Klasifikasi Chord Musik Menggunakan Gabungan Fitur Domain Waktu, Frekuensi, dan MFCC

Abstract

This study presents a chord classification system that combines audio features from three different domains: time, frequency, and Mel-Frequency Cepstral Coefficients (MFCC). The purpose is to improve the accuracy of identifying musical chords from audio signals, which often contain overlapping sounds and instrument variations. The dataset used consists of major and minor chord audio clips sourced from Kaggle. Each audio file undergoes preprocessing, including resampling and signal normalization, followed by feature extraction from the three domains. The extracted features are then merged into a single vector and classified using the Random Forest algorithm. The model is evaluated using accuracy, precision, recall, F1-score, and confusion matrix. Results show that the model performs well in detecting major chords (F1-score 0.83), but has lower recall for minor chords (F1-score 0.68). The overall accuracy is 77%, indicating that combining features from multiple domains enhances classification performance. This method shows potential for future development in audio signal analysis and music recognition systems.