摘要
针对高强度噪声背景下活动话音无法准确检测的问题,提出了基于多特征的语音活动检测算法,详细论述了该算法中语音信号的采样量化、预加重、分帧和加窗等预处理技术,分析了检测算法设计中的动态门限更新、短暂停顿平滑等关键因素,并总结出了多特征语音活动检测算法的流程图。通过基于硬件平台的算法测试和仿真分析,结果验证了该算法的合理性和有效性,对于复杂背景噪声环境下的活动话音检测有着重要的实用意义。
Under the high-intensity background noise,the voice activity can＇t be detected accurately,Aiming at this problem,a voice activity detection algorithm based on multi-features is proposed.In this paper,the technologies such as speech signal sample quantization,pre-emphasis,frame windows and other pre-processing are discussed in detail,the key factors such as dynamic threshold updating,pause smooth in algorithm design are analyzed,and the program flow chart is concluded.The result of test and simulation analysis on hardware platform proves the rationality and effectiveness of algorithm,which has important practical significance for voice activity detection under complex background noise environment.
出处
《无线电工程》
2011年第10期24-26,共3页
Radio Engineering
