In this work, we present a system of speech synthesis from a text written in Standard Arabic, we explain the methodology used for its design and we demonstrate the principale stages of its conception which can be shown in the folowing two points; 1) The creation of the sound Database (BD) from the elaboration of a corpus. 2) And the acoustic generation of the text previously acquired. To insure this important step whih is the passage between the world of the writing to that of the speaking, we raise two techniques of Orthographic and Phonetic Transcription (by rules; in which we used 12 rules and by lexicon where we used 6 exceptions words) as well as two methods of speech synthesis (by phonemes and by diphones). The results obtained are shown at the end.