This paper describes the special aspects of the Chinese Mandarin speech synthesis, A comprehensive Chinese speech synthesis system is introduced which is capable of distinguishing the four spoken tones. Three levels of implementation of the Chinese speech are described. The first level is the phonetic-element based realization. The second is the syllable based realization. The third is the word based ( combination of the characters) realization. A lot of rules of variations of intonation are taken into account. The intelligibility and quality of the synthesized output speech are good.