Article Details

Title · Authors · Abstract · Keywords · DOI · References

Research on Speech Endpoint Detection Algorithm in Low Signal-to-Noise Ratio Environment

FU Hongzi1, CHEN Xiduan1, ZHANG Lin2

(1.Zhejiang Industry and Trade Vocational College, Wenzhou 325003, China; 2.Wenzhou Technician Institute, Wenzhou 325000, China)

Section:Electronic Engineering · 2025, Vol. 9, No. 11 · Pages:5-8 · 2096-4706(2025)11-0005-04 · CLC number:TP39;TN912.3 · Document code:A · DOI:10.19850/j.cnki.2096-4706.2025.11.002

Abstract

To enhance the robustness and accuracy of speech endpoint detection in low Signal-to-Noise Ratio environment, a speech endpoint detection algorithm based on multi-window spectral estimation and subband quadratic spectral entropy is proposed. This algorithm integrates the advantages of subband quadratic spectral entropy and multi-window spectral estimation. First, the quadratic power spectrum of each frame of noisy speech signal is computed. Then, subband analysis is performed to derive the subband quadratic spectral entropy. Finally, on this basis, multi-window spectral estimation of multi-sine-window is incorporated into the subband quadratic spectral entropy to form new feature parameters for determining the start and end points of speech endpoint detection. Through MATLAB simulation verification, this algorithm has better noise robustness in low Signal-to-Noise Ratio environment compared to traditional speech endpoint detection algorithms.

Keywords

low Signal-to-Noise Ratio; subband quadratic spectral entropy; multi-window spectral estimation; robustness

References

  1. [1] Lv Hao, Zhang Xiaobing, Cai Cheng An improved algorithm for speech endpoint detection under low signal-to-noise ratio [J]. Journal of Wanxi University, 2021, 37 (2): 45-50
  2. [2] Zhang Junchang, Zhang Dan, Cui Li A robust adaptive threshold based speech endpoint detection method [J]. Journal of Xi'an University of Electronic Science and Technology, 2015, 42 (5): 115-119
  3. [3] Guo Wu, Ma Xiaokong Active Speech Detection Method in Complex Noise Scenes [J]. Journal of Tsinghua University: Natural Science Edition, 2016, 56 (11): 1190-1195
  4. [4] Zhu Jianwei, Sun Shuifa, but Zhiping, etc Speech Endpoint Detection Based on Subband Quadratic Spectral Entropy [J]. Microelectronics and Computer, 2011, 28 (3): 77-80
  5. [5] Wu Bian, Wang Zhong, Liu Xingtao Research on Algorithm for Speech Endpoint Detection under Strong Background Noise [J]. Computer Engineering and Applications, 2011, 47 (33): 137-139
  6. [6] WU B F,WANG K C. Robust Endpoint Detection Algorithm Based on the Adaptive Band-partitioning spectral Entropy in Adverse Environments [J].IEEE Transactions on Speech and Audio Processing,2005,13(5):762-775.
  7. [7] Zhang Tianqi, Zhang Zhan, Quan Jingguo, etc Quadratic Spectral Method for Pitch Detection of Speech Signals [J]. Computer Applications, 2005 (4): 934-936
  8. [8] Han Tao, Wang Ling, Liu Hui An Improved Endpoint Detection Method for Speech Recognition [J]. Microelectronics and Computer, 2008 (5): 146-149
  9. [9] RIEDEL K.S,SIDORENKO A. Minimum Bias Multiple Taper Spectral Estimation [J].IEEE Transactions on Signal Processing,1995,43(1):188-195.
  10. [10] CAI T T,REN Z,Z H H. Optimal Rates of Convergence for Estimating Toeplitz Covariance Matrices [J].Probability Theory and Related Fields,2013,156(1/2):101-143.
  11. [11] HALEY C L,ANITESCU M. Optimal Bandwidth for Multitaper Spectrum Estimation [J].IEEE Signal Processing letters,2017,24(11):1696-1700.
  12. [12] Song Zhiyong MATLAB Speech Signal Analysis and Synthesis: 2nd Edition [M]. Beijing: Beihang University Press, 2017

Funding

General Research Project of Zhejiang Provincial Department of Education (Y202456016)

About the authors

Fu Hongzi (1988-), female, Han nationality, from Shangrao, Jiangxi, teaching assistant, master's student, research direction: speech recognition, computer vision; Chen Xiduan (1989-), male, Han ethnicity, Wenzhou, Zhejiang, lecturer, master's student, research direction: signal and information processing; Zhang Lin (1986-), male, Han nationality, from Wenzhou, Zhejiang, lecturer, undergraduate, research direction: artificial intelligence and embedded electronic development.