Double-end pronouncing robust structure and acoustic echo cancellation method

Abstract

The invention discloses a double-end pronouncing robust structure which comprises a far-end speech data caching module and a near-end speech data caching module, wherein the far-end speech data caching module is used for receiving and keeping far-end speech data, and the near-end speech data caching module is used for receiving and keeping near-end speech data; the far-end speech data caching module and the near-end speech data caching module are in signal connection with an autoregressive model; the autoregressive model comprises an adaptive filter which can carry out adaptive learning according to estimation errors, and the autoregressive model is connected with a residual echo suppression module which can output processed signals to the far end; and the adaptive filter carries out adaptive control in the process of double-end pronunciation, thereby reducing the learning speed. According to the invention, through adopting the update technologies of the autoregressive model and the adaptive filter, near-end speeches are estimated through the autoregressive model by using white noises, and the convergence speed is accelerated, meanwhile, the learning speed is subjected to adaptive control according to communication scenes, thereby avoiding the problem of filter divergence under the condition of double-end pronunciation. According to the invention, the quality of speech signals in full-duplex communication is ensured, and therefore the robust structure can be widely applied to the field of mobile communication.

Description

technical field [0001] The invention belongs to the technical field of speech enhancement in mobile communication, and in particular relates to a device and method for canceling acoustic echo in a mobile communication system, specifically, a double-end pronunciation robust structure and a method for canceling acoustic echo. Background technique [0002] Acoustic echo means that the far-end voice played by the near-end speaker is collected by the near-end microphone and then transmitted to the far-end, so that the far-end user can hear the voice of his previous speech. The presence of acoustic echoes seriously interferes with the communication quality. Acoustic echo canceller is an effective way to solve this problem. Therefore, in the mobile communication system, the echo canceller is an essential component. In order to keep the voice signal free from acoustic echo interference, the echo canceller must have a weighted terminal coupling loss of at least 46dB in the case of ...

AI Technical Summary

Problems solved by technology

However, the convergence speed of these variable step size methods will be affected when the ST and echo paths change

Images