Dengan Klasifikasi Nyamuk Berdasarkan Suaranya Metode Mel Frequency Cepstral Coefficients dan Jaringan Syaraf Tiruan

• Main Author: LUKMAN, Achmad
• Format: Theses and Dissertations NonPeerReviewed
• Published: [Yogyakarta] : Universitas Gadjah Mada 2013
• Subjects: Electronic and Instrumentation System
• Online Access: https://repository.ugm.ac.id/120602/; http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=60640
• Institution: Universitas Gadjah Mada

Digital signal processing are the most popular and sizable positive impact is digital sound field processing. Digital sound processing can be developed with a variety of applications that can facilitate human life, one of the research that can be made is the introduction of mosquito noise. Mel frequency is the process of getting the value of logarithmic energy in a filter. In his book, (Huang et al, 2005) states that the main parameters of the feature extraction Mel Frequency Cepstrum Coefficients (MFCC) is the number of filters that will be used. MFCC feature extraction is actually an adaptation of the human auditory system where the signal will sound in a linear filter for low frequency (<1000 Hz) and is logarithmic for high frequency (> 1000 Hz). This study used MFCC method for extracting features of the voice signal in the form of a mosquito that cepstrum coefficients contained therein are signal frequency mosquito noise that can be classified by artificial neural networks. The research is continued by making new software that can be used to support experimental classification mosquito noise signal with implemented signal processing method. As a comparison there are two classification method in implemented on the software, i.e. backpropagation artificial neural network method and learning vector quantization artificial neural network method. The software is used to classify three types of female aedes aegypti mosquitoes, the anopheles mosquitoes and culex pipiens mosquitoes. The test is performed with the variation of the number of filters in scenarios MFCC method at 13, 20, 25, 30, and 35 with number of training data of each 50 mosquitoes and number of testing data of each 40 mosquitoes. Furthermore, the accuracy result of backpropagation neural network will be compared to the accuracy result of learning vector quantization neural network. So that can give us brief description about the high accuracy of the proposed classification method.