Smart Spoken and the Diacritics Arabic Words Recognition System based on Qaidah Noraniah model

Abstract

This paper presents the Recognition System of spoken Arabic words and the diacritics based on Qaidah Noraniah concept. The Arabic language is one of the most popular languages in the world. There are not many speech recognition systems for the Arabic language. Because of its extensive vocabulary and morphology, it is more difficult to develop systems to recognize it. For some of these reasons, it uses agglutinative letters and diacritics (الحركات). The Qaidah Noraniah language model was designed to solve some of these problems. The model was trained using the CMU Sphinx program, and tested by dependent and independent speakers. The model achieved good results in recognizing single letters, recognizing letters with diacritics and the ability to differentiate between different diacritics of a single letter, as well as recognizing words not used in training where they consisted of letters used in training. The recognition rate achieved by the model for letters is 67.17% and 41.75% for words.