Phonetic Typewriter

Abstract

We have constructed a research model of phonetic typewriter named Sonotype since 1960. This equipment was designed based upon our research results of these several years, aiming at both universal and flexible ability in the processing of speech sound. The basic equipment was a device for one hundred Japanese monosyllables in which about 3000 transistors and about 5000 diodes were used. For the purpose of treatment of Japanese conversational speech, some parts of control system have been added considering the data obtained from the basic equipment; however, the experimental result has not yet been obtained. This equipment has several peculiarities together with some features similar to those reported in other papers. (a) All the control pulses are derived from the input speech sound wave itself. (b) Distinctive features of speech sound are obtained by using not only the results of ordinary spectrum analysis but also various kinds of features necessary for the recognition of phonemes: for instance, voiced, unvoiced, fricative, nasal, contracted, and so on. (c) Discrimination is made by digital technique such as order pulse, AND gate, OR gate and binary circuit. Of course it is necessary to convert analog quantity to digital one and vice versa.