You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A scalable unsupervised deep learning approach extracts behavioural and demographic insights from forest elephant vocalisations
The analysis in this paper can be partially recreated by running the two notebooks saved within the main repo.
'call_types_notebook.ipynb' - performs the call-type analysis for question 1.
'rumble_subtypes_behaviour_notebook.ipynb' - performs the rumble subtype analysis (question 2) and the behavioural and demographic analysis for question 3.
The notebooks load the audio files stored with 'audio_dir', the annotations for these audio files stored in 'data' and the python scripts in 'elephant scripts'.
7 sample audio files have been uploaded for demonstration purposes.