| 摘要 |
PURPOSE:To execute voice recognition in a short period and with high accuracy by discriminating a keyword waveform from an inputted voice waveform, and recognizing an effective voice from the voice waveform in the vicinity of the keyword waveform. CONSTITUTION:In a RAM 28 of a voice processing part 2, voice waveforms of 'Yen' and 'shees' are stored as keyword waveforms in advance. A sub-CPU 24 discriminates whether the keyword waveform is contained in the waveform inputted from a microphone 20 and stored in a voice waveform storage part 29 or not. When it is contained, the waveform immediately before its waveform is recognized as an effective voice waveform by a voice recognizing part 25. In such a way, a noise waveform can be eliminated, and the voice recognition of high accuracy can be executed in a short period. |
