摘要
对一个用语音遥控的视听系统而言，视听系统本身所发出的声音常常造成语音识别的错误．本文所采用的自适应声干扰抵消技术能有效地降低这种声干扰．基于自适应声干扰抵消最佳性能的分析，导出了抵消性能与权系数的个数以及混响时间的表达式，并实时地实现了带有1024个权系数的自适应声干扰抵消系统．在一般居室中，声干扰的抵消可达26dB．
For an AV system under the remote control of speech, errors of speechrecognition are often observed caused by the sound produced by the AV system itself.Self-adaptive acoustic interference cancellation is used to reduced the acousticsinterference effectively. Based on the analysis of the optimum cancellation, therelations between cancellation and weight number or reverberation time have beenderlved. A real-time self-adaptive acoustic interference cancellation system with 1024weights is realized. In an ordinary living room, 26 dB acoustic interferencecancellation could be obtained.
出处
《应用声学》
CSCD
北大核心
1997年第4期1-6,26,共7页
Journal of Applied Acoustics
关键词
视听系统
声干扰
自适应抵消
语音识别
Acoustics interference produced by AV system, Self-adaptive interference cancellation, Real-time, High-speed digital signal processing, Grouped LMS algorithm
