The systems of recognition and speech synthesis, essentially, concern to intellectual systems. They should, at first, understand speech, despite of possible interference's, inaccuracy diction, deviation from normative syntax etc., and, secondly, to provide qualitative speech having high articulation and naturalness of a voice signal. The man integrally uses as language knowledge (phonetics, lexicon, syntax, semantics, prosody etc.), and not language, i.e. knowledge of data domain of the dialogue. Main psychology-acoustic by the characteristic of speech its articulation, i.e. degree of correct perception of sounds, words and sense of speech is. The maximum articulation is characteristic of continued speech perception   phrase articulation. If the man perceives the isolated words, percent of articulation appears less. It is even more reduced at perception of the isolated phonetic units of speech such as syllables. However, the majority of the approaches is based on sequential execution of recognition of words and syntactical analysis. But the experience of the linguists shows, that the man in the beginning seizes sense of a phrase and only then begins purely the task of comprehension of speech, which frequently is considered as the procedure independent of a stage of recognition. This point of view naturally reduces in weak outcomes. From psychology speech perception becomes obvious, that the recognition and comprehension   is two tightly coupled procedures.