Seal call spectrogram example
Example acoustic visualisation for demonstrating vocal structure and review workflow.
Research
Scientific work behind the consultancy.
Halicho Marine is grounded in active research on underwater seal vocal repertoires, acoustic structure, environmental context and monitoring methods.
Paper under review
Ecological and acoustic structure of underwater seal vocal repertoires.
Ecological and Acoustic Structure of Underwater Seal Vocal Repertoires Revealed Using Contrastive Representation Learning
A first-author research paper currently under review by the journal Bioacoustics.
The study investigates the latent structure of a manually annotated underwater seal vocal repertoire using spectrogram-based representation learning. It analyses 2,825 annotated call events and compares representation-learning approaches to understand call-family structure, subtype organisation and environmental relationships.
The work suggests that seal vocal repertoires may be hierarchically organised into acoustically meaningful call families, with repertoire composition varying across recording hour, tidal phase and low-frequency noise conditions.
BioacousticsRepresentation learningUMAPCall subtypesEnvironmental context
Acoustic examples
Spectrograms make seal vocal structure visible.
Short spectrogram examples can be used to demonstrate call structure, review decisions and the acoustic patterns underpinning repertoire analysis.
Seal call spectrogram example
Example acoustic visualisation for communicating call morphology and classification context.
Conference work
Collaborative acoustic monitoring research.
Automated Classification of Vocalisations from Wild and Captive Seal Populations
Collaborative conference paper with the University of Bath and Celtic Sea Power exploring labelled seal call classes and spectrogram-based classification. The work described S1, S2, S3, SP4 and SP5 classes and evaluated model performance across those categories.
Use on this website is descriptive of prior collaborative research and does not imply ownership of any partner commercial detector.
Power and Accuracy Trade-Offs for Machine Learning Methods Applied to Detection of Underwater Sound Sources
Conference paper comparing model design choices for underwater acoustic classification, including trade-offs between waveform and spectrogram-based approaches, computational cost and performance.
SubSea Soundscape / S3
Experience contributing to regional acoustic evidence-base thinking for the Celtic Sea, including soundscape monitoring, marine mammal evidence, machine learning, data integration and consenting relevance.
The research direction is not just automated detection. It is biologically interpretable repertoire analysis for better passive acoustic monitoring.