Mixture-Constrained Max Pooling Improves Separation-Based Bird Species Classification
2026-07-03 • Sound
Sound
AI summaryⓘ
The authors work on improving how we identify different bird species from field recordings, which is hard because birds often sing over each other and some labels are missing. They use two sound separation methods combined together to split mixed bird calls before trying to identify the species. To reduce mistakes caused by separation, they introduce a technique that limits guesses based on the original sound mix, helping to avoid false alarms. Tests show that their combined approach works better than individual methods and that separating sounds helps find more real bird species but can also cause some false detections.
bird species classificationsource separationmixture invariant training (MixIT)ensemble methodsfalse positivesmax poolingaudio signal processingmulti-species detectionfield recordingssound mixture analysis
Authors
Yuzhu Wang, Kalle Lahtinen, Patrik Lauha, Shiqi Zhang, Panu Somervuo, Otso Ovaskainen, Tuomas Virtanen
Abstract
Bird species classification from field recordings remains challenging due to overlapping vocalizations and incomplete species labels. We study source separation as a preprocessing for bird species classification to improve multi-species detection. Specifically, we employ an ensemble of two separators, FTRNN and TF-Locoformer, both trained with mixture invariant training (MixIT). To address the false positive gain caused by separation errors in separated outputs, we propose mixture-constrained max pooling (MCM), which clips the predicted probability from each separated channel based on the corresponding species probability in the original mixture. The classifier is applied to each separated output and the original mixture independently, and MCM aggregates the predictions into a final per-species probability. Experiments on two real-world datasets show that the ensemble outperforms individual separators and MCM outperforms standard max pooling across multiple metrics, and reveal that separation leads to both true positive gain for present species and false positive gain for absent species.