飞机舱音记录器非话语信号盲分离性能

飞机驾驶舱话音记录器(Cockpit voice recorder,CVR)记录的舱音信号,通常是语音声、警告声、开关按钮声和背景噪声等混合而成的.目前国内对该类信号的分析和辨别主要是计算机译码后进行人耳辨听,存在不易准确分辨出各种独立的声音信号的缺点.本文提出采用基于高效快速的独立分量分析(Efficient variant of fastI-CA,EFICA)算法和可调整权值的二阶盲分离(Weight-adjusted variant second-order blind identification,WA-SOBI)的混合算法对舱音信号进行分离实验.采用不同算法的仿真结果比较表明,混合盲处理算法具有更为优越的分离性能.

Blind Separation Performance of Non-voice Acoustic Signal for CVR

The acoustic signal recorded by cockpit voice recorder (CVR) on airplanes is a mixed signal combined by voices, switching knob, alarm sound, and noises. So far, in domestic, the analysis and the identification of these acoustic signals are mainly dependent on the human audition, which is difficult to separate independent sounds. A combined algorithm based on efficient variant of fastICA (EFICA) algorithm and weight-adjusted variant second-order blind identification (WASOBI) algorithm is proposed to separate the specific sound from the acoustic signal. Simulation results using these different algorithms are compared. Simulation results show that the combined algorithm has better performance in separating independent sounds from the acoustic signal.