The beginning of the thesis presents introduction to the latest standard for the development systems with speech capabilities, VoiceXML. The description of the advantages when using VoiceXML instead of older technologies used in interactive voice response systems, follows in next chapters. Next part of the thesis deeply covers issues about how to use the available hardware, technologies and projects as a building blocks for creation of the interactive voice portal, with explained possibilities of system localization to e.g. Croatian language. This follows with the description of the ways how spech recognition and speech synthesis works. User centered design, with the special quotes on system usability part, is presented in the third part of the thesis. The techniques for improvements of the system usability, especially for voice systems, are covered more precisely. Because of the nature and "no interupts" requirements, interactive voice response systems have to pass very complex tests, which are taken from the beginning of the project development. The forth part of the thesis describe different methods and the ways that the voice systems can be tested and certified. Cetvrto poglavlje opisuje razne metode te na koje nacine je moguce testirati i certificirati glasovne sustave. Overview of the current development of voice and web technologies, which can be applied to voice systems, with the description of the new paradigm "v-commerce", are explained in the last part of the thesis.