Аннотация:The researches in the field of recognition of images testify, that there is a significant divergence between results in spheres of recognition of visual and acoustic images. It is no coincidence that at development of a number of expert systems functioning on the basis of oral speech input, the researchers address to the information incorporated in the images of spectrograms, joining it further with certain acoustic keys. This paper describes the concept of receipt of useful phoneme-acoustic information on the basis of nature of visual images (and its parts) configuration by means of associative connections specially for similarity of local-time character.