基于正弦+噪声模型的语音分析合成技术  

Speech Analysis and Synthesis Technology Based on Sinusoidal + Noise Model

作　　者：孙俊海 SUN Jun-hai(School of Information Engineering,East China University of Technology ,Nanchang 330013,China)

机构地区：[1]东华理工大学信息工程学院,江西南昌330013

出　　处：《电脑知识与技术》2019年第3期195-196,共2页Computer Knowledge and Technology

摘　　要：语音信号分析合成技术是语音信号处理的一个重要组成部分,一直是近年来研究的热点,特别是,近年来随着多媒体通信技术的迅速发展,人们对语音重建质量提出了更高的要求。为此,该文通过采用正弦模型和噪声模型相结合的分析合成方法,通过提取每一帧信号的幅度、频率、相位和残差等参数,合成每一帧语音信号,然后通过叠加的方式重建语音。研究表明,该模型是一种很好的语音分析合成模型,能合成高质量的语音。Speech signal analysis and synthesis technology is an important part of speech signal processing and has been a hot topic in recent years. In particular, with the rapid development of multimedia communication technology in recent years, people have put forward higher requirements for the quality of speech reconstruction. To this end, this paper combines the sinusoidal model and the noise model to analyze the synthesis method, extracts the amplitude, frequency, phase and residual parameters of each frame signal, synthesizes each frame of speech signal, and then synthesizes the speech by superposition. Research shows that the model is a good synthesis model of speech analysis, which can synthesize high quality speech.

关 键 词：语音重建 正弦模型 噪声模型 语音分析合成 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]