In recent years, the research and application on sensor network became a very hot topic. People use sensors to collect information from environment, which is not accessible or not convenient to access. In this thesis, by building up a sensor network based real time eavesdropping system, called SensorEar, we show that sensor network can bring up very serious and detailed threat to personal privacy. We implement the system on Mica2 of Crossbow Inc., which is a small, cheap, resource limited, commercial-off-the-shelf, and TinyOS based platform. The SensorEar system features on its high data rate, which exceeds the limitation of the Mica2. It is achieved by the application of a self-designed multi-rate compression algorithm, and a multi-hop and multi-channel transmission scheme. The performance of this system is evaluated on several aspects such as sampling rate, transmission rate, packet loss rate and sound quality.