This paper presents ideas, methodology and progress report of the project aimed at developing educational software for pronunciation training. The software is supposed to be used by people studying English as a foreign language. It may turn out handy in distant education courses and for additional phonetic drilling when the human teacher is not available. The system which is being developed at Moscow State Linguistic University (MSLU) pursues three major goals: firstly, detecting and classifying phonetic and phonological errors of users, secondly, assessing user's pronunciation, and thirdly, explaining the errors and formulating recommendations how to correct the errors made by the user during the training session. The work on the described project is conducted at the Department of Applied and Experimental Linguistic of MSLU and is financially supported by the grant of the Ministry of Education of the Russian Federation.