Listening to the Sea's Whisper.
Innovative underwater monitoring using pressure sensors to untangle the sounds of the sea and protect our aquatic world.
The Cocktail Party Problem of the Sea
The sea is full of pressure signals—from waves, currents, and marine life to passing ships. Isolating a faint, distant signal is like trying to hear a single voice in a crowded room. PRESENSEA is developing advanced algorithms to solve this "cocktail party problem," enabling us to listen to individual sources with unprecedented clarity.
Non-Invasive Sensors
Using passive, energy-efficient pressure sensors that listen quietly without emitting signals or disturbing marine ecosystems.
Source Separation
Applying machine learning to distinguish the subtle signature of a ship from the background noise of waves and currents.
Open-Source Library
Creating a documented, real-time algorithm library for low-cost platforms, accessible to the wider scientific community.
Decoding the Sea's Pressure Signature
Any movement in a fluid creates a pressure change. The source of such changes can be natural phenomena like waves and currents, or human-induced dynamics like moving vessels. These pressure changes can be observed using sensors that are passive, non-invasive, and energy-efficient. They are not sensitive to common acoustic noise, allowing them to focus on hydrodynamic pressure signals.
Natural Sources
- Waves
- Currents
- Iceberg Calving
Man-Made Sources
- Moving Vessels
- Wastewater Release
- Underwater constructions
Passive Pressure Sensor
Detects all pressure changes to create a complete picture of the underwater environment.
Project Objectives
A four-step journey from multi-sensor data collection to real-world marine validation.
Dataset
Collection and quality auditing of passive pressure, acoustic (PAM), and submarine DAS datasets across diverse sea states and locations.
Model
Physical and bio-inspired modeling of hydrodynamic waves and marine sources, including parametric cod grunt synthesis and shallow-water wave dispersion.
Algorithm
Design of signal-processing and machine-learning algorithms for source separation, synthetic-template detection, label-scarce retrieval, and vessel/wave tracking.
Validation
Validation of the developed methods across real-world marine deployments and benchmarks, including Baltic Sea submarine DAS trials and cross-site PAM datasets.
The Team
Combining expertise in signal processing and hydrodynamic hardware design.
Elizaveta Dubrovinskaya
Researcher
Bringing in-depth knowledge of acoustic signal propagation, processing, and underwater localization.
Maarja Kruusmaa
Supervisor
Leading the Centre for Biorobotics with vast experience in marine robotics and hydrodynamic sensors.
Hosted at the
Centre for Biorobotics, Tallinn University of Technology
Recent Publications
Peer-reviewed conference papers, preprints, and submitted manuscripts from PRESENSEA.
Journal Submission & Preprint · 2026
Detection of Atlantic Cod Vocalizations in Underwater Passive Acoustic Monitoring Data under Label Scarcity
Elizaveta Dubrovinskaya, Anna Bass, Maarja Kruusmaa
Submitted to Ecological Informatics · SSRN Preprint (doi:10.2139/ssrn.7524699)
Evaluates frozen audio foundation models (including BEATs and SurfPerch) and window-duration alignment for retrieving and probing rare Atlantic cod grunts under severe label scarcity.
Conference Paper · May 2026
Bio-inspired Synthetic Cod Grunts Generator and Detection Performance Evaluation on Passive Acoustic Monitoring Data
Elizaveta Dubrovinskaya, Maarja Kruusmaa
Baltic-Nordic Acoustic Meeting (BNAM 2026), Gothenburg, Sweden
Replaces site-specific empirical matched-filter templates with a biophysical parametric pulse-train cod grunt generator calibrated on NOAA SanctSound recordings for cross-deployment generalization.
