Journal of South China University of Technology (Natural Science Edition) ›› 2021, Vol. 49 ›› Issue (4): 59-64.doi: 10.12141/j.issn.1000-565X.200019

Special Issue: 2021年物理

• Physics • Previous Articles     Next Articles

Study on the Localization Spectral Characteristics at Different Distances

ZHONG Xiaoli1 LAI Zhuowei1 SONG Hao2 YANG Xiaofei1   

  1. 1. School of Physics and Optoeletronics, South China University of Technology, Guangzhou 510640, Guangdong, China;2. School of Management, Guangdong University of Technology, Guangzhou 510520 , Guangdong, China
  • Received:2020-01-13 Revised:2020-03-17 Online:2021-04-25 Published:2021-04-01
  • Contact: 钟小丽(1975-),女,博士,教授,主要从事心理声学和声信号处理研究。 E-mail:xlzhong@scut.edu.cn
  • About author:钟小丽(1975-),女,博士,教授,主要从事心理声学和声信号处理研究。
  • Supported by:
    Supported by the Natural Science Foundation of Guangdong Province(2021A151501187)

Abstract: The spatial distribution of the localization spectral characteristics at various distances was studied by using the spectral deviation algorithm and the correlation algorithm. The best-match angles at different distances were obtained, and a subjective localization experiment was further carried out to validate the audio effects. Calculation results show that, the spatial distortion of the localization spectral characteristics occurs at different distances, in which the distortion degree increases for decreasing distance and large deviation angle relative to the lateral direction.The further localization experiment shows that the calculated spatial distortions of the localization spectral characteristics lead to audible distortion of sound images, especially at the front and back directions. Meanwhile, compared to the correlation algorithm, the spectral deviation algorithm is more suitable in selecting the best-match angles for different distances in view of localization confusion rate as well as localization accuracy.

Key words: binaural hearing, sound source localization, head-related transfer function, spatial distortion

CLC Number: