作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (14): 140-142. doi: 10.3969/j.issn.1000-3428.2011.14.046

• 人工智能及识别技术 • 上一篇    下一篇

基于噪声倒谱阈值频谱估计的语音活动检测

李 宇 1,2,郭雷勇 1,2,谭洪舟 2   

  1. (1. 广东药学院医药信息工程学院,广州 510006;2. 中山大学信息科学与技术学院,广州 510275)
  • 收稿日期:2010-12-10 出版日期:2011-07-20 发布日期:2011-07-20
  • 作者简介:李 宇(1977-),男,讲师、博士,主研方向:语音、医学信号处理;郭雷勇,讲师、博士;谭洪舟,教授、博士、博士生导师
  • 基金资助:
    国家自然科学基金资助项目(60874060)

Voice Activity Detection Based on Noise Cepstrum Thresholding Spectral Estimation

LI Yu 1,2, GUO Lei-yong 1,2, TAN Hong-zhou 2   

  1. (1. College of Medical Information Engineering, Guangdong Pharmaceutical University, Guangzhou 510006, China;2. School of Information Science and Technology, Sun Yat-sen University, Guangzhou 510275, China)
  • Received:2010-12-10 Online:2011-07-20 Published:2011-07-20

摘要: 针对低方差频谱估计的语音活动检测(VAD)中Welch频谱估计方法计算量大的问题,提出利用倒谱阈值方法估计VAD中的噪声功率谱。该方法在静音时期为噪声的倒谱设置阈值,利用快速傅里叶变换计算频谱,再更新VAD中的判决阈值。算法复杂度分析与仿真结果表明,该方法的检测性能与Welch方法相当,计算量降低约18%,同时降低整个VAD的时间复杂度。

关键词: 语音活动检测, 频谱估计, 倒谱阈值方法, 功率谱密度, 快速傅里叶变换

Abstract: Aiming at high computational complexity in Voice Activity Detection(VAD) which uses low variance spectral estimation, this paper proposes a method using noise cepstrum thresholding spectral estimation to compute noise Power Spectral Density(PSD). In speech absent period, a threshold is set for smoothing the noise cepstrum, then noise PSD is estimated by using Fast Fourier Transform(FFT), and detecting threshold in VAD system is updated. Computational complexity analysis and simulation results indicate that compared with Welch method, the computation of the method is reduced by 18%, and total time complexity of VAD is decreased.

Key words: Voice Activity Detection(VAD), spectral estimation, cepstrum thresholding method, Power Spectral Density(PSD), Fast Fourier Transform(FFT)

中图分类号: