Koel

Meet the team

Team members: Jackson Ng (Data Science), Ritwik Jha (Data Science), Ervin Cheng (Software Engineer), Nazrul Syahmi (Software Engineer), Indraneel Paranjape (Software Engineer)

Problem statement

Due to climate change and urbanisation, Singapore’s nature areas are under threat of adverse changes and deforestation. Without proper intervention, animals face habitat loss, forced local relocation (possibly into less-than-ideal areas), migration, or death. Singapore is home to over 500 native species of birds and many unique species such as the Malayan flying lemur, and acts as a pitstop for migratory species like the Common Greenshank. Thus, it is essential for us to preserve our biodiversity and natural heritage in the face of our rapid urbanization and climate change.

When Government organisations and other enterprises work on projects involving nature, they perform an Environment Impact Analysis (EIA) to mitigate harm to the environment, and this involves the identification of wildlife species in a given area. However, data processing is a major hurdle in this process. While data collection using audio recorders is the preferred cost-effective method, there is a severe lack of tools performing species identification, especially for species common to South-East Asia. Without such tools, audio data collection and processing becomes very laborious and ineffective, potentially reducing EIAs accuracy and leading to construction decisions that are less informed and may risk a net harm to our biodiversity. This is the problem we address with Koel.

Proposed solution

To create an application that automates animal identification from bioacoustic data. Nature groups, academic researchers and government employees will be able to submit audio files to be batch processed to receive insights on biodiversity data in the region.

Number and types of species from all submissions will be made available on a geospatial visualization tool to enable biodiversity tracking and environmental impact assessments without labourious manual processing.

This tool improves the accuracy and efficiency of EIAs (Environmental Impact Assessments), and helps reduce uninformed decision-making that may introduce a net harm to our natural environment.