带噪混叠语音信号盲分离算法研究作者:黄珊 拧扣机杜庆治来源:《软件导刊》2018年第01期豆浆器
摘要:
盲源分离也称盲信号分离,是指在源信号和传递信道的参数均未知的情况下,仅根据输入源信号的统计特性,通过观测信号恢复各个源信号的过程。语音信号的盲分离技术在计算机听觉、语音识别、语音增强等领域具有重大的研究意义。现有的有关语音信号盲分离研究基本不考虑噪声的影响,然而在现实生活中,接收到的语音信号不可避免地混有各种噪声。因此,对于带噪声混叠语音的盲分离方法研究具有十分重要的现实意义。针对带噪声混叠语音信号,提出一种基于稀疏编码和EFICA的分离方法。首先用稀疏编码去噪方法消除带噪混叠语音信号中的噪声,然后将经过去噪处理后的观测信号用EFICA方法进行盲分离。Matlab仿真实验结果表明,该算法对带噪声混叠的语音进行盲分离效果良好。 关键词:
微型液位开关混叠信号;语音信号;盲源分离;稀疏编码;EFICA
DOIDOI:10.11907/rjdk.172151
中图分类号:TP312
文献标识码:A文章编号文章编号:16727800磁分离(2018)001007403
Abstract:BSS is refers to the process of recovery each source signal only by the statistical characteristics of each input observation signal under the condition of the original signal and parameters of transmission channel is unknown,安全监控 which is also called the separation of blind source signal. The BSS of speech signal is of great significance to the study of computer hearing, speech recognition and speech enhancement. At present, most studies of the BSS of speech signal without considering the impact of noise, while in real life environment, the speech signals are mixed with all kinds of noise inevitably. So, there is of great practical significance to study the blind separation of noisy speech mixtures.Aimed at the noisy speech mixtures, a BSS method based on sparse coding and EFICA is proposed in this paper.First, the noisy speech mixtures is eliminated noise by the method of sparse coding, and then to blind separate the speech
地震的模拟实验
mixtures with EFICA. The simulation result shows that this method can achieve good effect in BSS of noisy speech mixtures.