Grey-Headed Flying Fox Vocalisation Machine Listening

This project employs Machine Learning to listen to Grey-Headed flying fox vocalisations, and takes the liberty of interpreting these into corresponding human-perceptible concepts in real-time based on known research into communication methods of these species. Grey-headed flying foxes are known to science to have up to 20 vocalisations with distinct meanings within five vocalisation ‘types’ (Christesen & Nelson, 2000). Of these 20 vocalisations, seven have so far been accurately learnt and translated for this project. This project follows a practice-based audio machine learning research design model, from audio capture through to classification with the help of bat carers and vocalisation experts, ending in a trained model free for public use. The public can take this project and, with the use of a microphone, decode flying fox vocalisations that it hears in real-time. While this tensorflow model has undergone a rigorous training process, I question how these learnt classifications are translated, and whether creative practice can be helpful in designing ‘creative interpretations’. I attempt this through poetic narrative and visual design, where each accurately heard vocalisation is turned into on-screen poetic animated text that aims to evoke the meaning behind each communication, such as a mating vocal, territorial vocal, a growl. This gives me the artistic license to interpret known vocalisations through creative practice. This project’s knowledge contributions include a dataset of categorised flying fox vocalisations, a trained tensorflow machine for listening to and identifying vocalisations, and a final artwork that interprets this data into digital poetry in real-time. Further research will be undertaken into the nature of creative machine listening for animal vocalisation interpretation.

Website

‘The Bat Translator’ (2021-), Alinta Krauth. Image courtesy of the researcher.

Bat Translator

EVENTS & EXHIBITIONS

Sites of Connection

13 Aug- 30 October 2021 | About the event

Location | USC Art Gallery

____

Created by and part of PhD by Alinta Krauth

Principal Supervisor – Prof Donna Hancox, Associate Supervisor A/Prof Jen Seevinck