| 摘要 |
<p>PURPOSE:To improve the recognition rate of equality on utterance related to a specific talker. CONSTITUTION:A voice signal outputted from a voice input/output part 10 at a register mode time is registered as registered voice information, and the voice signal outputted from the voice input/output part 10 at a recognition mode time is compared with the registered voice information as unknown voice signal, and when the equality of them is decided as personal information, by comparing the registered voice information in which a section where no amplitude is changed much in a voice waveform though the time is changed is absorbed with unknown voice information, the fluctuation in the utterance speed of the same person is absorbed in a unite of single tone, and thus, the recognition rate is improved.</p> |
