The linguistic processor in a multi-lingual text-to-speech and speech-to-text conversion system

This paper sumn1arizes the work being undertaken in ESPRIT project #860, 'Linguistic Analysis of European Languages'. The major objectives of this project are to provide statistics on the frequency of use of words, n-graphs, n-phones, and ambiguities (both in spelling and in pronunciation) as well as to develop a language modelthat can help in resolving ambiguities and that can perform a number of additional tasks in office automation and in speech technology. The language model will be developed and all attendant information will be provided in a unified way for a large number of European languages. In this presentation attention will be focussed on the language model.