Coefficient updating method for adaptive filter in adaptive echo cancellation

Abstract

The invention discloses a coefficient updating method for an adaptive filter in adaptive echo cancellation, and the method comprises the steps: estimating an echo signal y(k) according to a reference signal x(k) of a sound generation unit through the adaptive filter; calculating a residual signal e(k) according to the echo signal y(k) and a signal d(k) picked by a microphone; calculating a power spectrum xPow(f) of a frequency domain reference signal X(f); calculating a power spectrum ePow(f) of a frequency domain residual signal E(f); calculating the correlation power spectrum xePow(f) of a conjugate residual signal E*(f) of the frequency domain residual signal E(f) and the frequency domain reference signal X(f); calculating the correlation coefficient cohxe(f) of X(f) and E*(f); enabling a median cohxemed of cohxe(f) to serve as the adaptive step factor of the adaptive filter at each frequency point, and updating the coefficient of the adaptive filter. According to the technical scheme of the invention, the method enables the correlation coefficient between the residual signals and the reference signal to serve as the adaptive step factor so as to update the coefficient of the adaptive filter, and can achieve the effective echo cancellation of each voice state.

Description

technical field [0001] The invention relates to the echo processing technology in the process of real-time voice communication and human-computer interaction, in particular to an adaptive filter coefficient update method in adaptive echo cancellation. Background technique [0002] In the process of real-time voice communication, the sound from the speaker of the communication terminal will always be picked up by the microphone of the terminal. If it is sent out without processing, the other party can always hear his own voice; The sound emitted is picked up by the microphone, and the voice of the controller is picked up at the same time. If the voice from the interactive terminal is not eliminated in the signal picked up by the microphone, the interactive terminal will introduce strong interference when recognizing the voice of the controller, reducing the The success rate of recognition is reduced, which eventually causes difficulty in interaction. Therefore, in the proces...

AI Technical Summary

Problems solved by technology

[0004] The problem with the existing solution is that: on the one hand, the updating of the adaptive filter coefficients is not related to the filtering effect (residual signal), and the adaptive step factor is not adjusted in real time according to the echo intensity, which cannot be well guaranteed in various The Accuracy of Echo Cancellation in the Voice State

Images