Arabic Vowels Recognition by Modular Arithmetic and Wavelets using Neural Network

The recent rising increase of loudening activity in mobile communication domain draw new opportunities and shed some lights for applications of speech recognition including words and sentences. Text to speech or vice versa is incredibly critical issues in many computer applications. Where, English language has achieved vast success of the major part of interest. However, Arabic language speech recognition has been less attractive than English; because of its many nature difficulties, in term of, several alphabets forms and various dialects

There are 22 Arabic countries with around 350 millions Arabic speakers living in it or distributed all over the world. For this reason, Arabic language considered one of the most important and widely spoken languages in the world. Arabic is Semitic language that is characterized by the existence of particular consonants like pharyngeal, glottal and emphatic consonants. As well as Arabic language has some phonetics characteristics that are built around pattern roots (CVCVCV, CVCCVC, etc.).

The 28 letters can be used in a set of 90 additional combinations, shapes, and vowels. The 28 letters enclose consonants and long vowels such asى and ٱ (both pronounced as /a:/), ي (pronounced as /i:/), andو (pronounced as /u:/). The short vowels and some other phonetic pronouncing like consonant doubling (shadda) are not introduced using letters directly, but by diacritics. The diacritics are short strokes, where each can be located above or below the consonant. Arabic discretization is interpreted by three groups: short vowels, doubled case endings form, and syllabification marks. First set, short vowels are written as symbols above or below the letter in the word using diacritics. We have short vowels: fatha: it pronounce as /a/ sound and is an slanted dash above the letter, damma: it pronounce as /u/ letter sound and has form of a comma above the letter and kasra: it pronounce as /i/ letter sound and is an oblique dash under the letter as tabulated in Table 1.

Table 1. Diacritics above or below consonant letter

Therefore, it is essential to realize that, what we usually denote to “Arabic” is not single linguistic variety; rather, it is a set of separate dialects and communities. Classical Arabic is an older and literary figure of the language, exemplified by the type of Arabic used in the Quran, the holly book for Islam. Modern Standard Arabic (MSA) is a version of Classical Arabic based on a modern vocabulary. MSA is a formal standard popular to all Arabic-speaking countries and communities. It is the language utilized in the newspapers, radio and TV, in official speeches, in courtrooms, and in any kind of formal communication.

Though, it is not utilized for everyday speech, informal communication, which is classically applied in one of the particular dialects. The dialects of Arabic may roughly speaking be divided into two sets: Western Arabic, which consists of the dialects spoken in Morocco, Algeria, Tunisia, and Libya, and Eastern Arabic, which may be further divided into Egyptian, Levantine, and Gulf Arabic countries. These different dialects differ significantly from each other and from Modern Standard Arabic. Differences influence all levels of language, i.e. pronunciation, phonology, vocabulary, morphology, and syntax.