fanchirp变换域语音增强分析 word格式.docxVIP

Fan-chirp 变换域语音增强研究中文摘要Fan-chirp 变换域语音增强研究 中文摘要在噪声环境下，语音处理系统的性能会大大降低，严重影响语音处理质量。因此， 语音增强作为消除干扰噪声的语音前端处理，是非常有必要的。本文主要研究了 Fan-chirp 变换域的语音增强算法，第一次将 Fan-chirp 变换应用到了语音增强算法中， 基于 Fan-chirp 变换的特点，研究了 chirp 速率的估计方法和 Fan-chirp 变换域的噪声 估计算法，结合了梳状滤波器和基于超高斯混合模型的最小均方误差估计，实现了 Fan-chirp 变换域的语音增强算法。具体的研究工作如下：首先，由于 Fan-chirp 变换需要 chirp 速率作为参数，本文提出了一种估计 chirp 速率的方法，提出了三种不同的计算测度，通过循环遍历寻找测度极值的方法，可以 在低信噪比下准确估计 chirp 速率。其次，基于 Fan-chirp 语谱能量集中的优点，提出了一种 Fan-chirp 变换域的噪声 估计算法，与两种经典的噪声功率谱估计算法进行了对比，表明本文算法能有效地估 计噪声。最后，研究了 Fan-chirp 变换域的语音概率模型，得出了含噪语音服从超高斯混 合分布的结论，结合频域梳状滤波器和基于超高斯模型的最小均方误差估计，实现了 Fan-chirp 变换域的语音增强。实验中对不同信噪比的白噪声和 babble 噪声的含噪语 音进行了语音增强，将本文方法与两种经典的语音增强算法进行了对比，各种主客观 评价指标表明了本文方法的有效性。关键词：语音增 强；Fan-chirp 变换 ；噪 声估 计；最小均方误差估计；超 高斯 混合 模型作者：郑似辰 导师：吴红卫AbstractFan-chirp 变换域语音增强研究中文摘要Speech Enhancement in the Fan-chirp Transform DomainAbstractPerformance of speech processing system will be degraded under noise, which imposes a significant consequence on speech quality. Thus speech enhancement is of great necessity for noise rejection as a front-end processing of speech systems. This thesis studies speech enhancement in the Fan-chirp transform (FChT) domain, which is the first application of FChT into the speech enhancement domain. Based on the feature of Fan-chirp spectra, we propose an estimation method of the chirp rate as well as the noise estimation algorithm in the FChT domain. By combining a frequency-domain comb filter with the minimum mean-square error estimator based on the super-Gaussian mixture model, speech enhancement is implemented in the FChT domain. The main works are explained as follows:Firstly, we propose an estimation method for the chirp rate, which is a key parameter of FChT. The chirp rate can be obtained by nearly exhaustive search of the maximum of three proposed measures, and the estimates are accurate even in low signal-noise-ratio.Secondly, we present a noise estimation algorithm in the FChT domain, which benefits from the high concentration of Fan-chirp spec