Author

Dr. Bassel Al Khatib – Eng. Mouhamad Kawas - Eng. Ammar Al Nahhas – Student: Reem Kannous and Rama Bondok

Published in

Journal of Theoretical and Applied Information, Vol. 95, No. 3, 15^th February 2017

Abstract

The Arabic language is characterized by its vocal variations. Making its pronunciation a difficult task for Arabic learners. In this paper, we show how we built a mobile application that can detect mispronounced words and guide the user to the correct pronunciation. Foreigners and children can learn Arabic pronunciation in a friendly manner using our application. Our mobile application is customized to help them learn The Holy Quran recitation in particular.
The process of the application compares the user sound (sound signal) of a single word with the set of correct recordings of this word pronunciation. This paper proposes the use of MFCC features to extract features from the speech signal. It also demonstrates the use of a modified version of DTW algorithm to compare the features of the user and the teacher.

Keywords: Mispronunciation Identification System, Mel-Frequency Cepstrum Coefficients, Dynamic Time Warping, Speech recognition.

Link to read full paper

http://www.jatit.org/volumes/Vol95No3/3Vol95No3.pdf