语音业务中鲁棒性VAD算法分析  

Study of Robust VAD Algorithm in Speech Operation

作　　者：郭莉[1] 殷南[1] 王炳锡[1] 

机构地区：[1]中国人民解放军信息工程大学信息科学系,河南郑州450002

出　　处：《电声技术》2005年第9期41-45,共5页Audio Engineering

基　　金：国家自然科学基金(6037038)

摘　　要：采用话音激活检测(VoicedActivityDetection,VAD)技术的目的是检测语音通信时是否有话音存在,检测到静音时加以抑制,使其不占用或极少占用信道带宽,检测到话音时才对其进行压缩编码与传输。鲁棒性语音识别系统、数字移动通信和因特网实时语音传输等领域要求在恶劣声学环境条件下进行VAD检测,以节省带宽并抑制噪声,因此VAD技术是目前语音处理领域的重要问题。文中给出的几种最新VAD算法(EZCR-VAD,STAT-VAD和E-VAD)是在低信噪比环境下的话音检测具有很好的鲁棒性的算法。VAD （Voice Activity Detection） is used to determine the presence of speech in communication. The speech coder will be suppressed or operate at a highly reduced bit rate during silent segments, and the speech will be transmitted during speech segments. The emerging technologies in speech processing,such as robust speech recognition system, digital mobile communications and real-time internet speech transmitting,are strong driving forces for the development of VAD algorithms operating in adverse acoustic environment,so as to suppress noise and reduce transmitting bandwidth,thus VAD technology has become a major issue in speech processing. Several new VAD algorithms （EZCR-VAD,STAT-VAD and E-VAD）discussed in this paper can perform robust voice activity detection at very low signal-to-noise ratios （SNR）.

关 键 词：话音激活检测 信噪比 过零率 信息熵 

分 类 号：TN912[电子电信—通信与信息系统]