complex process. Each Sape, crafted by various makers, differs in size, materials, and design, leading to variations in their quality. Despite individual methods employed by Sape makers to assess quality during...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2024
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/15397/2/Wong_Tee_Hao.pdf http://studentsrepo.um.edu.my/15397/1/Wong_Tee_Hao.pdf http://studentsrepo.um.edu.my/15397/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Sape, a traditional musical instrument in Malaysia, is meticulously handcrafted through a complex process. Each Sape, crafted by various makers, differs in size, materials, and design, leading to variations in their quality. Despite individual methods employed by Sape makers to assess quality during production, a standardized guideline for quality inspection remains absent. This research aims to delineate the primary factors influencing Sape quality, employing both qualitative and quantitative methodologies. Initial stages involved gathering insights from seasoned Sape makers and players through questionnaires and focus group discussions, revealing material as the foremost quality determinant in the Sape. Subsequently, the focus shifted to investigating common woods used in Sape construction, specifically Adau, Tapang, and Merbau, representing light, medium, and heavy hardwood categories, respectively. Rectangular wood samples simulating Sape soundboards were created, and sound data was recorded through flexural vibration tests. Expert evaluations of the sound quality were conducted via listening tests. Utilizing MATLAB's MIRToolbox, 18 acoustic properties were extracted from the wood samples. Statistical analyses were employed to identify the most reliable quality ratings. To address dataset imbalances, Synthetic Minority Oversampling Technique was used, enhancing dataset quality before training 40 machine learning classification algorithms. Among these, the Gaussian-kernel Support Vector Machine stood out, achieving remarkable performance with 88.18% validation and 93.37% test accuracies. This model was employed to build a MATLAB-based Sape sound quality classifier. Utilizing the Shapley Additive Explanations interpretation method, the analysis emphasized the importance of selected features in predicting wood acoustic quality, highlighting "Spectral Roll-off 85%" as the most crucial predictor of sound quality. Finally, a user-friendly Graphical User Interface was developed to aid Sape makers in assessing wood quality objectively, enhancing the process of selecting high-quality Sape instruments.
|
|---|