According to the saturated diving
helium speech interpretation method and
system based on multi-objective optimization, in at least one
system comprising a diver and a filter, a diver working language phonetic symbol
library and a common working language word
library are established, and the diver reads to generate a phonetic symbol standard speech
library, a phonetic symbol
helium speech library and a common working word speech library; and designing an
impulse response coefficient of the filter by using a multi-objective optimization
algorithm, correcting and interpreting the
helium voice
signal after tag sampling word by word, and continuously updating the
impulse response coefficient to complete the complete interpretation of the helium voice. According to the method, individual phonetic symbol voice library characteristics of a diver in different environments are set, and
system parameters of a filter are designed by adopting a multi-objective optimization
algorithm and are continuously updated and corrected, so that the dependence on the number of
machine learning samples is eliminated, saturated diving helium voice interpretation is stabilized, the performance of a helium voice
interpreter is always in an optimal state, and the reliability of the system is improved. The self-adaptability of the helium voice
interpreter at different diving depths is improved, and complete interpretation of helium voice is realized.