Authentication is generally required on systems which need safety and privacy. In common, typed username and password are used and applied in authentication system. However, this type of authentication has been identified to have many weaknesses. In order to overcome the problem, many proposed authentication system based on voice as unique characteristics of human. We implement Dynamic Time Warping algorithm to compare human voice with reference voice as the authentication process. The testing results show that the system accuracy of the speech recognition average is 86.785%.