Acoustics, balance, and chimpanzees – The ABCs of developing a deep learning-based automated acoustic detector for wild chimpanzee (Pan troglodytes) loud calls
Bibliographic record
Abstract
1. Passive acoustic monitoring (PAM) is a powerful tool for wildlife monitoring, but the time and expertise required to process large volumes of data pose significant challenges. Automated acoustic detectors improve efficiency by speeding up data processing. Class imbalance, resulting from fewer target signals relative to noise, complicates development and can negatively impact performance. However, training datasets should also reflect the conditions of real-world PAM datasets. 2. We developed an automated acoustic detector for chimpanzee loud calls while addressing class imbalance. We predicted that greater data diversity and high-quality data (clear signals, minimal noise interference) would enhance network performance and that class imbalance, by supporting diversity, is essential for functionality. We built training datasets with data recorded in wild settings and applied a temporal convolutional neural network approach using Deep Audio Segmenter (DAS). We trained networks using datasets containing varying levels of noise (50%, 75%, 90%, 99%) and also tested the effectiveness of frequency removal in improving performance. 3. The network performances varied significantly, with F1 scores of 0.44 to 0.86, exceeding a previous study (5% F1). The most imbalanced dataset produced the best performing network, capturing 90% of pant-hoot events and annotating them with 90% (SD = 20.9) accuracy. The results showed that increased class size was associated with greater intraclass diversity and improved precision (0.41–0.83). The networks showed consistently high recall rates, especially when frequency removal was not applied (0.89–0.92). 4. This study stresses the importance of class size and diversity in developing automated acoustic detectors. It also highlights the value of high-quality data for accurate pattern recognition of the target signal and the importance of the noise class for effective class decoupling and detector functionality. This research supports the advancement of PAM in chimpanzee studies, opening new opportunities to integrate remote sensing for efficient wildlife monitoring.
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 | 0.003 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.004 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".