Identifikasi Alat Musik Tradisional Nusa Tenggara Timur (NTT) Menggunakan Metode Mel Frequency Cepstral Coefficients (MFCC) dan K-Nearest Neighbor (KNN)

Abstract

Traditional musical instruments such as gongs and drums from East Nusa Tenggara (NTT) are an important part of the local cultural heritage, but their existence is increasingly threatened due to lack of documentation and difficulties in the sound tuning process. This study presents a digital-based approach to classify the accuracy of the sound of these musical instruments by utilizing the Mel Frequency Cepstral Coefficients (MFCC) feature extraction technique and the K-Nearest Neighbor (KNN) classification algorithm. A total of 238 sound data were recorded and processed to remove noise. The results of the feature extraction were then grouped into three categories: "not right", "almost right", and "already right". The system managed to achieve an accuracy of 92.93% with the best parameter configuration. This solution is expected to help craftsmen in the sound tuning process and contribute to the preservation of regional culture.