In this paper, we propose a system level design for blind source separation to separate mixture automatically. We use a fast time frequency mask technique by combining the binary time-frequency mask and spectral subtraction method. Our primary object is to separate the noise source from the targeted source and reduce the noise influence. Not only the audio quality is concerned but also the reduction on complexity is maintained in our proposed algorithm. We also evaluate the separated signal in terms of SDR and SIR on both design environments. The simulation results show that the proposed algorithm can provide high performance on audio source separation algorithm. We have implemented the overall system both on PCs and an embedded system.