Diverse synthesis methods and text-to-speech (TTS) architectures are being developed and applied almost every day. This tendency raises the need for durable program systems that effectively assist research and development in this area. A flexible development system for multilingual text-to-speech and general speech research is introduced. The system was developed for use with the Multivox and Profivox concatenative speech synthesis systems, but its architecture makes it theoretically appropriate for a wide variety of purposes and different TTS systems. The system architecture and the functions of the development system are described. Keywords: TTS development system, speech research tools, system architecture, SGML derivative, object oriented design