Intelligent Processing of Voice Messages in Civil Aviation: Message Recognition and the Emotional State of the Speaker Analysis

The text deals with the problems of recognition of speech messages pronounced in the prescribed form in accordance with aviation regulations. In addition, during the analysis of the phrase of an aircraft crew member, a forecast is made about the emotional state of such a person. The basis is the frequency representation of signals associated with the known time Fourier transform. Algorithms based on correlation analysis, neural networks with backpropagation of errors and Gaussian mixture models are compared. The correlation algorithm provides the best results when recognizing a specific phrase pattern (reaching 94% with a signal-to-noise ratio of 2), however, it is inferior to the neural network algorithm and the algorithm based on the Gaussian mixture model when predicting the emotional state of the speaker.