The first chapter is an introduction made in the thesis. The notion of language is defined and it describes the level of speech analysis: acoustic, phonetic, phonological, morphological, syntactic, semantic and pragmatic. Also, it describes the most common technology for speech synthesis: formants synthesis and concatenation synthesis.The second chapter describes eSpeak, as the system for speech synthesis. This section through eight subchapters describes files, attributes, rules and other elements that must be defined when creating a new or improving an existing speech synthesis of a language. The first subchapter provides general information about the software. The second subchapter contains a list of all supported languages in the system. In the third subchapter, defines the way you create new voice files in the system and describes the attributes that can be assigned in eSpeak voice. The fourth subchapter contains an explanation of how the software defines the intonation in a language. Through the presentation of _rules and _list files, the fifth chapter describes how eSpeak converts text to phonemes. The sixth and seventh subchapter are dealing with the definition of phonemes and tables of phonemes in the system and their attributes. The last subchapter of the second subchapter provides information on using MBROLA voices in the eSpeak system.The third chapter describes eSpeakedit, an additional software that is closely related to the eSpeak, and is designed for users that improve speech synthesis for some language. Through a series of descriptions and graphics from the system, it is describing all the functionality, elements of graphical user interface and how to use the aforementioned system. The fourth chapter presents the practical part of this thesis. Through modification of the _rules and _list files for the Croatian language, the pronunciation of the text in Croatian has been upgraded.