Method and its system for eliminating stereophonic echo based on voice signal separate model

Abstract

The present invention provides a stereo echo cancellation system based on speech signal separation, comprising an echo canceller, a microphone, and a loudspeaker, and the echo canceller includes a periodic power detection module, an impulse response difference measurement module, a speech signal separation module, and Talking detection module, echo path estimation module, adder; A method of echo cancellation realized by using the above system, including the following steps: when there is no two-way intercom, estimating or updating the difference of the impulse response of the near-end room and the echo path Impulse response; separate the single-channel echo from the multi-channel echo; combine the source signal from the far-end room to obtain the estimation of the impulse response of the echo path, and then obtain the estimation of the separated single-channel echo; the near-end The received signals of the room microphones are separated and processed, and the corresponding estimated values ​​are respectively subtracted from the remaining single-path echoes, so as to realize echo cancellation. The invention has the advantages of simple structure, small calculation amount, strong real-time performance, easy hardware realization and good voice communication quality.

Description

technical field [0001] The invention relates to an echo cancellation technology in a closed speech environment, in particular to a stereo echo cancellation method and system based on a speech signal separation model. Background technique [0002] With the development of multimedia communication technology, video teleconferencing and vehicle-mounted hands-free wireless telephone systems have become increasingly popular. In such a closed voice environment composed of speakers, rooms and microphones, there are acoustic echoes that affect the normal progress of communication, so it is necessary to set up an echo canceller to reduce the interference of echoes. With the improvement of people's requirements for fidelity, in order to enhance the immersive feeling of the participants, it is essential to use stereo sound. However, due to the increase in the number of speakers and microphones, the resulting multi-channel echo cancellation question. Compared with the single-channel ec...

AI Technical Summary

Problems solved by technology

With the improvement of people's requirements for fidelity, in order to enhance the immersive feeling of the participants, it is essential to use stereo sound. However, due to the increase in the number of speakers and microphones, the resulting multi-channel echo cancellation question

Compared with the single-channel echo cancellation, the multi-channel echo cancellation represented by the stereo situation exists because of the strong correlation between the input signals. The solution of the echo canceller is not unique, and the offset is large, that is, the near-end cannot be uniquely identified. The impulse response of the room is inherently problematic and does not meet the application requirements well

Many of the existing echo cancellation patent technologies are aimed at single-channel echo cancellation. The use of traditional adaptive filters has achieved relatively good results, but the traditional method of directly extending the single-channel situation to multiple channels cannot meet the requirements. actual application needs

The Chinese invention patent application with the application number 200610034214.6 discloses a "multi-channel echo cancellation system based on SIMO channel blind identification", which is a technology for multi-channel echo cancellation. The separation technology is applied to the multi-channel echo cancellation model; however, this technology is difficult to overcome the defects of the huge amount of calculation and the slow operation speed of the blind signal separation technology itself, and the initial construction matrix and subsequent construction matrix that need to be used in this technology Both occupy a large storage space and will also lead to high hardware costs

Images