The TTS (Test-To-Speech) systems are finding increasingly important place in today’s IT-driven world that has transformed into the Global Village. But like many other IT tools and technologies, TTS systems are severely limited in their lingual scope. Since the human languages vary greatly in their written forms and phonology, most of the tasks within a TTS system become language-dependent. Natural sounding speech output is the aim of a TTS system become language-dependent. Natural sounding speech output is the aim of a TTS system. Majority of the commercial speech synthesizers use the speech-unit concatenation. Most of the basic algorithms to generate concatenative speech output are language-independent, which harness a database or language-dependent lookup tables called phonetic lexicon that contains the basic speech units. Although Sindhi language is modern language of the time and at the same time it is the language of communication and modern electronic media, there exists a vacuum as far as the development of modern electronic media, there exists a vacuum as far as the development of modern information technology tools for the language is concerned. In this paper, The authors describe phonology of Sindhi language and propose a phonetic lexicon that may lead to natural sounding concatenative synthesis of Sindhi speech.