论文部分内容阅读
提出了一个基于最小统计及谱减法的语音增强方法。不采用端点检测,在语音帧内及噪声帧内都进行噪声更新。为了实时应用的需求,提出了一种加快最小值更新的方法,实验表明,该方法能有效削弱变电平噪声。
A speech enhancement method based on minimum statistics and spectral subtraction is proposed. Endpoint detection is not used, and noise is updated both in the speech frame and in the noise frame. In order to meet the demand of real-time application, a method to speed up the minimum update is proposed. Experiments show that this method can effectively reduce the level noise.