Make Your Own Recogniser
As part of the 2025 Australian Acoustic Society Conference we are running a workshop for building automated call recognisers. The goal of this workshop is for each participant to leave with a working call recogniser for a single species-call of their choice. It covers some basic theory of convolutional neural networks, embeddings and embedding based classifiers. We are using Perch embeddings, and Ecosounds for serving audio and verifying detections.
This workshop is run by QUT ecoacoustics group with support from the Australian Research Data Commons, and from Google Deepmind, the developers of the Perch embedding model and Agile Modelling workflow.
