Find what exists.
Know what it permits.

Caller identity within and across call types; method comparator for evaluation splits, not a behavior/context dataset.

Reproducible acoustic comparisons and identity analysis.

Inspect labels and group keys; the sample is not yet suitable for a context-association benchmark.

Near-match for group/day/block-level vocal-sociality analyses; not an event-aligned sound-to-behavior benchmark.

Embedding search, annotation and lightweight classifier workflows.

Frozen acoustic representation candidate.

Bird-vocalization presence detection; not observed behavior.

45-task, 37,718-example evaluation suite for acoustic perception, biological-category recognition, scene understanding and in-context learning.

Training supervision across four bioacoustic capability tiers; paper reports 43,922,135 audio-language conversations. The project overview says 6.5 million distinct audio clips; the current card's resolver index says 4,745,972 distinct references; definitions not reconciled.

Behavioral playback controls for claims about meaning and call order.

Methods comparator for within-cohort caller identification, contact-call convergence and receiver response; not a semantic or phone-audio benchmark.

Species-specific feasibility and future baseline design.

Species-specific feasibility and future baseline design.

Species-specific feasibility and future baseline design.

Species-specific feasibility and future baseline design.

Species-specific feasibility and future baseline design.

Noncommercial methods comparator for synchronized audio/video and observable behavior labels; modality-ablation design.

Potential species-specific vocal-development reference; not an event-context or semantic-translation benchmark.

Non-commercial methods comparator for retrieving camera-trap clips from human-written queries about visible, labeled behavior; it is not an audio-language or animal-meaning benchmark.

Potential reference for transparent query parsing and evaluation over structured camera-trap behavior annotations.

Candidate source for studying acoustic associations with caller, addressee and video-observed social context in Egyptian fruit bats.

Audit the schema and aggregate structure of video-observed caller, addressee, behavior and social-context labels; not a translation benchmark.

Methods comparator for sequence-level context classification and a possible source of derived encodings and annotations.

Primary methods and benchmark reference for sparse event detection, call-type classification and event-level evaluation.

Non-commercial methods reference for rare-event detection and time-localized call-category labels in continuous audio.

Reference implementation for sparse, rare-event bioacoustic detection and call classification.

Published model checkpoint reference for MeerKAT call/event classification.

Method reference for studying reported use of learned human names by companion parrots; not an acoustic dataset or animal-language benchmark.

Interface-design reference for separating caregiver-reported context from recorded sound, researcher coding and model predictions.

Methods reference for wild-parrot call categorization and sequence structure; not a translated-message or phone-recording benchmark.

Reference for call-event, speaker/pair, sequence-position and acoustic-measurement fields in a wild-parrot duet study.

Methods reference for controlled extracellular recording of fungal electrical activity; not a biological-language or Talk2Nature model result.

Possible methods-reproduction input for extracellular fungal electrophysiology, not an animal behavior or communication benchmark.

Potential implementation reference for reproducing the paper's STFT and power-spectral-density analysis.

Determine whether video-derived FileID annotations can be joined to recording time, channel, treatment and session for a leakage-safe grouping audit.

Review subject/treatment descriptions and annotation definitions before assessing FileID-to-session and animal-group splits.

Understand what a targeted contact-call playback can establish about receiver response and what it cannot establish about meaning.

Reproducibility and evaluation-split comparator for context prediction from bat ultrasound calls.

Identify group leakage, background/site confounds and evaluation controls relevant to animal-audio models.

Find reusable data-access and preprocessing groundwork for the public Egyptian fruit-bat archive, then determine whether it supports any part of the reported context-prediction workflow.

Understand source separation of overlapping animal vocalizations as an upstream processing step, and keep reconstruction evidence distinct from context or communication claims.

Trace the paper's stated code release for source separation, mixture generation and downstream evaluation.

Use a proposed conceptual framework to keep claims about animal names distinct from identity, context association and receiver response.

Potential reference for a bounded reproduction of receiver response to plant-like acoustic cues; not an audio corpus or animal-language training set.

Potential source for checking the published behavioral tables and analysis provenance; distinct from a sound corpus or animal-language training set.

Methods context for clustered photographs, bottle-level experimental units and small-sample claims; not an animal-communication resource.

Methods reference for event-level song/call type annotation, repertoire analysis and spatial vocal variation. Not a semantic-translation or independently observed behavior benchmark.

Potentially inspect analysis-validation code for reproducibility after exact record identity and code license are verified. No code execution is planned at intake.

A critical reading for claims about forest network prevalence, resource transfer and “mother tree” effects; not a communication dataset or model benchmark.

