A good starting point is speech synthesis, particularly diphone synthesis