The medical teaching needs auscultation training, so a model people auscultation training system was designed based on capacitive sensing principle. The PIC32 CPU with charging time measuring unit was used as the system core. Capacitance sensors were set in different parts of the model, the sampled signal was digitalized and processed, the cancelling jitter algorithm and dynamic average filtering was used for improving signal, and then the simulation audio was played. At the same time, the acquisition data was sent to the workstation through Zigbee RF module for being processed. The experience results showed that the system could simulate the audio signal from the different model parts, and it’s useful for raising the training effect; the algorithms of dynamic average filtering and cancelling dithering played important role for keeping on the system stable.