Acoustic echo cancellation

In telecommunication systems, quality of voice is getting more important. Quality of voice is related to independency of noise, echo or the state of the place where communication takes place. In voice processing, there are software enhancements, during transmission of voice via voice processing undesired data like noise and echo can be cancelled. Acoustic echo is a type of echo that is formed due to the acoustic characteristics of the place where communication takes place. Acoustic echo varies due to types of communication site, that is mobile, stationary, open or closed place. Acoustic echo cancellers comprise of adaptive echo canceller, double talk detector (DTD) and nonlinear processor (NLP). During voice transmission, unknown characteristic of transmission system is predicted by benefiting use of adaptive filtering methods, adaptive filters creates a prediction signal of the echo, that is added to the system and by subtracting the prediction signal echo is removed.In this thesis, adaptive filtering methods LMS, NLMS, VSLMS, VSNLMS and RLS are used. NCC double talk algorithm is applied and under different types and levels of noisy environmental conditions, results are analysed in terms of ERLE, MSE and computational complexity.