In the “post-epidemic era”, the requirement of online meeting grow rapidly and related technologies attracts a lot of attention. As an efficient method to improve the sound clarity and reduce noise, sound source positioning can strengthen the signal in the direction of sound source, while reduce the noise from other directions. In this work, a sound source position system based on FPGA was proposed, which was consisted of microphone matrix as signal input and FGPA as hardware platform to accelerate the localization algorithm. The microphone matrix was constructed by MEMS. On-chip system was built on Nexys Video, a FPGA development board carrying DDR3. The XADC of Artix-7 series FPGA was used to collect the electrical signals output by the microphone matrix and the obtained data was stored in DDR3. A sound source location algorithm acceleration module was designed in the FPGA and it acquired date from DDR3 via the stream transmission bus. Finally, the azimuth and angle of the sound source were outputted. Experiments demonstrated this system could achieve accurate date of the sound source by a commercial microphone matrix.