Sistem Pengecaman Ucapan Huruf Jawi Menggunakan Rangkaian Neural : Perbandingan Penggunaan Fungsi Pengaktifan Sigmoid Dan Hiperbolik Tangen

NIK MOHAMAD, NIK HAJARATUL ASHIKIN (2003) Sistem Pengecaman Ucapan Huruf Jawi Menggunakan Rangkaian Neural : Perbandingan Penggunaan Fungsi Pengaktifan Sigmoid Dan Hiperbolik Tangen. Other thesis, Universiti Teknologi Malaysia.

[img]
Preview
PDF
nik_hajaratul_ashikin_.PDF

Download (3466Kb) | Preview

Abstract

Speech recognition has been one of the most research-applied fields in this new era.This is simply because technology are developing rapidly.Speech recognition is an applied field of artificial intelligence which has focused by many researchers to develop a new technology Speech recognition,in general will help researcher engineers, computer scientist,and others to solved all the difficult problems. The objective of this project is to study speech recognition of Jawi words from 'Alif until 'Fa' using neural network. This study is focused on the used of two activation functions that are sigmoid activation function and hyperbolic tangent activation function in neural network. This study is to identify which activation function is better and how neural network can be used in Jawi speech recognition.To solve this problem, neural network are tested using try and error method on the speech recognition system that has been build. This study is done by three people who are using the same data from the same source. This study find that neural network can be a good Jawi speech recognition system. This study also finds that sigmoid activation function is better than hyperbolic tangent activation function. Results for the whole recognition proved that sigmoid activation function is better with 60.8 % recognition and 40% recognition for hyperbolic tangent activation function for 'Alif until'Jim'.

Item Type: Thesis (Other)
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Computer Science and Information System > Computer Graphics and Multimedia
Depositing User: Unnamed user with email knizam@utm.my
Date Deposited: 04 Jul 2013 07:09
Last Modified: 04 Jul 2013 07:09
URI: http://ir.fsksm.utm.my/id/eprint/1454

Actions (login required)

View Item View Item