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。 Flight simulator is a necessary training equipment for pilots. The sound system as an important part of flight simulator directly affects the fidelity and immersion of flight simulation. As the first step in sound system development, sound source acquisition has become the most tedious part of the entire development process. Based on this background, an efficient sound source extraction method is sought for low cost recording. The method of extracting sound source based on short-time Fourier transform is proposed, its advantages and disadvantages are pointed out, and the sound source is extracted through examples. In view of its shortcomings, this paper presents a spectrum modeling and synthesis technique, briefly describes its implementation steps and basic theory, and points out that this technique can be directly used to extract the sound source although it realizes the separation of the sound source. However, the theory is complex and many steps need to be complicated Signal processing algorithm. Therefore, based on the fact that the engine sound of the aircraft is homophonic and noisy, a simplified spectrum modeling and synthesis technique is proposed and used to extract the homophonic and noise generated by the engine starting from the original recording of aircraft engine start-up, which proves the feasibility of the method Sex. The proposed sound source extraction method is not only suitable for flight simulator, but also suitable for the development of other simulator sound system, which has great value in engineering application.