基于动态机制的虚拟声环境构造系统的仿真

Simulation system to construct virtual acoustic environment based on dynamic mechanisms

为了获得与临场相似的模拟实际声场的听觉感受,提出了一种基于动态机制的虚拟声环境构造系统实现方法.该方法通过频域插值,获取任意位置的房间脉冲响应,利用用户运动位置的前后相关性,实现运动预测机制.采用四叉树的层次数据结构来表示物体运动.结合与头相关的传递函数(HRTF)渐入渐出和神经网络,以多层前馈神经网络作为手段,求取了任意基于动态机制的室内声场HRTF数据.仿真结果表明,该方法较好地解决了运动声源HRTF数据的动态产生和前后声信号段平滑过渡问题,具有更好的真实性和精确度,实现了基于动态机制的室内声学仿真方法.

To obtain the hearing feeling on the virtual sound field similar to the practical sound field, a realization method of constructing virtual acoustic environment based on dynamic mechanisms was proposed. Room impulse response of random position was obtained through frequency domain interpolation in the method, and motion predicting mechanism was realized utilizing the fore-and-aft relativity of the user ＇s motion position. The object motion was represented using hierarchical data structure of quad tree. After the gradually coming in-out with head related transfer function （HRTF） and neural network was mutually combined, any HRTF data in the indoors sound-field based on dynamic mechanisms was access adopting multi-layer feed-forward neural network. Simulation results show that the method solves dynamical product of HRTF data based on moving sound sources and the smooth transition of the fore-and-aft sound sequence signals. The method is more reliable and accurate than other methods, and realizes indoor acoustic simulation based on dynamic mechanism.