Web-based Kadazandusun speech recognition system

Arrvierri Vespha (2022) Web-based Kadazandusun speech recognition system. Universiti Malaysia Sabah. (Unpublished)

	Text WEB-BASED KADAZANDUSUN SPEECH RECOGNITION SYSTEM.24pages.pdf Download (557kB)
	Text WEB-BASED KADAZANDUSUN SPEECH RECOGNITION SYSTEM.pdf Restricted to Registered users only Download (3MB)

Abstract

This project investigates a model for the classification and develops the web-based system of speech recognition for the Kadazandusun language. It is specifically designed for Universiti Malaysia Sabah (UMS) students and lecturers that are studying or teaching the Kadazandusun language. Nowadays, only the elderly of the Kadazandusun ethnic are fluent in their language and the majority of the ethnic’s youth cannot speak it fluently yet still understand their language, while some cannot understand the language spoken. This caused the existential crisis of the Kadazandusun language to arise. Furthermore, there are not many speech recognition systems that were developed for the language itself. There is only a little information of research regarding Kadazandusun speech recognition. The main purposes of this project are to investigate, recognize, and evaluate the Kadazandusun language speech recognition based on Mel-frequency Cepstral Coefficient (MFCC), Feed Forward Neural Network, and Principle Component Analysis. A website application is developed which integrates the model for speech recognition system. Based on the investigation (training and testing), the MFCC and Neural Network produce 89.22, 86.82, 87.54, 85.92, and 85.61 mean for classification accuracies using 11, 12, 13, 14, and 15 of MFCC coefficients respectively. Coefficient 11 was chosen as MFCC coefficient therefore it can be used as basic speech recognition for the Kadazandusun. Future works include collecting more data to enhance the usability of admin page, improving the pre-processing method, and hyperparameter tuning, as well as adding new feature to this system.

Item Type:	Academic Exercise
Keyword:	Classification , Kadazandusun , Speech recognition , Mel-Frequency Cepstral Coefficient , MFCC , Hyeperparameter , Feed forward neural network
Subjects:	P Language and Literature > PL Languages and literatures of Eastern Asia, Africa, Oceania > PL1-8844 Languages of Eastern Asia, Africa, Oceania > PL5001-7511 Languages of Oceania > PL5001-7101 Austronesian, Papuan, and Australian languages > PL5051-5497 Malayan (Indonesian) languages Q Science > QA Mathematics > QA1-939 Mathematics > QA71-90 Instruments and machines > QA75.5-76.95 Electronic computers. Computer science > QA76.75-76.765 Computer software
Department:	FACULTY > Faculty of Computing and Informatics
Depositing User:	DG MASNIAH AHMAD -
Date Deposited:	18 Jul 2022 19:19
Last Modified:	18 Jul 2022 19:19
URI:	https://eprints.ums.edu.my/id/eprint/33292

Actions (login required)

View Item