人工影响天气作业飞机中语音增强算法的研究

Research of Speech Enhancement Algorithms of Weather Modification Aeroplane

飞机实施人工影响天气作业,具有覆盖范围广,作业效果好的优点,是未来人工影响天气工作的发展方向;但人影作业飞机内部工作环境复杂,工作人员的正常语言交流会受到飞机引擎、机舱内设备、飞机外部等噪声的影响,甚至无法正常交流;因此,需要设计一种语音增强装备或设施来消除噪声,方便人影飞机上工作人员的正常沟通交流;从理论上阐述了参数法、非参数法、统计方法及其他语音增强算法;重点研究了维纳滤波法、谱减法和基于短时对数谱幅度的MMSE算法,并对这3种语音增强算法都进行了实验仿真,对比了各种算法的优缺点;针对人影飞机的特定工作环境,提出基于短时对数谱幅度的MMSE算法比较适合用于人影飞机上进行语音增强,消除噪音。

The implementation of weather modification with aeroplane can cover a wide range and produce better results. So it is the fu ture trend of the development of weather modification. However, the internal working environment of the aeroplane is so complex that the normal communication of the staff would be affected by the noise of aircraft engines, inside equipments and external aeroplane, Sometimes they can＇ t communicate fluently. Therefore, it is need to design a speech enhancement equipment or facilities to eliminate the noise and con venient staff communication. Elaborates theoretically the parametric method, the nonparametric method, the statistical method and other speech enhancement algorithms. Focus research on Wiener filtering algorithm, spectral subtraction algorithm and MMSE algorithm based on short--time logarithm spectrum. The three algorithms are simulated and are compared. The experimental results demonstrate that the MMSE algorithm based on short--time logarithm spectrum is suitable for speech enhancement aiming at the specific circumstance.