Adaptive Noise Cancellation Combined Tree-Structured Sub-Band Filtering and Cross-Talk Adaptive Filtering: A Two Microphone Approach

摘要

In this paper, we have proposed an efficient LMS (Least Mean Square) adaptive noise cancellation algorithm based on sub-band techniques. It has a double-microphone structure and combines tree-structured sub-band filtering with two LMS based cross-talk adaptive filters. Simulation results show that it can improve the SNR (Signal to Noise Ratio) from 5 - 10 dB to 15 - 20 dB in an ambient environment and has been implemented on TMS320C3x DSP. It can real time running on the DSP with computing power about 5 MIPS and run time memory below 1K words.