AI-Assisted Fish Sound Detection in Underwater Recordings

Ecology · Research & Development

What it collects

About a measurement

Anonymized data

Passive acoustic recordings from underwater hydrophones capturing fish sound events. These are environmental audio measurements with no personal information attached. Training data was collected in British Columbia.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Run by: Fisheries and Oceans Canada (DFO)
Where: No fixed location
Kept: Not stated by the Helpful Places.
Shared with: Accountable organization
Your copy: You cannot see the data it holds about you. What you can do

What it is for

Ecology

Research & Development

FishSoundFinder is an open-source software tool that uses machine learning to automatically detect fish sounds in passive acoustic recordings. It helps Fisheries and Oceans Canada scientists estimate fish presence without disturbing aquatic environments. The system does not collect or process any personal information. It was developed using fish sound data collected in British Columbia and has been tested in regions including Florida.

What it collects and what happens to it

Data taken in

About a measurement

Anonymized data

Passive acoustic recordings from underwater hydrophones capturing fish sound events. These are environmental audio measurements with no personal information attached. Training data was collected in British Columbia.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Processing

Classification & Prediction

Random Forest and Convolutional Neural Network (CNN) algorithms trained on manually identified fish sounds are used to classify acoustic segments, predicting whether fish sounds are present in passive recordings.

Government of Canada AI Register — entry 2526-DFO-MPO-001

What it does

Sensing (Perceptive AI)

Human decides

The system senses fish sounds in raw audio recordings, converting passive acoustic signal data into structured detections of fish presence. Scientists review the outputs to draw ecological conclusions.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Deciding (Analytical AI)

Human decides

Random Forest and CNN models classify audio segments as containing fish sounds or not, producing presence/absence scores from structured acoustic features. A human researcher interprets and acts on these classifications.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Outputs

About a measurement

Anonymized data

Detection outputs indicating the presence or absence of fish sounds within acoustic recording segments, enabling estimates of fish presence over time and across locations. No personal information is produced.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Run by

Fisheries and Oceans Canada (DFO)

The federal department responsible for safeguarding Canadian waters and managing aquatic resources. It deploys FishSoundFinder for use by Government of Canada employees conducting passive acoustic monitoring research.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Built by

FishSoundFinder

FishSoundFinder is the open-source software tool developed through this project. It is listed as the vendor in the register, reflecting the publicly available software package that implements the AI detection algorithms.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Kept for

Not stated by the Helpful Places.

Shared with

Not available to me

As a member of the public, you have no personal data in this system. The system processes only environmental acoustic recordings and involves no personal information, so individual access rights do not apply.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Available to the accountable organization

Detection results and processed outputs are available to Fisheries and Oceans Canada scientists and GC employees who use the tool for ecological research and fish population monitoring.

Government of Canada AI Register — entry 2526-DFO-MPO-001

Stored

Not stated by the Helpful Places.

How to read the colours

Can it identify you?

Anonymized data: Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data: Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data: The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides: This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes: This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous: This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the Helpful Places, not the vendor.

Your rights

Right to Algorithmic TransparencyFishSoundFinder is open-source software, meaning the algorithms and model architecture are publicly available for inspection. Members of the public can review the source code to understand how fish sound detection works. No personal data is involved, so individual data rights do not apply.

Risks and safeguards

No risks or safeguards have been published for this system yet.