A Single-Channel Mix Signal Separation Technique
Xiefeng Cheng, Nianqiang Li, Yuhan Cheng, Zhengyu Chen · 2007
The paper presented a new signal processing technique to accomplish blind source separation when given only a single-channel TEOAEs mixture signal. One signal source can be generated by a set of weighted linear superposition of the time domain sub-band functions with independent component characteristic. By combining the independent sub-band function components into the single mixture signal, making the single mixture signal is transformed into a multi-dimensional vector from one-dimensional. Thus ICA can be applied to separate the extended single-channel mixture signal. It is researched too that principles of makeup of the independent sub-band function. The separation results of TEOAEs mix signal demonstrated the effectiveness and adaptability of the proposed method.