Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Engineering and Sciences

First Advisor

Veton Kepuska

Second Advisor

Ivica Kostanic

Third Advisor

Marius Silaghi

Fourth Advisor

Carlos E. Otero


Arabic texts suffer from missing diacritics (short vowels) which become obstacles for new learners. Speech Recognition is the translation of words spoken to text through intelligent computer programs. As of today, it has been integrated into many computer systems. Arabic Speech Recognition has made progress over the years, but it is still not as good as English speech recognition due to the problem of short vowels not being recognized. This is mainly because the Arabic language is unlike the English language in the nature because it is a Semitic language. This is reflected in different characteristics such as grammar, morphology, and the number of vowels. If these differences are taken into consideration, Arabic Speech recognition may give more accurate results. This research finds a way to recognize the short vowels by benefiting from the characteristics of the Arabic language. The research claims to investigate one of the important reasons which is the differences in the number of vowels. There are at least twenty vowels in the English language which are close to each other in pronunciation, the Arabic language only has three short vowels which are far from each other, and by elongating those short vowels when they are pronounced, the long vowels are produced. Other researches said that the vowels could be recognized using formants. The formants measurements of English vowels are close, so it is hard to use those measurements to recognize one from the others. The formants measurements of Arabic vowels are far from each other, so it is possible to recognize them. This research applies this idea using the Euclidian distance method to measure the distances between formant values to recognize Arabic short values as isolated or inside Consonant-Vowel-Consonant-Vowel-Consonant-Vowel pattern words. Since, there are no available Arabic corpus which include Arabic vowels and Arabic words with the required pattern, an Arabic corpus was built through collecting data from adult male native Arabic speakers. In addition to having CVCVCV pattern words and isolated vowels, the corpus also included all the sounds of the Arabic language so that it can be used by any future researcher that needs the raw data. Necessary programs with the purpose of using formants to recognize short vowels were designed and developed using the MATLAB software, the programs extract the formants using the LPC method and calculate the mean values for each one of the short vowels using the words in the corpus, then use those means to recognize the isolated short vowels and the short vowels within the words in the corpus. The results showed that if highly qualified volunteers were chosen to read the Arabic text then higher rates of recognition for isolated short vowels and the short vowels involved in words can be achieved. This research revealed that, some of the characteristics of a language can be used for vowel recognition or to enhance the existing methods for speech recognition.


Copyright held by author