关键词: 声源定位, 时延估计, 互功率谱相位

Abstract: Aiming at low SNR of delay estimation of traditional Cross-power Spectrum Phase （CSP） method in the strong noise environment, an improved CSP method to solve this problem is proposed in this paper. The traditional method of CSP is studied and the speech signal energy on the ratio of total energy is analyzed. In order that the speech signal is not drowned in the strong noise environment, a nonlinear parameter changing with Signal to Noise Ratio（SNR） is defined. By adjusting the non-linear parameter to control the size of weighting function, thereby reducing the influence of noise, the anti-noise performance of the algorithm is improved. Three methods of time-delay estimation are compared on matlab platform by using eight linear array microphone voice signal acquisition in different SNR environments, and simulation results show that the improved CSP method has better positioning performance than traditional CSP and SCOT method in the strong noise environment.

Key words: sound localization, time-delay estimation, cross-power spectrum phase