| 摘要 |
FIELD: sound identification means. ^ SUBSTANCE: method involves sound conversion into electric signal including recording of each sound phoneme in the form of amplitude-frequency response, measurement of frequencies typical of given phoneme of formants which are assigned "0", "+1", and "-1" depending on position relative to X-axis, presentation of electric signal in the form of amplitude-frequency response, its analysis with aid of set of narrow-band frequency filters which is divided into groups affording frequency coverage of bands of adjacent frequency filters connected to inputs of formant recorders whose outputs are connected to input of phoneme recorder; recorded set of formants is compared with predetermined set of formants, and recorded sound is determined, outputs of phoneme recorder being connected to printer. Groups of narrow-band frequency filters are selected so that part of these filters in group correspond to frequencies of one of formants of particular sound; in absence of signals from all recorders printer writes space character; sound recording level is adjusted prior to sound-to-signal conversion, and amplitudes of formants of most frequently encountered and stable pre-recorded phonemes are compared with formants of phonemes being recorded. ^ EFFECT: enhanced resolving power. ^ 1 cl, 1 dwg |