Conference Submission · OFC 2027
Long-Period Water Waves Sensed by a 6-km Shallow Submarine Telecom Cable in the Baltic Sea
Elizaveta Dubrovinskaya, Bauyrzhan Zhakanov, Laura Piho, Kaida Kaeval
Optical Fiber Communication Conference (OFC 2027)
Analyzes phase-sensitive OTDR submarine cable measurements across the shallow ice-covered Suur Väin strait, resolving 8–14 s water waves, ferry signatures, and arrival angles.
Project Updates
Follow our journey as we deploy, test, and share our findings.
June 26, 2025
A Guide to Open DAS Datasets
A guide to open Distributed Acoustic Sensing (DAS) datasets, explaining the technology and listing key resources for underwater and terrestrial applications.
June 12, 2025
Feeling the Pressure? A Guide to Underwater Sensors
A practical guide to selecting, waterproofing, and deploying low-cost COTS pressure sensors for underwater applications, from DIY projects to professional research.
June 9, 2025
The Baltic Cocktail Party: Stirred by science, served with smarts
Welcome to the Baltic Cocktail Party — where the guests are often invisible, and the real conversations happen under pressure. This isn’t your average party...
Open-Source Software · 2026
das-audit: DAS Data Quality Audit Tool
A lightweight, low-RAM Python CLI and interactive Plotly HTML dashboard for automated quality assessment of Distributed Acoustic Sensing (DAS) datasets (.ds, .sd, .zarr, .h5), covering metadata validation, spectral baselines, coupling coherence, f-k aliasing, and ship-wake analysis.
MSc Thesis · June 2026
Detection of Atlantic Cod Vocalizations under Label Scarcity
Master's thesis by Anna Bass (TalTech, supervised by Elizaveta Dubrovinskaya). Investigates SurfPerch foundation model embeddings, similarity retrieval, clustering, and 1-s tiled window alignment for Atlantic cod grunt detection.
BSc Thesis · June 2026
Multiscale Underwater Acoustic Event Analysis & Visualization
Bachelor's thesis by Alina Murtazina & Sofia Zolina (TalTech, supervised by Elizaveta Dubrovinskaya). Delivers an interactive multiscale visualization and candidate-event analysis system for synchronized DAS and hydrophone arrays.
Project Information & Funding
MOB3JD1212 — Pressure Sensor Based Sea Monitoring
Project description
PRESENSEA (PREssure SENsor based SEA monitoring) is an interdisciplinary research project developing passive, non-invasive underwater monitoring methods based on pressure sensor arrays. While satellites and surface radar observe the sea surface, many environmental and human-induced dynamics—including waves, currents, moving vessels, underwater construction, and wastewater discharge—occur beneath it. In coastal waters, weak signals from distant targets are masked by stronger hydrodynamic disturbances, creating an underwater "cocktail party problem." Combining signal processing, machine learning, and hydrodynamic hardware from TalTech's Centre for Biorobotics, PRESENSEA separates and tracks individual pressure sources in real-world marine environments.
Project objective
The main objective of PRESENSEA is to develop and validate a real-time algorithm library for underwater source separation, detection, and localization tailored to the physics of pressure signal propagation in water. The project collects multi-condition field datasets using an autonomous, telemetric pressure sensor array, builds coupled physical simulation and experimental emulation models of hydrodynamic signals, and validates the algorithms on low-cost computational platforms in at least two distinct sea environments.
Project result
PRESENSEA delivers an autonomous, non-invasive pressure sensor monitoring prototype alongside a documented, real-time algorithm library for source separation, vessel tracking, and sea-state estimation on low-cost hardware. The project produces open-access, FAIR-compliant experimental and modelled datasets in standard marine formats, separating natural hydrodynamic flows from ship-induced wake waves to support coastal engineering, maritime safety, and marine environmental monitoring in line with the EU Mission "Restore our Ocean and Waters."
Project grant amount:
€ 124 675,20
Funding programme: Mobilitas 3.0 / Mobilitas Pluss (Estonian Research Council, Grant no. MOB3JD1212, 01.12.2023 – 30.04.2027), co-funded by the European Union.