| 摘要 |
PROBLEM TO BE SOLVED: To provide a device and a program for utterance section detection to detect an utterance section at high speed and with high accuracy. SOLUTION: The device for utterance section detection to detect the utterance section from the input voice comprises: a sound analysis means for converting an input voice to a sound feature amount; a continuous voice recognition means for serially calculating cumulative likelihood in each sub-word while synchronizing with the input voice, by using the sound feature amount obtained by the sound analysis means, and a sub-word network composed of a sound model and a language model which are established beforehand; and an utterance section detecting means for serially detecting an utterance start point and an utterance end point in the input voice from the cumulative likelihood in each sub-word. COPYRIGHT: (C)2007,JPO&INPIT |
