扬声器组空间汇聚特性及声像感知一致的扬声器阵列精简

Three-dimensional (3D) audio and video technology for home applications has become an iconic symbol of global multimedia industry. The key is to using minimum arrangement of loudspeakers to reproduce the 3D auditory perception consistent with the original 3D acoustic audio system. However, the traditional methods ignored that the influence, caused by different loudspeaker triplets, could lead to the distortion of the reproduced sound field and the directional deviation of the phantom source perceived by binaural effect. So it is time-consuming and expensive for the traditional methods to use lots of subjective evaluations to distinguish the experience of spatial sound reproduced by the simplified loudspeaker arrangements. That is, it should study the multichannel simplification constrained by consistent directional perception of the phantom source. To solve this problem, multichannel simplification, constrained by the distortion of the reproduced acoustic field, will be carried out firstly. Three studies are included: the constraint rule between the positions of the loudspeaker triplets and the distortion of the reproduced acoustic field, the analysis of the convergence characteristics of the loudspeaker triplet constrained by the distortion of the reproduced acoustic field, and the simplification of the multichannel audio system constrained by the selection of the optimized loudspeaker triplet. Secondly, combining the Just Noticeable Difference (JND) of the sound spatial orientation and space-time correlation of the sound trajectory, consistent directional perception of the phantom sources will be carried out on the simplified 3D audio system. There are three studies: the constraint rule between the positions of the loudspeaker triplets and the directional deviation of the phantom source perceived by binaural effect, the analysis of the convergence characteristics of the loudspeaker triplet constrained by the directional perception deviation of the phantom source, and the consistent directional perception of the phantom source constrained by the selection of the optimized loudspeaker triplet. The related research is expected to make the distortion of the reproduced sound field within head be decreased by 30%, the average MOS scores of subjective evaluations of sound location be increased by 30% and try to make people get better 3D spatial sound ambience in home environment.

面向家庭应用的三维音视频技术成为全球多媒体行业标志性符号，其关键在于利用较少扬声器回放与原3D音频系统听觉感知一致的三维声。传统方法忽略替代扬声器组空间位置选取对重建声场失真及虚拟声像双耳感知定位的影响，导致通过人工测试确定最优扬声器精简阵列的工作量巨大，亟待研究感知一致约束的三维多声道音频精简技术。本项目首先研究替代扬声器组空间位置变化对声场重建失真的约束规律，分析声场重建失真约束的替代扬声器组汇聚特性，提出最优替代扬声器组选取准则，指导扬声器阵列自动精简；其次结合声音空间方位恰可感知特性及运动轨迹时空关联特性，研究替代扬声器组空间位置变化对虚拟声像双耳感知定位失真的约束规律，分析双耳感知定位失真约束的替代扬声器组汇聚特性，提出优化替代扬声器组选取准则，指导虚拟声像感知定位。本项目所提出方法预期在人头区域声场失真降低30%，虚拟声像感知定位主观测试结果提高30%，提升家庭环境3D听觉体验

传统3D音频系统能给听音者带来3D声场效果，但是扬声器数量太多，限制了其应用范围，当需要支持面向家庭环境的三维声场重建时，传统方法忽略替代扬声器组空间位置选取对重建声场失真及虚拟声像双耳感知定位的影响，重建中扬声器位置摆放主要依赖经验，并未开展扬声器位置优化方法的理论研究，导致通过人工测试确定最优扬声器精简阵列的工作量巨大，并且原始三维音频系统提供指定声音对象的声信号和对象空间位置（或运动轨迹）与虚拟声像感知不一致，空间沉浸感严重下降。.针对上述需求与挑战，本项目研究扬声器组空间布局在指定区域重建声场失真约束条件下的三维音频系统自动精简，并当原始三维音频系统提供指定声音对象的声信号和对象空间位置（或运动轨迹）时，研究虚拟声像感知一致性约束条件下的虚拟声像定位，提升听音者三维声效体验。在区域声场重建低失真的替代三扬声器组自动精简方面，建立替代三扬声器组区域声场重建失真模型，分析了替代三扬声器组空间汇聚属性，提出最优替代三扬声器组选取方法，设计了一种三维音频自动精简算法。当声道数精简到10时，我们所提处方法比最新研究成果，对应人头区域声场重建平均失真的平均值降低11.67%，有更好的重建效果。在三维空间位置（轨迹）感知一致的虚拟声像定位方面，分析替代三扬声器组空间位置变化对虚拟声像双耳感知定位失真的约束规律，提出双耳信号重构对应空间汇聚特性约束的最优替代三扬声器组选取准则，对声源实现了基于双耳感知的虚拟声像定位，所提出方法比最新研究成果的声场重建平均失真降低50.79%，主观评测平均CMOS分达到1.99分，所提出方法能更好地表达原始声源的空间方位，提升家庭环境3D听觉体验。.在本项目的实施过程中，在本领域期刊发表论文2篇和国际会议上发表论文4篇；申请发明专利7项，其中授权1项；培养博士研究生1名，硕士研究生6名；邀请两位国内专家访问交流。