Experimental Study of Applying Algorithms Based on Neural Networks to Enhance the Efficiency of Telecommunication Systems

The research is carried out in order to develop algorithms for the software of devices used in telecommunication systems of dispatch and operational command communication. The use of modern methods and algorithms for information processing using neural networks can significantly improve such characteristics as speech intelligibility and ease of use. The main purpose of the study is to identify the design features of speech recognition units based on single-board computers for further implementation in loudspeaker communication systems. The results of calculating the metrics of the speech recognition library under study, as well as for a number of data smoothing algorithms, are presented, and their effectiveness is evaluated on single-board computers. The results of the study are planned to be used to modernize existing and develop new devices for telecommunication systems of dispatching and operational command communication.