随机共振在强噪声环境中语音增强应用

Application of Stochastic Resonance Mechanism to Speech Enhancement in Strong Noise Environment

传统的语音增强方法是在保持语音可懂度和清晰度的前提下,尽可能地从带噪语音中提取需要的纯净语音,而在强噪声环境中,语音信号表现为弱信号,去噪变得困难。基于Hodgkin-Huxley神经元阈上非周期随机共振原理,提出一种自适应调节,添加最佳噪声来进行语音随机共振,从而实现语音增强。Matlab实验结果表明,在强噪声环境中实现对语音信号增强,信噪比提高明显,且效果优于传统算法。方法具有一定鲁棒性,提供了在强噪声环境中增强语音信号的新思路。

The traditional speech enhancement methods usually improve the quality of the noisy speech by extracting the clean speech from it as much as possible, but for speech polluted by strong noises , speech signal is weak, so it is hard to denoise in this way. Based on the theory of suprathreshold aperiodic stochastic resonance in Hodgkin - Huxley Model, a new self-adaptive adjusting, adding optimum noise speech enhancement method is given. Matlab results indicate that speech signal enhancement can be achieved in strong noise environment, SNR is increased greatly, and better than traditional method. This method is robust, provides a new idea for enhancing the speech signal in strong noise environment.