在最近的研究中，笔者等提出了产生前方垂直方向虚拟源的扬声器信号馈给法，称为“局域Ambisonics”信号馈给法，并已证明该方法对理想倾听位置，可以产生期望的垂直定位动态因素，较传统的分立—对信号馈给法有较大的改进。在实际倾听过程中，听者的头部平移会导致双耳声压及相应的定位因素的变化，从而影响合成虚拟源的准确性和稳定性。本文将进一步分析局域Ambisonics信号馈给法中，头部移动对合成定位因素的影响，并用心理声学实验验证该信号馈给法的稳定性。结果表明，对于扬声器布置范围内的目标声源，当听者偏离了理想倾听位置，但在侧向移动不大于0.20m的合理范围内，重放仍能产生合适的定位因素，并得到平均误差基本少于5°的合理的虚拟源定位效果。因而局域Ambisonics信号馈给法对头部平移具有一定的稳定性。

In one of our recent work, a signal panning method called “local Ambisonics” for creating vertical virtual source was proposed. It was proved that the local Ambisonics panning can create desired dynamic cue for vertical localization for a subject at the sweet point, and thus local Ambisonics panning is better than the traditional pair wise panning method. In actual listening, however, the translation of subjects head may change the binaural sound pressures and the corresponding localization cues, consequently influence the accuracy and stability of the synthesized virtual sources. In this paper, the influence of subjects head translation on the summing localization cues in local Ambisonics panning was analyzed, and a psychoacoustic experiment on the stability of virtual source localization was conducted. The results indicate that, for the target sound source within the loudspeakers arrangement and an appropriate content of head translation less than 0.20m, the local Ambisonics panning is still able to create appropriate localization cues, and the subjects still can perceive virtual sources in reasonable locations with mean error mostly less than 5°. Hence, local Ambisonics panning exhibits better stability against subjects head translation.