Comparative Analyses and Experiment Verification on Cockpit Background Sound' Characteristic Frequency

Because the characteristic frequencies of cockpit background sound recorded by cockpit voice recorder (CVR) is the key evidence in investigating accident causes for wrecked airplane. And it is crucial for investigator to verify the characteristic frequencies through different methods. To obtain exact characteristic frequencies for cockpit background sound, systematic research are made. Firstly,the CVR signals is classified into speech signals, non-speech signals; then, cockpit voice decoding system (CVDS) is developed according to the audio principles. Through the CVDS,the cockpit background sounds are differentiated & heard, and as an example of a background sound signal, an overspread warning signal initial spectrum characteristics are obtained. At the same time,to get more exact spectrum characteristics of the signal, three algorithm methods (wavelet transform-WT, chirp Z- transform-CZT, correlation analyses-PSD) are applied to the signal respectively, their characteristics frequency (maximum frequency) are acquired and almost identical. Finally, to prove the algorithm results, the experiment verification to characteristic frequency of over speed warning signal and other three kinds cockpit background sounds from airplane cockpits are checked out and analyzed in whole anechoic room with the aid of LMS SC305 instrument and its software. Research indicates that the experiment characteristic frequency spectrums of the cockpit background sounds are identical with the three algorithm methods. The concludes of the paper provides better approaches for civil aviation experts to comprehend the cockpit sound signals characteristics and reveal wreckage aircraft causes.