Sistem Pengenalan Chord Musik Otomatis dengan Algoritma Hidden Markov Model dan Pitch Contour Extraction

Abstract

Music is a form of art that is created through a series of sounds that are put together so as to create a tonal accompaniment that can be understood by humans. Music is an art that involves the arrangement of sounds related to tone, rhythm, and timbre. At present, music has experienced a fairly rapid development both in terms of genre and structure of the music itself. technological developments, especially in the field of computers, are increasingly rapid. Many new technologies are emerging at this time. Artificial intelligence is a modern technology that focuses on creating systems that react like humans. This research utilizes the model of Hidden Markov Model (HMM) and Feature Extraction of Chromagram audio obtained using Short Time Fourier Transfer (STFT) for Chord detection. Chord that can be classified consists of 24 chords with a division of 12 Major Chord and 12 Minor Chord. The dataset used is the result of audio recorded by the author which contains 474 collections of musical instrument recordings with various recording styles using a piano, bass and guitar. This research obtained an accuracy of 79%, precision 85%, recall 78%, and F1-score 80% of all Chord classes.