A review of software for note onset detection
Nicholas J. Bailey, Keziah Milligan · ENLIGHTEN (Jurnal Bimbingan dan Konseling Islam) · 2015
Performance study is impossible without reliable data acquisition. Manual note segmentation, an important prerequisite for performance analysis, is both time inefficient and unreliable, so an automatic method would increase both the amount of data that can be analysed and the accuracy of the results. There exist various pieces of software purporting to detect note onsets, which will be reviewed and compared here. A variety of test audio is used — both twelve EDO and microtonal, with a focus on instruments with non-percussive onsets, like voice or trumpet — to determine the strengths and limitations of each detection method. A new method is proposed, which seeks to mimic the way the ear and brain perceive music, by modelling hair cells and the basilar membrane in the ear and employing backtracking.