语音增强的目的是从带噪语音中提取尽可能纯净的原始语音。目前语音增强的方法很多,本文对一段未知的带噪语音信号采用基于短时谱估计的改进的谱减法、维纳滤波法、最小均方误差法这三种常用的语音增强技术进行降噪处理,并根据实验结果对这三种语言增强技术的降噪效果进行了比较。 The purpose of speech enhancement is to extract the purest possible original speech from the noisy speech. At present, there are many methods of speech enhancement. In this paper, three commonly used speech enhancement techniques, such as improved spectral subtraction based on short-time spectrum estimation, Wiener filtering and least mean square error, are applied to noise reduction of a noisy speech signal. , And compared the noise reduction effects of the three language enhancement technologies based on the experimental results.