Performance analysis of acoustic echo cancellation using Adaptive Neruo Fuzzy Inference System

Abstract

Removal of echo from respiratory signal could be a classical problem. In recent years, adaptive filtering has become one in all the effective and popular approaches for the process and analysis of the respiratory signal. Adaptive filter allow to find time varied potential and to trace the dynamic variations of the signals. Besides, they modify their behavior consistent with the input. Therefore, they can find form variations within the ensemble and so they will get a much better signal estimation during this project work respiratory signals generated synthetically. After that, the echo has been mixed with respiratory signal. That echo has been invalidated from the respiratory signal by victimization accommodative filter algorithms (LMS and RLS) And Adaptive Neruo Fuzzy Inference System. This the performance analysis of the project techniques is completed in terms of signal Echo Return Loss Enhancement (ERLE), Signal to Noise Ration (SNR), Mean Square Error (MSE) and Convergence Rate. These properties depend upon a couple of parameters such as: variable step-size(for the LMS), for getting factor (for the (RLS). Also, it's true for each algorithms that the filters length is proportional to MSE rate and it takes longer to convergence for each algorithms. Comparison is formed between varied kinds of LMS and RLS algorithms supported their performance analysis. Then the simplest adaptive filter algorithmic program is compared with the performance of ANFIS.