Note from the editor: "Expositions of Romanian scientists on the design of text-to-speech synthesis and natural language understanding and generation systems"

The special issue of the International Journal of Speech Technology is devoted to the assorted work of twelve Romanian speech scientists who have brought sharp analytical insights and new methodological approaches to the design of high performing text-to-speech (TTS) and natural language understanding and generation systems. Dragos Burileanu and his co-authors, Catalin Ungurean and Aurelian Dervis, confront head-on the problems arising from the fact that lexical stress is a fundamental component of the pronunciation of a word, making its correct placement crucial for prosody prediction and the resultant generation for high-quality TTS. Burileanu and his co-authors propound a statistical approach to lexical stress assignment for Romanian TTS synthesis, which is premised on 'n-gram language models at character level and a modified Katz back-off smoothing technique to solve the problem of data sparseness during training'. Radu Ion and Dan Tufis of the Romanian Academy, Institute for Artificial Intelligence, take a close look at how linguistic ambiguity, expressed at many different levels.