Machine Learning Algorithms for Automatic Birdcall Recognition

Graduate Research School For Candidates Prospective Candidates Available Projects Machine Learning Algorithms for Automatic Birdcall Recognition

Machine Learning Algorithms for Automatic Birdcall Recognition

Title of Project

Machine Learning Algorithms for Automatic Birdcall Recognition

Advisor/s

Dr Mangalam Sankupellay

College or Research Centre

College of Business, Law & Governance

Summary of Project

Rapid advances in recording and computing technology have made it possible to leave unattended acoustic sensors in exposed locations for months, of continuous recording. It is impossible to listen to all that is collected (In the Eco-Acoustics Research Lab alone, there is about 32 years of acoustic recording). Therefore, much research has been devoted to automated and semi-automated methods of acoustic analysis, particularly in automated recognition of birdcalls, as birds are a good barometer of a habitat's health. However, automated birdcall recognition is a difficult task because the content of environmental recordings is unconstrained. One of the main issues is the lack of a large number of appropriately tagged sample birdcalls to be used as training data for automated birdcall recognition techniques. This is because tagging birdcalls in audio recording is expensive and time consuming. The objective of this research is to investigate algorithms to detect birdcalls.

Key Words

Machine learning; deep learning; artificial intelligence; acoustics; big data

Would suit an applicant who

Is interested machine learning, big data, acoustics

Updated: 1 year ago