查看文章

In this study, combination of acoustic and temporal features has been used for multiclass classification of bird species, since it become an immense requirement for eco environmental sounds in the wildlife. Dataset used is ‘British Bird Song Dataset’ consists of 43 species and 103 samples. Acoustic (Minimum/maximum pitch, intensity and first, second, third and fourth formants and bandwidths, and average of pulse) and temporal features (Mel-frequency cepstral coefficient) have been analyzed. The dimension reduction technique has been used to significantly reduce the dimensionality issue. Three classifiers are used in this study and they are support vector machine, nearest neighbor, and decision tree. The extracted results of classification accuracy are in the range 32–68%. Support vector machine has produced promising results in the range 64–68%.