Sound Signal Analyses of Cockpit Voice Recorder Based on Cockpit Sound Decoding System

摘要

Cockpit voice recorder (CVR) play an indispensable role in airplane accident investigation. However, CVR signals are complex, some valuable signals ,especially background sound signals are sometimes difficult to be identified. To solve the problems, firstly, CVR signals are systematically classified into speech signals, non-speech signals and the electro-magnetic fields & triboelectric effect to the CVR; voice print methodology and sound spectrum analyses methods are also described. Then, A kind of cockpit sound decoding system(CSDS) is put forward according to computer audio principle; CSDS’ function principle flow chart is shown. Next, as an example of real CVR signals, differentiating & hearing analyses are accomplished through making using of CSDS, some typical signals are picked off from the real CVR signals, which of the typical signals include landing gear extension, switch, wheel landing and voices of pilot. Finally, through signals anatomizing ,noise reduction techniques and sound spectrum analyses, valuable analysis results are obtained, these results are meanings, exact time, maximum frequency and energy amplitude. All the results are validated by laboratory and provide a better approach for civil aviation experts to comprehend the cockpit sound signals characteristics and reveal wreckage airplane.