通常实验测量只能得到有限个离散空间位置的头相关传输函数( HRTF) ，并需要通过空间插值得到未测位置的 HRTF．常用的插值方法包括相邻线性插值、全局插值、双线性插值和三角插值．在 HRTF 的实际测量过程中，受试者的轻微头动会引入测量误差，从而影响到空间插值的准确性和稳定性．文中从头部偏移和头部转动带来的测量误差入手，以计算信号失真比( SDR) 作为插值误差的量度，分析和对比头动对以上4 种空间插值效果的影响．结果显示，头动前后，全局插值优于其余 3 种插值方法，且在一定的 HRTF 频带范围和头动程度内保持了较好的插值效果．由于 HRTF 空间插值方法和多通路声重放中信号的馈给方式相似，因此文中还利用所得方法和结果，在低于1．5 kHz 的频率范围，分析了水平面均匀布置的 6 或 8 扬声器重放合成声像的稳定性．

The head- related transfer functions (HRTFs) which can be measured through experiments,usually dis-tribute in discrete directions,and the other HRTFs in unmeasured directions,have to be estimated by means of spatial interpolation.The commonly- used interpolation methods include the adjacent linear interpolation,the global interpolation,the bilinear interpolation,and the spherical- triangular interpolation.In practice,potential subjects' head movements in the HRTF measurement may result in error,thus influencing the accuracy and stability of the spatial interpolation.In this paper,the errors caused by slight head movements in the HRTF measurement are eva-luated,and the effects of the head movements on the above four kinds of spatial interpolation are compared by calcu-lating the signal- to- distortion ratio.The results indicate that,the global interpolation outperforms the others before and after the head movements,and is of excellent performance in a certain frequency band range of HRTF and with appropriate head movements.Due to the similarity of the HRTF spatial interpolation to the signal panning or mixing methods for the multi- channel sound,the proposed methods and the corresponding results are extended to analyze the stability of the sound image reproduced from the sound field signal mixing with 6 or 8 horizontal uniformly-arranged loudspeakers below 1.5kHz.