作　　者： (）;

机构地区： 华南理工大学物理与光电学院

出　　处： 《华南理工大学学报（自然科学版）》 2018年第8期130-133,150,共5页

摘　　要： 头相关传输函数表征了声源到耳的传输特性,在虚拟声重放及其相关领域具有广泛的应用.针对头相关传输函数的多维特征,提出了基于独立成分分析的中垂面谱特征的提取和建模方法,并采用主观辨别实验验证了模型的听觉有效性.计算结果表明,中垂面的平均谱失真随着展开阶数的增加呈逐渐下降的趋势,前6阶独立成分的计权组合可有效表征中垂面谱的主要特征.实验结果表明,在±45°的仰角范围内,文中提出的中垂面谱模型不会引起听觉感知的畸变. Head-related transfer functions （HRTFs） , which represent the sound transmission process from sound sources to human ears, are widely used in virtual sound reproduction and relevant fields. In order to utilize the nmhiple dimensionalities of HIITFs, an extraction and modeling method of median-plane spectral characteristics is proposed on the basis of independent component analysis （ICA）. Moreover, a subjective discrimination experi- ment is carried out to validate the applicability of the proposed model. Calculation resuhs show that the mean value of median-plane spectral distortion decreases gradually with the rise of expansion order, and a weighted combina- tion of the preceding six-expansion components represents the main part of median-plane spectral characteristics. Furthelanore, experimental resuhs show that no auditory degradation is perceived when the proposed ICA-based median-plane spectral model is employed at the elevations from 45. to -45.

关 键 词： 头相关传输函数 独立成分分析 中垂面谱 声重放

领　　域： []