Dataset for Manx shearwater (Puffins puffins) individual acoustic identification

This data archive is for the Imperial College London MRes thesis "Predicting Manx Shearwater Individual Number in Acoustic Recordings Using Machine Learning" by Zhanyi Lin in 2026. The Manx shearwater (Puffinus puffinus) is a seabird species with individual signatures in its vocalisations. The aim of this study is to develop a pipeline that can identify individual Manx shearwaters by their calls, as well as predict the number of Manx shearwater individuals in acoustic recordings. This is important for the conservation of Manx shearwaters and has the potential to be expanded to other seabird species. The data included in this archive comprise the training and test set embeddings generated by BirdNET, which serve as the input to the core pipeline. These embeddings are 1024-dimensional acoustic feature vectors extracted from 2.6–3-second clips detected by BirdNET. This dataset also includes checkpoint files for the core 1D CNN model. The code for this model can be found at "https://github.com/Antongsky/Shearwater_individual_recognition". In addition, this dataset contains the data used for model optimization and parameter adjustment.