Birdsong Recognition Based on Improved DTW

Abstract

In this study, the improved Dynamic Time Warping (DTW) algorithm is used in MATLAB to perform simulation classification and recognition simulation on the sounds of Nycticorax nycticorax and other 10 birds in the Wuzhishan Islands Reserve in Zhoushan, Zhejiang. By setting the slope of the path, the path is restricted in a fixed parallelogram. The algorithm combines with global restrictions can effectively speed up the path search, reduce the amount of calculation, reduce the memory requirements, reduce the recognition time, and thus improve the recognition efficiency. Experimental results show that compared with the traditional DTW algorithm, the improved DTW algorithm increases the recognition speed by 36.49%, the accuracy rate by 4.08%, and the recognition efficiency is greatly improved.