This paper explores a method for automatic speaker-independent recognition of human psycho-emotional states by analyzing the speech signal based on Deep Learning technology. The solution to the problem of creating a model capable of performing a multi-class classification of seven human psycho-emotional states (joy, fear, anger, sadness, disgust, surprise and neutral state) based on selected informative features of audio recordings in the form of melspectrograms and mel-frequency cepstral coefficients is described. These informative features are used to train two deep convolutional neural networks on the generated data set. The classifier model based on the developed neural networks is capable of classifying a speech signal, determining the probability of its belonging to each of the seven specified emotions. As a result of training on a deferred testing subsample, the proportion of correct recognition answers is accuracy = 0.93. The purpose of this research is to enhance flight safety and comfort by using deep learning technology. Recognizing the emotions of pilots and crew members can aid in identifying states of stress, fatigue, or even panic, enabling prompt responses to potentially hazardous situations and averting incidents and accidents. Overall, the method of automatic speaker-independent emotion recognition from speech signals using deep learning technology represents a promising research direction in aviation, one that could contribute to enhancing safety, comfort, and the efficiency of air transportation.