盲语音去混响研究综述  

A Review of Blind Speech Dereverberation Research

作　　者：曲万春 梅铁民 QU Wanchun;MEI Tiemin(School of Automation and Electrical Engineering,Shenyang Ligong University,Shenyang 110159,China)

机构地区：[1]沈阳理工大学自动化与电气工程学院,沈阳110159

出　　处：《微处理机》2024年第5期1-6,12,共7页Microprocessors

摘　　要：盲语音去混响作为语音信号处理领域的一个重要研究问题,在通信、人机交互、人工智能、生物医学工程和消费电子等多个领域具有广泛应用。根据声源是否已知,盲语音去混响主要分为两种方法:在声源未知的情况下对接收信号进行盲解卷积,或在已知声源的情况下对声源信号进行盲预补偿。本文在详细梳理该领域发展历程与研究成果的基础上,对这项技术的应用前景、主要研究方法、研究成果以及存在的问题进行了深入分析和论述,为该领域的进一步深入研究打下基础。Blind speech dereverberation,as a crucial research topic in the field of speech signal processing,has widespread applications in various domains including communication,human-computer interaction,artificial intelligence,biomedical engineering,and consumer electronics.Based on whether the sound source is known,blind speech dereverberation primarily employs two approaches:blind deconvolution of the received signal when the source is unknown,or blind pre-compensation of the source signal when it is known.This paper provides a comprehensive review of the development and research achievements in this field,offering an in-depth analysis and discussion of the technology's application prospects,main research methods,significant findings,and existing challenges.By doing so,it lays a foundation for further in-depth research in this area.

关 键 词：盲去混响 盲SIMO系统辨识 卡尔曼滤波 语音信号 

分 类 号：TN911[电子电信—通信与信息系统]