Multi-voice signal activities detection method based on terahertz radar
CSTR:
Author:
Affiliation:

(College of Electronic Science and Engineering, National University of Defense Technology, Changsha 410073, China)

Clc Number:

TN95

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments
    Abstract:

    Aiming at the problem the traditional voice acquisition method is difficult to correctly detect the voice activity at the desired voice source when multiple voice sources exist simultaneously, a multi-voice activities detection method based on terahertz radar was proposed. The short wave length of terahertz radar was utilized to improve the detection sensitivity for micro vibration. At the same time, the high range resolution brought by large broadband was used to distinguish different voice sources. After obtaining the voice signal, the signal quality was further improved for signal activities detection by spectral subtraction wavelet joint denoising method. The experimental results show that, when there are interfering voice source on the transmission path, the proposed method can more effectively distinguish multiple voice sources and obtain the voice activity at the desired voice source than the microphone-based method.

    Reference
    Related
    Cited by
Get Citation

WANG Yuanhao, WANG Hongqiang, YANG Qi, DENG Bin, LUO Chenggao. Multi-voice signal activities detection method based on terahertz radar[J]. Journal of National University of Defense Technology,2022,44(1):28-33.

Copy
Share
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
History
  • Received:January 25,2021
  • Revised:
  • Adopted:
  • Online: January 19,2022
  • Published: February 28,2022
Article QR Code