首页 | 本学科首页   官方微博 | 高级检索  
     检索      

低信噪比下基于谱熵的语音端点检测算法
引用本文:李晔,张仁智,崔慧娟,唐昆.低信噪比下基于谱熵的语音端点检测算法[J].清华大学学报(自然科学版),2005,45(10):1397-1400.
作者姓名:李晔  张仁智  崔慧娟  唐昆
作者单位:清华大学,电子工程系,微波与数字通信技术国家重点实验室,北京,100084
基金项目:国家自然科学基金资助项目(60272020)
摘    要:为提高语音端点检测系统在低信噪(0 dB以下)下检测的准确率,提出了一种基于谱熵的端点检测算法。将每帧信号分为16个子带,选取频谱分布在250~3.5 kH z并且能量不超过该帧总能量90%的子带,计算经过语音增强后的子带能量以及各子带信噪比,根据各子带信噪比的不同调整其在整个谱熵计算过程中的权重,然后平滑谱熵,以最终的谱熵作为端点检测的依据。实验结果表明,此方法在较低的信噪比下能够显著地提高端点检测的准确率。对坦克噪声,检测效果明显优于G.729中的端点检测算法,即使在-5 dB的信噪比下,仍然可以达到95%以上的检测率。

关 键 词:语音信号处理  端点检测  谱熵  语音增强  信噪比
文章编号:1000-0054(2005)10-1397-04
修稿时间:2004年10月11

Voice activity detection algorithm with low signal-to-noise ratios based on the spectrum entropy
LI Ye,ZHANG Renzhi,CUI Huijuan,TANG Kun.Voice activity detection algorithm with low signal-to-noise ratios based on the spectrum entropy[J].Journal of Tsinghua University(Science and Technology),2005,45(10):1397-1400.
Authors:LI Ye  ZHANG Renzhi  CUI Huijuan  TANG Kun
Abstract:Voice activity detection(VAD) in low signal-to-noise ratio(SNR environments is improved with an algorithm based on the spectrum entropy.Each frame is first divided into 16 bands with selection of bands with frequencies between 250 Hz and 3.5 kHz and energies below 90% of the total energy.The energy and the SNR of each band after speech enhancement are then calculated with the entropy band weight adjusted according to it's SNR.The smoothed entropy is then used for the voice activity detection.Test results show that the method significantly increases the voice activity detection ratio.For example,it works the detection accuracy is above 95% even with-5 dB noise whish is better than the G.729 algorithm for tank noise.
Keywords:speech signal processing  voice activity detection  spectrum entropy  speech enhancement  signal to noise ratio
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号