摘要
为提高在低信噪比下语音增强系统输出的语音质量,提出一种基于随机共振理论的谱减语音增强方法。利用二次采样随机共振对低信噪比语音信号进行预处理,使微弱语音、噪声与非线性系统产生协同效应,减小噪声的同时放大微弱语音信号;对经随机共振处理过后的语音信号采用谱减法进行二次降噪,优化低信噪比下的语音增强的效果。模拟仿真结果表明,随机共振与谱减法的复合方法可明显提升低信噪比下输出语音信号的输出信噪比及感知语音质量得分。
To improve the speech quality of speech enhancement in low signal-to-noise ratio(SNR)environment,a speech enhancement algorithm based on stochastic resonance(SR)and spectral subtraction was put forward.The twice sampling SR system was adopted as speech preprocessing to make synergistic effect among weak speech signals,noise and non-linear systems,which not only reduced noise but also amplified weak speech signals.And the spectral subtraction was used to improve the effects of speech enhancement twice.Experimental results show that the proposed method can improve SNR and enhance the perceptual evaluation of speech quality(PESQ).
作者
蔡文坚
王辅忠
张慧春
卢欢欢
CAI Wen-jian;WANG Fu-zhong;ZHANG Hui-chun;LU Huan-huan(College of Science,Tianjin Polytechnic University,Tianjin 300387,China)
出处
《计算机工程与设计》
北大核心
2018年第2期499-502,541,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(61271011)
安徽大学光电信息获取与控制教育部重点实验室开放课题基金项目(OEIAM201405)
关键词
随机共振
二次采样
谱减法
语音增强
信噪比
stochastic resonance
twice sampling
spectral subtraction
speech enhancement
signal-to-noise ratio
