飞行模拟器音效系统声源提取方法

Methods of sound source extraction for audio system of flight simulator

飞行模拟器是训练飞行员的必须装备,音效系统作为飞行模拟器的重要组成部分,直接影响飞行仿真的逼真度和沉浸感。声源采集作为音效系统开发的第一步,已成为整个开发过程中最繁琐的一环。基于这一背景,寻求在低成本录音情况下高效的声源提取方法。提出基于短时傅里叶变换声源提取方法,指出其优缺点并通过实例来提取声源。针对其缺点,提出频谱建模合成技术,简述该技术的实现步骤和基本理论,指出该技术虽然实现声源的真正分离,可以直接用来提取声源,但理论复杂且很多实现步骤需要复杂的信号处理算法。因此,依据处理对象飞机发动机声音是谐音与噪声这一事实,提出简化的频谱建模合成技术并使用该技术从飞机发动机起动原始录音中提取出发动机起动生成的谐音与噪声,证明该方法的可行性。提出的声源提取方法不仅适用于飞行模拟器,还适用于其他各种模拟器音效系统的开发,有很大的工程应用价值。

The flight simulator is a necessary device for training pilots. As a key component of flight simulator, audio system has direct impact on fidelity and immersion of flight simulation. Sound source extraction is the preparation step for audio system development and has become the hardest part in the development process. Based on this background, efficient sound source extraction methods were pursued in low-cost recording condition. The method of sound source extraction based on short-time Fourier transform was used. Shortcomings and advantages of the method were pointed out. To solve these weak points, spectral modeling synthesis technology was used, and the basic theory and implementing steps of the technology were discussed. Although real separation of sound sources can be realized by using this method, the theory is complicated and Many steps need to be implemented by using complex signal processing algorithms. Therefore, according to the fact that aircraft engine sound sources are harmonic sound plus noise, simplified spectral modeling synthesis technology is used to extract engine start harmonic sound and noise. The final result shows that this method is valid. This method of sound source extraction can not only be applied to flight simulator, but also to all audio system developments of other simulators. Hence it has actual using value in engineering.