Nondestructive voice quality stereo echo cancelling method

Abstract

The invention discloses a stereo echo cancellation method without loss of speech quality. The method is: performing FFT transformation on a frame of stereo left and right channel audio input signals at the nth moment to obtain its corresponding frequency domain signal; adding random to the frequency domain signal The phase noise is used to obtain the frequency domain input signal of the left and right channel speakers in the room that receives the speech signal; the IFFT transform is performed on the speaker frequency domain input signal to obtain the time domain input signal of the left and right channel speakers in the room that receives the speech signal; the speaker frequency domain input signal is used as a stereo echo For the offset reference signal, the frequency domain stereo echo cancellation algorithm is used to remove the echo in the microphone input signal in the room that receives the voice signal at the nth moment, thereby obtaining a voice signal without echo. This method is combined with the frequency domain algorithm of the stereo echo cancellation system to reduce the amount of calculation. At the same time, due to the insensitivity of the human ear to sound phase, adding phase random noise to the left and right channel signals has no impact on speech quality.

Description

1. Technical field [0001] The invention relates to a method for canceling stereo echo without loss of speech quality. 2. Background technology [0002] For some video conferencing systems, in order to transmit spatial information of speech, a stereo system is usually used, and the corresponding echo cancellation device should also use a stereo echo cancellation system. Unlike the single-channel echo cancellation system, due to the strong correlation between the left and right stereo channel signals, the system has a non-unique minimum value, which greatly affects the performance of the echo cancellation. The existing method to solve this problem is to perform non-linear processing on the left and right channel signals to reduce the correlation between them, for example, adopt half-wave correction to the left and right channel signals. In order to reduce the correlation between the left and right channels, the nonlinear change used cannot be too small, so that the speech qua...

AI Technical Summary

Problems solved by technology

In order to reduce the correlation between the left and right channels, the nonlinear change used cannot be too small, so that the voice quality after the nonlinear change is severely degraded, which is a major shortcoming of the existing method

Images