Overview

• Griffith University’s team led by Katie Turlington published the peer‑reviewed method in Methods in Ecology and Evolution and released the tool publicly for free.
• Field tests in South‑East Queensland reported roughly 90% correct identification of distinct sounds with substantially less effort than manual analysis.
• The approach clusters similar signals by time, frequency and amplitude, enabling exploratory work without reference libraries or pre‑labelled data.
• The system detected biological sounds even under constant flow noise, a common obstacle in river recordings.
• The R‑based workflow is described as scalable to very small and very large datasets and is intended for broader validation across diverse ecosystems.