This paper introduces the speech synthesis system developed by USTC for Blizzard Challenge 2011. USTC attended all the English tasks including a hub task and a spoke task. We developed a hidden Markov model (HMM) based unit selection system for both the tasks. And also some new techniques are employed in our speech synthesis system construction. Results of some internal experiments comparing these techniques are shown and analyzed. The evaluation results of Blizzard Challenge 2011 prove that our system performed well in the similarity and naturalness evaluation, and the differences in intelligibility between our system and the better systems are not significant.