Automatic Classification of Flight Calls of Crossbill Species (Loxia sp

A new method for automatic classification of bird songs is proposed. This method is based on tracking of the first harmonic components of the spectrogram. From the tracking data some features were selected. The final classification was then performed using a neural network. As a test case we use a set of flight call recordings for crossbills that were publicly available. The preliminary results indicate that we can classify correctly well over 70% of the crossbills using their flight calls