语音识别系统通常是将在安静的环境下训练得到的参数应用于实际环境中。如果实际环境也是安静的 ,则语音识别系统可以令人满意地工作。然而 ,当实际环境中有噪声存在时 ,语音识别系统性能急剧下降。为了让语音识别系统在安静的环境和有噪声的环境中都获得令人满意的工作性能 ,研究了一个将语音增强器和语音识别器级连起来的系统。该系统中 ,语音增强作为前端处理用于提高识别器输入端信号的信噪比。通过 3种不同的增强算法用于纯净语音和3种类型带噪语音的实验结果分析比较表明 ,这一方法对纯净语音的识别精度几乎没有任何改变而大大提高了系统的抗噪声性能 Speech recognition system is usually used in quiet environment training parameters applied to the actual environment. If the actual environment is also quiet, the speech recognition system can work satisfactorily. However, when there is noise in the actual environment, the performance of speech recognition system drops sharply. To make speech recognition systems work satisfactorily in both quiet and noisy environments, a system that connects speech enhancers to speech recognizers has been studied. In this system, speech enhancement is used as front-end processing to increase the signal-to-noise ratio of the signal at the input of the recognizer. The experimental results of three different enhancement algorithms for pure speech and three types of noisy speech show that this method improves the anti-noise performance of the system with almost no change in the recognition accuracy of pure speech