| 摘要 |
PROBLEM TO BE SOLVED: To provide a voice signal processing device capable of accurately detecting a target voice section regardless of amplitude of a target voice.SOLUTION: Delay subtraction processing is applied to an input voice signal to form first and second directional signals having dead angles in first and second prescribed directions, and by using the two directional signals, coherence is obtained. Coherence is compared with a target voice section determination threshold to determine whether the input voice signal is a target voice section or a non-target voice section; even when coherence is compared with hangover imparting threshold larger than the target voice section determination threshold, and the result of determination changes from the target voice section to the non-target voice section, a result of determination to be a target voice section is continued just for a prescribed hangover length. |
