Blind audio source separation based on a new system model and the Savitzky-Golay filter

Abstract Blind source separation (BSS) is a research hotspot in the field of signal processing. This scheme widely applied to separate group signals from given set observations or mixed signals. In present study, Savitzky-Golay filter smooth signals, adopt simplified cost function based on noise ratio (SNR) and obtain demixing matrix accordingly. To this end, generalized eigenvalue problem solved without conventional iterative methods. It founded that proposed algorithm has simple structure can be easily implemented diverse problems. The obtained results demonstrate good performance model for separating audio cases with high ratios.