Deteksi Onset pada Beberapa Genre Musik Menggunakan Metode Deviasi Fasa

Amsaro, Enny (2017) Deteksi Onset pada Beberapa Genre Musik Menggunakan Metode Deviasi Fasa. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
2913100006-Undergraduate_Theses.pdf - Published Version
Restricted to Repository staff only

Download (195MB) | Request a copy

Abstract

Berbagai metode telah digunakan dalam mendeteksi onset sinyal musik, namun sebagian besar hanya terfokus pada besaran spektrum (magnitude) dan mengabaikan spektrum fasa dari sebuah sinyal. Hal ini menjadi sebuah perhatian dikarenakan proses pengkodean struktur tempo sinyal musik justru terjadi pada spektrum fasa. Deviasi fasa merupakan metode yang memanfaatkan informasi dari spektrum fasa berdasarkan nilai turunan kedua fasa sinyal musik. Fasa tersebut didapatkan dari hasil transformasi ke domain waktu-frekuensi menggunakan algoritma Short-time Fourier Transform (STFT) secara overlap. Setelah mendapatkan fungsi deteksi onset, dilakukan proses pemilihan puncak (peak-picking) dengan menerapkan adaptive threshold. Performa sistem deteksi onset diukur menggunakan f-measure dan diujikan terhadap genre musik jazz, reggae, disko, metal, dan klasik. Pada genre jazz, reggae, dan klasik nilai f-measure tertinggi diperoleh pada lebar window STFT 4096 sampel, yaitu sebesar 73.10%, 82.58%, dan 79.44% sedangkan pada genre metal dan disko nilai f-measure tertinggi diperoleh pada lebar window STFT 2048 sampel, yaitu sebesar 80.93% dan 71.68%. ========================================================= There are various methods that have been used in music signal onset detection, but most of them are only focused on the magnitude of the spectrum as the only information source. Mostly, the phase spectrum of a signal is ignored and this becomes a concern because much of the temporal structure of a signal is encoded in the phase spectrum. Phase deviation is a method that uses information from the phase spectrum according to its second difference. This phase spectrum is obtained from the signal transformation into the time-frequency domain by using overlapping Short-time Fourier Transform (STFT). After getting the onset detection functions, the peak selection process (peak-picking) is conducted by applying an adaptive threshold. The performance of onset detection system represented by f-measure and be tested against several music genres such as jazz, disko, reggae, metal, and classical. In the genre of jazz, reggae, and classic, the highest f-m-measure values obtained from STFT using window length of 4096 samples, i.e. 73,10%, 82,58% and 79,44% whereas in the genre of metal and disko, the highest f-measure values obtained from STFT using window length of 2048 samples, i.e. 80.93% and 71.68%.

Item Type: Thesis (Undergraduate)
Additional Information: RSKom 621.382 2 Ams d
Uncontrolled Keywords: Deteksi Onset, Short Time Fourier Transform, Deviasi Fasa, Adaptive Threshold, Peak-picking.
Subjects: M Music and Books on Music > M Music
Q Science > QA Mathematics > QA404 Fourier series
Q Science > QA Mathematics > QA76 Computer software
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms
Divisions: Faculty of Electrical Technology > Computer Engineering > (S1) Computer Engineering
Depositing User: Enny Amsaro .
Date Deposited: 08 Feb 2018 03:14
Last Modified: 08 Feb 2018 03:35
URI: http://repository.its.ac.id/id/eprint/46839

Actions (login required)

View Item View Item