| 摘要 |
PURPOSE:To simplify an easy changing of an voice input into a sentence, by deciding the break of the words or sentences in case the time between the supplied phonemes becomes longer than a fixed time and securing an automatic insertion of the punctuation marks. CONSTITUTION:The voice input applied to the frequency component analysis circuit 1 receives a full rectification after passing through the voice BPF as well as a BPF that analyzes the input into each frequency component to be applied to each logarithm compressor 6. The output of the compressor 6 is applied to the primary feature extracting circuit 7 where the local top or valley of the spectrum is detected by the threshold logic. The output of the circuit 7 is applied to the sound element feature extracting circuit 8 to presume the sound element. The sound element is applied successively to the logic recognition circuit 9 and the sequence logic recognition circuit 10. Thus the word corresponding to the input voice is decided and delivered. The full wave rectification output of the circuit 1 is applied to the counter device 11, and the time corresponding to the counter value is compared with the time set previously. When the value of comparison becomes larger than the set value, the possibility of a period or punctuation mark is decided. Then a period or punctuation mark checked by a prescribed dictionary is inserted through the circuit 9. |
