Knock knock, who's there? Identifying wild species‐specific fish sounds with passive acoustic localization and random forest models
Bibliographic record
Abstract
Passive acoustic monitoring (PAM) is a useful non-destructive tool for evaluating species presence, diversity and abundance. However, in marine environments, a dearth of tools and methods for identifying wild, species-specific fish calls makes quantitative PAM assessments for specific fish species challenging. We tested a novel passive acoustic localization array with paired audio/video for identifying wild, species-specific fish sounds in a high-diversity region of British Columbia, Canada. We then used random forest models incorporating 47 sound features to test the feasibility of differentiating species-specific fish calls. We identified calls for eight soniferous fish species, five of which had never been documented or described, including vermillion (Sebastes miniatus), canary (S. pinniger), and black rockfish (S. melanops). Random forest models were able to differentiate fish knocks and grunts to the species level with high accuracy (80% for knocks, 88% for grunts). The models struggled to differentiate species knocks when sample sizes were low. The Gini impurity index and partial dependence probability plots showed species-specific differences in call features that measure low frequencies and central frequencies. We also provide a comprehensive set of species-specific call characteristics for 47 sound features which can be used to parameterize fish sound detectors. Our study outlines a robust method for collecting and differentiating wild species-specific fish sounds from a high-diversity region with many closely related soniferous fish species. This research can be used to design a species-specific fish sound detector for quantitative estimates of species presence, diversity and range. These adaptable methods can also be applied elsewhere using the same 47 sound features and random forest models to identify species-specific fish sound parameters.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".