| 摘要 |
PURPOSE:To provide a voice recognition device whose reduction of recognition rate is small even under the environment of a large noise and capable of being used in various environments. CONSTITUTION:This device is provided with means 102 dividing a voice into plural frequency bands having different center frequencies, means 103 obtaining the degree of correlativity included in respective signal wave forms obtained by division, means 104 detecting power components of respective signal wave forms obtained by the means 102 dividing the voice into plural frequency bands, means 105 normalizing synchronization frequency components included in respective divided signal wave forms using the power components and a means 106 adding the normalized synchronization signal components to the respective signal wave forms. |
