小波语音增强算法在说话人识别中的应用

干净语音环境下识别率很高的说话人识别系统，在有噪声语音环境下识别性能显著降低。针对这一问题，将小波语音增强算法应用于说话人识别系统，提出一种结点阈值去噪新方法。语音增强主要目的是从带噪语音中尽可能地提取纯净的原始语音。在不同信噪比条件下进行实验，结果表明，提出的方法比传统的阈值法能更好地提高语音质量。

Application of Wavelet Enhancement Algorithm in Speaker Recognition

To the problem that a speaker recognition system with high performance in quiet environment will become deficient with unacceptable recognition performance in noisy environment,wavelet transformation is applied in speaker recognition system,and a node dependent thresholding algorithm is proposed.The aim of speech enhancement is to extract pure information.Experiment results on different signal noise rates show that this algorithm can improve the speaking quality better than traditional thresholding method.