Abstract Our project focuses on identifying bird species by detecting audio recorded bird songs. Being able to identify bird songs can be very useful from environmental safety and research perspective. This can help monitoring birds in different habitats and keep track of endangered birds. It is often difficult to identify birds manually just by observing and hearing. An automated process can ease the job hugely.Features can be extracted from bird song using different audio treatments and then classical ma- chine learning technique can be used to match them with previously trained database of birdsongs.The experimental results compare the performance obtained in different situations, encompassing the complete audio signals. Signal processing will be the main method of training and matching them with sample data. Python provides pyaudio library for capturing audio signals. Overall the process encompasses training the machine using recorded bird songs then test it against sample audios to check precision.
Pereira et al. (Tue,) studied this question.