A Method of Adaptive Deviation Compensation Echo Cancellation Based on Least Cubic Absolute Value

Abstract

The invention belongs to the technical field of adaptive echo cancellation and discloses an adaptive deviation compensation echo cancellation method based on least absolute third. The method comprisesthe following steps: sampling the current signal, and processing the sampled signal to obtain the noise vector eta(n) and the residual signal e(n) at the current moment n; obtaining the deviation compensation vector rho(n) at the current moment n according to the noise vector eta(n), the residual signal e(n) and the filter tapping weight coefficient vector w(n) at the current moment n; judging whether a call is ended, if no, obtaining the filter tapping weight coefficient vector w(n+1) at the next moment (n+1) according to the deviation compensation vector rho(n) at the current moment n, enabling n to be equal to n+1, and repeating the steps; otherwise, ending repetition. Through the adoption of the method, mixed noise in far-end signals can be eliminated on the basis of rapid realizationof fast convergence and low steady-state misadjustment, the defect that the traditional echo cancellation method has larger deviation during echo cancellation is overcome with the method, and the echo cancellation effect is better.

Description

technical field [0001] The invention belongs to the technical field of adaptive echo cancellation, and in particular relates to an adaptive deviation compensation echo cancellation method based on the minimum cubic absolute value. Background technique [0002] When the user uses the hands-free function of the communication device or holds a multi-person network audio conference, the phenomenon of acoustic echo often occurs, that is, the caller can hear his own voice. The acoustic echo generation process is that the spatial distance between the microphone and the speaker makes part of the sound from the speaker enter the microphone after multiple reflections. This part of the sound is picked up by the near-end microphone and then transmitted to the far-end signal output, so the caller hears himself the sound of. This will affect the call quality and user experience, and more seriously, it will cause vibration and howling, which will seriously reduce the voice quality of the ...

AI Technical Summary

Problems solved by technology

However, when the remote sampling signal is mixed with noise, this mixed noise will make the traditional LMAT algorithm deviate in the echo cancellation, and it is difficult to achieve a satisfactory echo cancellation effect

Images