Fuzzy Approach to Speech Synthesis