| 摘要 |
PURPOSE: To receive a freely prepared text and convert it into an easyto-see speech which has correct accent and a correct pause by interpreting an abbreviated word and correcting an error, and translating special terms used by the speech handicapped in common. CONSTITUTION: The text is inputted to a word detector 11, which discriminates a word or term. The detected term is inputted to word stores 12-14 which are connected in series in a shift register style. A processor 15 functions to analyze and correct the inputted term and also embed accent and a pause at positions which are correct in terms of syntax, and applies the corrected term to a synthesizer in proper timing. The rule which discriminates and alters the term gives effects to the recognition of abbreviated words, the recognition of special terms used in a community that a user participates in, the recognition of misspelling, mistyping, and noise, etc. Therefore, the text generated by the voice handicapped person can be repeated and served without any attendant. |
