带噪语音信号基音检测技术研究-信息与通信工程专业论文.docx

万方数据 万方数据 摘要 摘 要 随着信息社会的高速发展，语音技术的应用越来越广。准确分析语音信号从 而提取出表征语音本质特征的特征参数是语音信号处理的重要组成部分。基音周 期是描述语音信号激励源的极其重要的一个参数，准确提取语音信号的基音周期 对语音信号处理各个领域（比如语音合成、语音编码、语音识别、说话人识别、 语音分离等）来说都意义非凡。常用的基音检测算法对干净语音的检测效果还可 以但是在低信噪比下检测误差大为下降，甚至无法检测。然而在实际应用中，不 可避免的会受到背景噪声的干扰，因此本文针对带噪语音信号的基音提取进行研 究，旨在寻找鲁棒性和准确性更好的算法。在前人研究的基础上主要做了以下工 作： 1、介绍了常用基音检测算法的优缺点，并分析仿真了基音检测相关的基本理 论。 2、研究了基于频域增强和改进的子带谱熵预处理算法。通过改进的谱减法对 语音增强从而减少噪声的干扰，分析研究了改进的加权子带谱熵对清浊音进行判 决。实验证明该预处理算法有较好的性能。 3、对 WAUTOC 算法进行改进，得到了抗噪性能更好的算法。首先用循环平 均幅度差函数（CAMDF）代替平均幅度差函数（AMDF）；然后建立循环平均幅 度和函数（CAMSF），用循环平均幅度和函数（CAMSF）代替自相关函数（ACF）， 提出了基于 CAMDF 加权的 CAMSF 基音检测算法；最后对常用的中值平滑后处 理技术做了改进。 4、结合时域和频域，研究了一种基于谐波正弦自相关和时域匹配的基音检测 算法。通过谐波正弦自相关模型及最小二乘拟合技术提取含噪声语音信号的一个 基音谐波，然后利用循环平均幅度和函数（CAMSF）加权的脉冲序列的时域匹配 方案得到对应谐波的谐波次数从而提取基音周期，最后利用动态规划进行基音跟 踪以获得更平滑的基音轨迹，实验证明了该算法能更准确的提取含噪声语音的基 音周期。 关键词：基音检测，循环平均幅度和函数，谐波正弦自相关 I ABSTRACT ABSTRACT With the rapid development of the information technology, voice technology is being widely used with increasing frequency to many areas. Analyzing speech signal precisely and then extracting the characteristic parameters that characterize the essence of speech is an important part of speech signal processing. Pitch period is a very important parameter to describe the excitation source, and extracting pitch period accurately is of great significance to various field of speech signal processing such as speech synthesis, speech coding, speech recognition, speaker recognition, speech segregation and so on. The commonly used pitch detection methods have good performance for clean speech, but the effectiveness of these methods decreases drastically even unable to detect with low SNR. However, in practical, speech is inevitably influenced by background noise, so this paper focuses on the pitch detection for the noisy speech, and the objective is to develop algorithms with better accuracy and robustness. Based on the research and study of the predecessors, the main works of this paper are as follows: The advantages and disadvantages of the commonly us