Hardware design for blind source separation using fast time-frequency mask technique

In this paper, we propose a fast time-frequency mask technique that relies on the sparseness of source signals for blind separation (BSS) to separate mixture two input sounds in single signal automatically. Due signals, can be distinguished when it is transformed into domain. Most previous methods did not mention effect different angles accuracy. To overcome such problems, first define features which are normalized level-ratio and phase-difference. Next, use our method decrease variance Direction Arrival (DOA). This reduce so iterations k-means. Finally, generated according clustered features. Our does require any prior information or parameter estimation. The motivation proposed design incorporate BSS system with some smart voice appliances. application scenario, all non-human voices may appear regard as interference. We Signal Distortion Ratio (SDR) Interference (SIR) make comparison. Based system, then present hardware design. TSMC 90-nm CMOS process. As cost-effective result, consumes about 120 K gates executes frequency 10 MHz. power consumption only 2.92 mW low considerations.