The article discusses the methodology of system analysis in speech recognition for autonomous information systems with a voice interface. We proposed the novel computationally efficient algorithm of decision-making and speech processing with fuzzy phonetic decoding based on the theoretic information approach. In this algorithm the relationships between different phonemes are used to improve the recognition accuracy. This algorithm is implemented in information decision support system with speech processing. Its functional scheme is described. In the initial stage we perform an initial learning of the operator to pronounce each phoneme stably. We experimentally compare our algorithm in isolated words recognition problem for Russian language with the known hidden Markov models and deep neural networks. It was demonstrated that the proposed approach makes it possible to achieve both high accuracy and low runtime and space complexities required to store an acoustic model.
In this paper we study the temperature regimes of power LEDs for use in various types of light-emitting devices. It is shown that the use of thermoelectric coolingbatteries for high-power LEDs are not very effective. The most significant contribution to LED cooling is achieved by increasing the size of the radiator. The results of this study is needed for the subsequent mathematical modeling and control systems operating modes of light-emitting systems