This paper deals with recognizing the species of the bird from the sound signal recorded from the bird’s environment. First, the actual part of the signal input that corresponds to the bird sound is selected and the remaining part of the signal is assumed to be silence and removed. The resulting signal contains the bird sound from start to end along with some noise from the surroundings. These noises are eliminated and the features of the actual sound are extracted by using Fast Fourier Transforms(FFT). The features are used to form a model that is compared with already existing models for various bird sounds. The model that matches with the highest probability is chosen and the bird species is determined from it.