The rules for a Korean text-to-speech system based on the formant synthesis method are proposed. This paper describes all the aspect of the synthesis rules for Korean language. The rules are required for two parts. Firstly, for a linguistic process which converts the input text into phonetic unit. Secondly, for a synthesis process which generates and concatenates the synthesis unit to form a data package for spoken output message. All these rules have been implemented on a personal computer with a digital signal processing board which is designed for real time processing.