The goal of our work is the study of some peculiarities of speech comprehension in noise, stability of speech signal towards noise effect and allocation of elements which influence correct recognition. The first aim was to find such a level of correlation signal/noise at which 100% recognition of the material visibly fails. Second we tried to analyze the effect of speaker`s individual speech characteristics on speech comprehension. And our third aim was to make an attempt to allocate those elements in speech signal which are the most importaint for a listener in the process of speech recognition in noise. 
The speciality of our research is that it is held on the base of two languages: Russian and Greek. Part of the research connected with the Greek language was carried out with the help of Wire Communications Laboratory at University of Patras, Greece. Choosing of Greek language as a second language for the research can be explained by those facts that it is, like Russian, is a flective language and the place of a word stress is free. We were mostly interested in comparison of the results obtained for Russian and Greek languages. One of the main goals was to discover that features which are common for these two languages and those which present different tendencies.