著者
高村 聡 松本 和教 関口 芳廣
出版者
山梨大学
雑誌
山梨大學工學部研究報告 (ISSN:05131871)
巻号頁・発行日
vol.42, pp.44-50, 1991-12

We are developing a wild birds twitter-identification system which can work in noisy enviroments. In the natural world, there are several types of noises, for example, the sound of stream, the sound of leaves swayed by the wind and twitters of other birds. Then we can't use specific features of the twitters of birds to identify them. In this system, the number of zerocrossing and the mean value of overall spectrum of twitters of birds are used to determine the kind of birds. This system can successfully identify 34 twitters out of 35 twitters which are twittered by 7 kinds of birds and recorded in natural world.