摘要 |
<p><P>PROBLEM TO BE SOLVED: To suppress a decrease in speech recognition rate of a speech interaction device. <P>SOLUTION: The speech interaction device 100 is provided with a sound output part 3 for outputting a sound to the outside as an output sound and a sound input part 4 for inputting an external sound as an input sound, decides whether the input sound inputted to the sound input part 4 is the output sound outputted from the sound output part 3, and performs speech recognition for the input sound only when deciding that the input sound is not the output sound. Consequently, the speech recognition is not performed for the input sound which is outputted as the output sound from the sound output part 3 and inputted to the sound input part 4 and then misrecognition is prevented to suppress a decrease in speech recognition rate. No adaptive filter is used and then an estimation error by an adaptive filter is not generated, thereby suppressing the decrease in speech recognition rate. <P>COPYRIGHT: (C)2006,JPO&NCIPI</p> |