Applied research in acoustic wildlife monitoring with AI

Sound recording is a cheap, rapid, powerful way to monitor many animal species and their interactions. Our research programme develops new AI methods for sound – directly within the applied context of acoustic wildlife monitoring.

0
PhD students
0
Countries
0
Organisations

Apply for a PhD position in Bioacoustic AI

Would you like to develop AI algorithms to help understand animal sound? To monitor biodiversity across Europe? To collaborate with AI and biodiversity experts, as part of a team and with partners in many European countries?

We’re training 10 PhD students to become full-stack bioacoustic AI experts and are now looking for PhD Doctoral Candidates. You can pre-register your interest for a position now.

Recruitment starts in August 2023

  • 10 positions available, funded by the EU

  • Positions in The Netherlands, Germany, France, Belgium, Czechia, Finland, and the UK

  • PhDs are planned to begin in early 2024

  • The funded PhDs will run for 3 to 4 years

About our research

We want to understand animals better to protect biodiversity

The biodiversity crisis is coming into focus. Yet, data for monitoring wild animal populations are still incomplete and uncertain. And there are still big gaps in our understanding of animal behaviour and interactions.

birds

bats

hyenas

insects

Animal sounds can be used to understand animal behavior

Sound recording is a cheap, rapid, powerful way to monitor many animals. Modern machine learning can dramatically improve its scale and precision.

Can AI-powered acoustic monitoring radically improve our understanding of wild animals and how to protect them?

AI can help monitor a variety of species

We develop new AI methods and new hardware-software integrations, directly within the context of acoustic wildlife monitoring. At the same time, we investigate behaviour/ecology monitoring of selected target species in their own habitats.

We combine four methods to monitor wildlife

Our “full-stack” approach to wildlife monitoring is unique: from low-level on-device processing to high-level ecological inferences.

AI and acoustic signal processing

AI methods are constantly leading to better and better recognition, but usually in a standard “supervised learning” model.

We will develop state-of-the-art AI task formulations that suit the special constraints of wildlife monitoring tasks, including the monitoring of hard-to-detect birds and mammals in European and tropical soundscapes. We will build upon innovative AI techniques such as human-in-the-loop, self-supervised and multitask learning.

Hardware devices

Wildlife sounds can be recorded using smartphones, or small dedicated recording devices (such as those made by our project partners!). But these devices usually can’t run recognition algorithms on-device, and nor can they synchronise between multiple devices to estimate the location of calling animals.

We will develop algorithms to run directly on these devices, to create a new generation of smart wildlife microphones. We will also make sure that such methods have a small ecological footprint, by minimising their power demands.

Animal vocal behaviour

Animal sounds offer us a window into the behavioural interactions among individuals and groups, and how these are shaped by the social and physical environments in which they live.

Using the AI techniques we develop, we will connect animal sounds with the social and behavioural “structure” of animals’ lives.

Ecology and environment

An important goal is to use acoustic monitoring to help protect wildlife in the era of environmental change. With increases in sensitivity and scale, automatic acoustic monitoring is now ready to become a new tool in the toolbox of governments, nature organisations, and communities to guide their policy and practices.

We will study this aspect through the deep ecological experience of our consortium. Through collaborations with partner organisations across all sectors, our PhD candidates will gain experience in real-life nature monitoring projects.

Applied research

We want to develop easy sound monitoring solutions to be used in real life

Bioacoustic monitoring with machine learning is not a new idea, but machine learning methods were not previously good enough for easy general use. Many people are searching for new ways to keep track of animals: governments, ecology organisations, and citizens.

We will develop improved methods, test the “full stack” in the wild, and finally establish Bioacoustic AI as a powerful and flexible new source for all kinds of evidence about animal life.