中文摘要：

Audio-only游戏中，游戏者的所有操作只能依靠虚拟声所包含的空间信息实现声像定位，因此对虚拟声的准确性和实时性有较高的要求．本研究基于HRTF的虚拟声基本原理，比较了HRTF的个性化匹配和压缩方法，并利用PCA技术对CI—PIC和MIT实验室的HRTF数据进行压缩与重建处理，实验证明该技术具有计算快速简单和处理效果好的优点，同时压缩重建前后的HRTF数据具有较小误差，既不影响虚拟声重发的准确性，又减少了HRTF的存储空间，能有效降低HRTF合成的计算复杂度，提高虚拟声应用的实时性．

英文摘要：

In Audio-only games, the players can only control the game process and operation based on the orientation information of the sound source in three dimensional virtual sound space, and so real time processing with high accuracy is needed in sound reconstruction. In the paper, based on the basic principle of the HRTF-based three dimensional virtual sounds, the individualized matching and compression method was analyzed and validated through the process progress of the CIPIC and MIT HRTF data using PCA method. The small error between the original and the compress-reconstruct HRTF data proved that the proposed method can be applied in real-time processing system with less storage space and simplified computational complexity to enhance the reconstruction speed and accuracy, which is favorable for the practical application of virtual sound.