Published Paper Details:

In this paper, techniques for identifying birds have been researched and an automatic system for recognizing their species has been devised. Significant study on taxonomy and other subfields of ornithology has been a challenging and difficult attempt for automatic identification of bird sounds without physical interaction. A two-stage identification approach is used in this work. The first step was creating an ideal dataset that included all of the bird species' sound recordings. The sound snippets were then put through a variety of sound pre-processing procedures, including pre-emphasis, framing, silence removal, and reconstruction. For each reconstructed sound clip, spectrograms were produced. In the subsequent stage, a neural network was set up and given the spectrograms as input. Using the input characteristics, the sound sample is categorized by Convolutional Neural Network (CNN), which also determines the type of bird. For the system mentioned above, a real-time implementation model was also created and put into practice.