ABSTRACT
Conservation of biodiversity is important for the existence of all living beings. Birds are an integral part of biodiversity. Monitoring a region for the presence of bird species is an important task required for the conservation of biodiversity. In the past 30 years, researchers have used signal processing and pattern recognition techniques to recognize bird species in a vicinity. This has helped the environmentalists to study species distribution, conservation planning and biodiversity of a region. Recently, deep neural networks have become increasingly popular in classification. This chapter encapsulates classification and identification of bird species through their sounds using convolution neural network (CNN). Two different approaches are considered for the same CNN architecture to perform this task. The first approach is to extract mel spectrograms and the other one is to extract mel frequency cepstral coefficients (MFCC) plots from the audio files. These two different inputs are provided to the CNN model, which in turn gave impressive results.
