?
Integral Robot Technologies and Speech Behavior
The monograph presents papers on the subject domain “Integral robot. Speech behavior”. These cover issues of a theoretical nature, including representation and processing of speech information in the human mind in the process of both text analysis and text generation, and specifically the need to use jointly working linguistic and extralinguistic models of the world, the use of temporal summation of signals along with spatial summation for information processing in natural neural networks. The use of fuzzy representations for decision making is also considered, as well as some issues of the current state of speech analysis and synthesis and related areas: image analysis and speech dialogue. These problems are addressed in terms of using both rules and artificial neural networks of various paradigms. A few relevant (based on neural networks with temporal summation of signals) implementations of these theoretical concepts are presented: software technology for automatic semantic analysis of texts TextAnalyst, which uses the mechanism of temporal summation of signals in text analysis, as well as a microelectronic implementation of associative filters, where temporal summation of signals takes place. Material is also presented on a promising microelectronic implementation of synaptic connections based on memristive structures, which will ensure the creation of an effective neurochip based on neurons with temporary summation of signals in the future. The materials considered provide an extensive review of the integral problem of the implementation of speech behavior not only in terms of existing technical solutions, but also from the perspective of information processing in the human mind, which makes it possible to deepen the understanding of the mechanisms of information processing in the human brain.
