[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82722-en":3,"doc-seo-82722-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82722,4398048949847,"Eliana","https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267",8,"Research & Report","Mixture-Constrained Max Pooling Improves Separation-Based Bird Species Classification","Bird species classification from field recordings remains difficult because vocalizations overlap and species labels are incomplete. The study treats source separation as preprocessing to improve multi-species detection. An ensemble of two MixIT-trained separators, FTRNN and TF-Locoformer, produces separated channels, while mixture-constrained max pooling (MCM) limits each channel’s probability using the species probability from the original mixture. Classifier outputs are computed per channel and then aggregated into final per-species probabilities. Experiments on two real-world datasets show the ensemble improves over single separators, and MCM surpasses standard max pooling while exposing true- and false-positive gains caused by separation errors.","MIXTURE-CONSTRAINED MAX POOLING IMPROVES SEPARATION-BASED  \nBIRD SPECIES CLASSIFICATION  \nYuzhu Wang 1, Kalle Lahtinen 1, Patrik Lauha2 ,3, Shiqi Zhang 1, Panu Somervuo2, Otso Ovaskainen3, Tuomas Virtanen 1  \n1 Signal Processing Research Center, Tampere University, Tampere, Finland  \n2 Faculty of Biological and Environmental Sciences, University of Helsinki, Helsinki, Finland  \n3 Department of Biological and Environmental Science, University of Jyvskyl, Jyvskyl, Finland  \narXiv :2607 .03221v1 [ ee ss .AS] 3 Jul 2026  \nABSTRACT  \nBird 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 classifieris 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.  \nIndex Terms— Source separation, unsupervised learning, bird classification, ecology, multi-species detection  \n1. INTRODUCTION  \nAutomatic bird species classification is central to passive acoustic monitoring for ecological surveys [1, 2, 3], but species recognition from field recordings remains challenging. Such recordings typically contain overlapping vocalizations from multiple species and diverse environmental noises [4, 5] . Training data are also largely weakly labeled, with annotations covering only the dominant species in each clip, while background species are often incompletely labeled [6] .  \nSource separation has been explored to improve bird species classification from field recordings. For single-species separation of Golden-Cheeked Warbler, separators trained on site-specific data combining target-species recordings with real local background noise outperform training on generic birdsong data [7] . However, this improvement in separation quality did not consistently translate into gains in downstream classification performance [7] . A GAN-based spectrogram translation approach has been shown to effectively identify, denoise, and separate target sound sources from real-world recordings [8], but its training requires manually annotated pixel-wise paired spectrograms. Scaling supervised approaches to real-world soundscapes remains challenging, as obtaining clean single-species recordings for a large number of species is difficult in practice. Mixture invariant training (MixIT) [9, 10] opens another avenue for bioacoustic applications, which trains  \nFig. 1. Visualization of the proposed system.  \nseparation models on mixtures of mixtures without clean reference signals. A TDCN++ separator trained with MixIT, combined with an ensemble of EfficientNet-B0 classifiers, demonstrated consistent classification improvement across multiple real-world soundscape datasets [11] . A key finding was that including the original mixture alongside separated channels for classification outperformed using separated channels alone.  \nSeparation can isolate species masked in the original mixture into individual output channels to boost their predicted probabilities (true positive gain) . At the same time, separation may introduce source leakage and artifacts, inflating predicted probabilities for species absent from the recor","cbCaibc7mzdj0Fa3","https://ap.wps.com/l/cbCaibc7mzdj0Fa3","pdf",3089692,1,5,"English","en",105,"# Abstract\n# Introduction\n# Methods","[{\"question\":\"Why is bird species classification challenging from field recordings?\",\"answer\":\"Field recordings contain overlapping vocalizations from multiple species and diverse environmental noise, while training data are often weakly labeled and may miss background species.\"},{\"question\":\"How does mixture-constrained max pooling (MCM) reduce separation-caused errors?\",\"answer\":\"MCM clips each separated channel’s predicted species probability based on the corresponding species probability computed from the original mixture, suppressing false positive gain from separation artifacts.\"},{\"question\":\"What do experiments show about true-positive and false-positive gains from separation?\",\"answer\":\"Separation can increase true positive predictions for species present, but it can also inflate predictions for species absent due to leakage and artifacts; MCM improves overall metrics by capturing the true-positive gain while suppressing false-positive gain.\"}]",1784182495,13,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"mixture-constrained-max-pooling-improves-separation-based-bird-species-classification","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/mixture-constrained-max-pooling-improves-separation-based-bird-species-classification/82722/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is bird species classification challenging from field recordings?","Question",{"text":75,"@type":76},"Field recordings contain overlapping vocalizations from multiple species and diverse environmental noise, while training data are often weakly labeled and may miss background species.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does mixture-constrained max pooling (MCM) reduce separation-caused errors?",{"text":80,"@type":76},"MCM clips each separated channel’s predicted species probability based on the corresponding species probability computed from the original mixture, suppressing false positive gain from separation artifacts.",{"name":82,"@type":73,"acceptedAnswer":83},"What do experiments show about true-positive and false-positive gains from separation?",{"text":84,"@type":76},"Separation can increase true positive predictions for species present, but it can also inflate predictions for species absent due to leakage and artifacts; MCM improves overall metrics by capturing the true-positive gain while suppressing false-positive gain.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":21,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":21,"slug":137},19,"General","general"]