Pengenalan Pola Sinyal Suara Manusia Menggunakan Metode Back Propagation Neural Network

  • Faradiba Faradiba Dosen Pendidikan Fisika, Fakultas Keguruan dan Ilmu Pendidikan, Universitas Kristen Indonesia

Abstract

This research have been designed a structure of artificial neural network (ANN) with using backpropagation to recognize signal pattern of human voice. The signal should first be processed with Linear Predictive Coding (LPC). Linear Predictive Coding (LPC) used for extraction characteristic. Producing matrix with 24 x50 orde which is then becoming input data for Artificial Neural Network Backpropagation (ANN-BP). There nets consist of 4 layers. Those 4 layers are : 1 input layers with 24 neuron, 2 hidden layers which are devided as 15 neuron at the first hidden layer and 10 neuron at the second, The last output with 5 neuron.For 5 training data, parameter characteristic value of net such us: Learning rate value (alpha) = 0,05 and mu value (μ) = 10-3with using by sigmoid bipolar activation function. The result of the research shows that the nets can recognize as 100 % of 25 trainning data, 74 % of 25 testing data.
Keywords : Artificial Neural Network, Back propagation, voice recognation

Published
Jul 5, 2017
How to Cite
FARADIBA, Faradiba. Pengenalan Pola Sinyal Suara Manusia Menggunakan Metode Back Propagation Neural Network. EduMatSains, [S.l.], v. 2, n. 1, p. 1-16, july 2017. ISSN 2527-7235. Available at: <http://ejournal.uki.ac.id/index.php/edumatsains/article/view/372>. Date accessed: 24 sep. 2017.
Section
Articles