||Wireless Sensor Networks (WSN) has recently become tremendous application and research area. Accordingly, localization issue of WSN is becoming a hot topic, which aims to recognize the position of concerned object. However, it is well known that WSN is quite complicated in a real environment due to the characteristic of ad-hoc network and also limited sensing ability of sensor itself. In this thesis, we build up a system using acoustic localization based on wireless sensor network, which can be adopted in an indoor environment. We carefully design a series of experiments using Micaz nodes and MTS310 sensors. After that, we propose a new algorithm called Cycle Distance Sum (CDS) used for acoustic localization especially for the detection of human voice. This thesis presents the entire construction of the whole process of acoustic localization, including time synchronization, sound detection, processing of sound waves, and localization algorithm. Also the evaluation of CDS will be presented. Our work should be beneficial to the researchers who want to develop a practical acoustic localization system in wireless sensor networks.