In this paper the system structure of DSR is first introduced, then the major techniques used such as the front-end feature extraction algorithm, feature compression, error correction and detection, as well as the server side speech reconstruction algorithm are analyzed, and finally the potential application of DSR is briefly discussed.