PhD Thesis: Acoustic identification of individual animals
Queen Mary University of London, 2025
📍 Based in London
Postdoctoral Researcher
in Computational Bioacoustics and Animal Communication
I am currently a Postdoctoral Researcher at Earth Species Project, working in computational bioacoustics and animal communication. My research focuses on developing machine learning methods to understand how animals communicate, with applications in conservation and biodiversity monitoring.
I come from a computer science background and began working in bioacoustics by analysing beehive sounds to understand how the bee colony is doing. This work sparked my interest in applying machine learning more broadly to decode animal vocal communication across species. I received my Ph.D. from Queen Mary University of London in 2025, where my dissertation focused on acoustic identification of individual animals using deep learning approaches.
What drives my work is a desire to understand the behavior of other animals. By developing computational tools to study how different species communicate, I aim to expand our knowledge of their worlds and, in doing so, foster greater empathy and acceptance towards other forms of life.
🔬 What I'm Working On
- interpecies interactions: Investigating evidence of vocal interaction across species
- Embedding Analysis: Understanding what embeddings are encoding in bioacoustic signals
- BioDCASE: Organisation and coordination as tasks chair
💡 Currently Learning/Exploring
Research: Explainable AI and interpretability methods, Communication networks in animals
Reading: The Ministry of Utmost Happiness by Arundhati Roy
News & Tidbits
May 2026
📢 The Third Bank of the River Lea: Listening to Waterbodies in East London @ NIME 2026
Photo by Courtney Reed
We invite everyone to join us on this full-day, in-person workshop at the 2026 New Interfaces for Musical Expression Conference (NIME) in London, UK. Together, we’ll discuss and engage with the multiple identities, phases, and histories of waterbodies. To explore the practicalities of human-machine-water relationships, we’ll take a listening sound walk alongside the Lee Navigation canals in East London, near the Loughborough University London (LUL) host campus. During our day out, we’ll query the plural perspectives that make up the River Lea and its diversion along the Lee Navigation, as a greater entity beyond the physical waterway, and how these can be listened to and engaged with through sound and audio technologies. Registration Deadline: 19 June, 23:59 (AOE). https://thechaoslab.github.io/thirdbank/
June 2026
Paper accepted at Interspeech 2026 - Beyond task performance: Decoding bioacoustic embeddings with speech features
Our work investigates what acoustic information is captured in the representations learned from widely used pre-trained bioacoustic models. By decoding a diverse set of acoustic features from these embeddings, we take a first step toward improving the interpretability of pre-trained models for bioacoustic research and informing model selection for several tasks. The preprint is available on arXiv.
May 2026
📢 The Third Bank of the River Lea: Listening to Waterbodies in East London @ NIME 2026
Photo by Courtney Reed
We invite everyone to join us on this full-day, in-person workshop at the 2026 New Interfaces for Musical Expression Conference (NIME) in London, UK. Together, we’ll discuss and engage with the multiple identities, phases, and histories of waterbodies. To explore the practicalities of human-machine-water relationships, we’ll take a listening sound walk alongside the Lee Navigation canals in East London, near the Loughborough University London (LUL) host campus. During our day out, we’ll query the plural perspectives that make up the River Lea and its diversion along the Lee Navigation, as a greater entity beyond the physical waterway, and how these can be listened to and engaged with through sound and audio technologies. Registration Deadline: 19 June, 23:59 (AOE). https://thechaoslab.github.io/thirdbank/
February 2026
📢 New tasks at BioDCASE 2026
BioDCASE 2026 will host 6 tasks! We were very happy to have received many great task proposals, which made the selection process quite challenging. In addition to the 3 tasks that launched this challenge last year, this edition we will introduce new tasks focusing on mosquitoes, bird counting and active learning. Find more details on the BioDCASE website
January 2026
A Robin sang us a song
after reading so many times about playback experiments, I finally had the chance of trying it myself... This cute fellow seemed very eager to respond to my youtube clips of robins singing.
December 2025
📢 Call for Task Proposals
BioDCASE is now accepting task proposals for the upcoming challenge. If you're working on bioacoustic research and have interesting datasets or problems to share with the community, please consider submitting a proposal!
October 2025
🎉 New Position at Earth Species Project
I'm very happy to say that I started as a postdoc researcher at Earth Species Project! I'll be continuing working on the intersection of computer science and bioacoustics, this time towards the decoding of animal communication.
Selected Publications
BioDCASE: Using data challenges to make community advances in computational bioacoustics
Acoustic identification of individual animals with hierarchical contrastive learning
ICASSP 2025 - IEEE International Conference on Acoustics, Speech and Signal Processing, 2025
Learning to detect an animal sound from five examples
Ecological Informatics, 77 (2023): 102258
Rank-based loss for learning hierarchical representations
ICASSP 2022 - IEEE International Conference on Acoustics, Speech and Signal Processing, 2022
Few-Shot Bioacoustic Event Detection: A New Task at the DCASE 2021 Challenge
DCASE 2021, 2021
Audio-based identification of beehive states
ICASSP 2019 - IEEE International Conference on Acoustics, Speech and Signal Processing, 2019
To bee or not to bee: Investigating machine learning approaches for beehive sound recognition
arXiv preprint arXiv:1811.06016, 2018