Converting Sound To MIDI Using Short Time Fourier Transform As A Virtual Midi Controller On Digital Audio Workstation

Abstract

Music is inseparable from everyday life, and its demand continues to increase with technological advancements. Amateur music producers are increasingly using digital equipment such as MIDI controllers to create music independently. However, MIDI controllers generally come with a high price tag, and not all musicians have the ability to play the piano or keyboard.

This research develops a method for converting sound into MIDI format using the Short-Time Fourier Transform (STFT) and analyzes the accuracy of pitch detection and the reliability of the generated information. Audio recordings from various musical instruments and human voices are used as input in a web-based system. The STFT method is applied to the audio signal to identify and convert pitches into MIDI notations.

The analysis results show that the STFT method is capable of producing adequate accuracy in pitch detection, reaching 24.216621%.Factors such as audio quality, STFT parameters, and threshold settings significantly influence the conversion results. Using good audio quality, selecting appropriate STFT parameters, and optimizing the threshold settings can improve pitch detection accuracy.