A method for adaptive communication echo cancellation for voice communication

Abstract

An adaptive communication echo cancellation method for voice communication, the steps are mainly: A, echo cancellation; B, tap weight vector update: B1, residual signal square sequence E 2 (n) calculation; B2, the calculation of the weighted median residual, to obtain the weighted median residual σ(n) at the current moment n; B3, the calculation of the estimated value of M, to obtain the M of the residual of the current moment n Estimated function value ψ(n); B4, calculation of zero attraction vector, obtains zero attraction factor of tap weight coefficient, thus constitutes zero attraction vector F(n) of current moment n, B5, update of filter tap weight vector, in Introduce the M estimation function value and the zero attraction vector into the update formula to obtain the filter tap weight vector W(n+1) at the next moment n+1; C, repeat. The method has fast convergence speed, low steady-state error and good impact resistance.

Description

technical field [0001] The invention relates to an adaptive echo cancellation method in voice communication. Background technique [0002] During a call (voice communication), the sound signal will be reflected back to the signal source after being delayed or deformed to form an echo, and the echo phenomenon will seriously affect the quality of the voice call. For example, when making a phone call, because the speaker and microphone are placed in the same space, the local near-end microphone will receive the far-end voice from the local speaker and transmit it back to the far end, which will cause the far-end speaker to hear himself The sound will degrade the quality of the call. This phenomenon widely exists in voice communication systems such as satellite communication, hands-free phone, and conference call system. It is necessary to take effective measures to suppress the echo signal, eliminate its influence and improve the quality of voice communication. Adaptive echo...

AI Technical Summary

Problems solved by technology

Since the adaptive echo cancellation system is usually a sparse system, the length of its response system can reach hundreds of symbols, but only a few effective factors are non-zero coefficients, resulting in slow convergence and low echo cancellation performance

Moreover, when the noise of the remote input signal cannot be ignored, the traditional least mean square algorithm will produce a biased estimate; resulting in slow convergence speed of the algorithm, large steady-state error, and poor echo cancellation effect

In addition, when there is impulsive noise, the "error signal" is huge, and the tap weight vector of the filter will produce huge updates of errors, resulting in increased steady-state error and slow convergence

Images