Global biodiversity is in crisis. Effective management of threatened biodiversity requires detailed knowledge of the population dynamics of species. Acoustic field observations of vocal animal species are an important tool for occupancy dynamics studies. Collecting animal sounds, generally involves surveying a study site in person (active surveys). However, remote or inaccessible areas, adverse climatic conditions, night-time observations and the high dependence on volunteers to survey, are major drawbacks that can easily blowout the budget. Passive surveys using automated recording devices at the survey site are starting to be used. However, these are still limited by the quality of the recording (e.g. capture background noises), the requirement of proximity to a 3G/4G network (e.g. if based on mobile technology) and the inability to easily capture additional environmental parameters specific for the survey site at the time of each survey (e.g. temperature) and key to understand the habitat condition. As a result, long-term monitoring of species is hindered by the prohibitive costs of the associated research time and the imperfection of the methods used. These issues seriously impede our ability to make reliable, informed management decisions.
In this project we carried out preliminary work to overcome the above challenges by designing an acoustic sensor setup for long term audio recording and a robust software for automatic species recognition. We carry out recordings of birds from the Mulligans Flat nature reserve in ACT and implement a software that automatically recognizes bird calls/species present in a given recording.
Project Results: https://cecs.anu.edu.au/sites/default/files/high-tech_biodiversity_rescue_final_workshop.pdf