基于机器学习的语音增强技术  

Language Enhancement Technology Based on Machine Learning

作　　者：杨涛[1] YANG Tao(Luohe Vocational Technology College,Luohe 462000,China)

机构地区：[1]漯河职业技术学院,河南漯河462000

出　　处：《电声技术》2024年第3期39-41,共3页Audio Engineering

摘　　要：主要研究基于机器学习的语音增强技术,以提升语音信号的质量。首先,介绍基于机器学习的语音增强系统框架。其次,详细探讨谱减法与深度神经网络(Deep Neural Network,DNN)相结合的语音增强方法的数学原理。最后,采用NOISEX-92数据集测试与评估提出的方法。实验结果表明,基于谱减法与DNN的语音增强方法在提升信噪比和语音清晰度方面取得显著的效果,能够有效提升语音通信质量。This paper mainly studies the speech enhancement technology based on machine learning to improve the quality of speech signals.Firstly,introduce the framework of a speech enhancement system based on machine learning.Secondly,we will explore in detail the mathematical principles of speech enhancement methods that combine spectral subtraction with Deep Neural Networks(DNN).Finally,the proposed method was tested and evaluated using the NOISEX-92 dataset.The experimental results show that the speech enhancement method based on spectral subtraction and DNN has achieved significant results in improving signal-to-noise ratio and speech clarity,and can effectively improve the quality of speech communication.

关 键 词：谱减法 深度神经网络(DNN) 语音增强 去噪 

分 类 号：TN64[电子电信—电路与系统]