Automatic grunt detector and recognizer for Atlantic cod (Gadus morhua).

Area of Science:

• Marine biology
• Bioacoustics
• Fisheries science

Background:

• Northwest Atlantic cod (Gadus morhua) populations are severely depleted due to overfishing.
• Passive acoustic technology offers potential for monitoring fish populations, but requires species-specific detection methods.
• Identifying spawning locations and persistence of Atlantic cod is vital for conservation and management.

Purpose of the Study:

• To develop an automatic grunt detection and recognition algorithm for Atlantic cod.
• To overcome limitations of manual analysis of large passive acoustic datasets.
• To facilitate the identification of cod presence in acoustic recordings.

Main Methods:

• A two-stage hypothesis testing algorithm was designed for grunt sound detection and recognition.
• The algorithm processes yearlong passive acoustic data recordings.
• Performance was evaluated based on detection probability across different signal-to-noise ratios (SNR).

Main Results:

• The algorithm achieved a 0.93 detection probability for cod grunts with SNR > 10 dB.
• A 0.8 detection probability was achieved for grunts with SNR between 3 and 10 dB.
• The detector significantly reduces the time needed to identify cod grunts in acoustic data.

Conclusions:

• The developed algorithm effectively detects and recognizes Atlantic cod grunts from passive acoustic data.
• This technology aids in identifying cod presence in U.S. waters, supporting stock assessment.
• The tool streamlines research and monitoring efforts for recovering cod populations.