Review the physiological and isotope-method arguments about carbon transfer through ectomycorrhizal networks; not evidence of a tree communication code.

A design and methods comparator for animal-operated interfaces and caregiver-mediated live versus playback interactions; not an acoustic communication benchmark.

Tablet interaction ergonomics and welfare-aware design questions for parrot-operated enrichment; not communication decoding or an acoustic ML benchmark.

Read the counterarguments and methods perspective on forest network structure and context-dependent material transfer; this Opinion is not an independent replication.

Primary source for the NatureLM-audio architecture and BEANS-Zero benchmark claims.

Official software for running and training the NatureLM-audio model.

Bioacoustic audio-language classification and captioning research comparator.

Audio-text training examples for the NatureLM-audio model.

Zero-shot, audio-plus-text test-only tasks for species or sound classification, detection, life-stage/call-type classification and captioning; a broad method comparator, not a receiver-response or communication benchmark.

Fetch/list/evaluate BEANS-Zero tasks and calculate reported benchmark metrics.

Test prompt robustness and instruction-following recovery in NatureLM-audio by interpolating its LoRA with the Llama base model.

Reproduce authors’ LoRA interpolation experiments and prompt/closed-set classification results.

Agent-assisted optimization of a fixed single-GPU language-model training setup; its validation bits-per-byte metric is not an animal-behavior measure.

Compare adult contact-call variation across wild, captive and reintroduced populations and describe nestling vocal development; not a context or meaning task.

Metadata and derived analyses for adult and nestling call selection/variation; no independently observed behavior labels or established semantic targets.

R scripts for adult call acoustic analysis and nestling vocal-development analysis, according to the Dryad file descriptions.

Compare frozen human-speech, animal-bioacoustic and general-audio representations on caller identification and call-type classification across four species.

Compare the relative acoustic structure of individual identity and group/site identity in learned parrot contact calls; methods comparator, not a context or meaning dataset.

Potential source for a separately scoped acoustic identity/convergence reproduction; not a behavior/context-labelled corpus.

Reproducibility reference for R Markdown analyses of acoustic identity and social-scale convergence.

Understand the scope and limits of a caregiver-reported companion-parrot vocal-learning survey; not an audio corpus or communication benchmark.

Learn from an established annotation and model-prediction review workflow before expanding Talk2Nature's own researcher-facing interface.

Potential separate research-workbench candidate; evaluate a standards-based exchange path without importing GPL code.

Provide an independent format-level check for a local Talk2Nature-to-AOEF export without adding a third-party application or runtime dependency.

Discover and access existing multi-taxon bioacoustic datasets through a shared Python interface.

Separate acoustic target/noise detection from species and call-type classification across animal-specific datasets.

Train and evaluate animal-specific target/noise, species or call-type classifiers using the published framework.

Candidate only for monk-parakeet target-versus-noise detection and split auditing; this deposit is not documented as the paper's call-type corpus.

Discover candidate bioacoustic datasets and compare what metadata is available before selecting any species-specific research question.

Reference for ultrasonic plant-emission measurement and stress-condition classification, not animal communication or translation.

Critical context for distinguishing plant acoustic emissions and physiological cues from experimentally supported receiver communication.

Potential plant-sound event and stress-condition methods reference; not a dataset for animal communication or phone-microphone claims.

Metadata-only reference to the classifier training function cited by the plant acoustics paper.

Methods reference for multilabel coarse soundscape context and cross-domain evaluation; not an animal communication or species-identification benchmark.

Reference implementation for the paper's soundscape experiments; potential later reproducibility candidate only after scientific and dataset review.

Author-provided model for broad soundscape-layer classification, not focal-animal detection, species identification or meaning inference.

Potential ecoacoustic context-classification reference with labels for anthropophony, biophony, geophony and silence; not a focal-animal communication corpus.

Methods reference for masked acoustic-token generation, acoustic style transfer, and evaluation of sperm-whale coda representations.

Reference implementation for the paper’s audio generation, fine-tuning and downstream evaluation methods.

Potential acoustic-structure/representation methods data; not a behavior-context or semantic-translation dataset.

Research checkpoint for WhAM inference; no Talk2Nature task or product use established.

Candidate packaging reference for sharing structured biodiversity data with a descriptor, tabular resources and explicit schemas.

Trace the guide's versioned profile/schema links and identify an authoritative, released artifact before any compatibility work.

Candidate vocabulary for describing biodiversity-related sound and other media, including rights, access and resource-level metadata.

Methods reference for testing receiver responses to call components, combinations and reversed call order.

Follow-up field test of call order and whether two call types are perceived as one source-linked unit.

Local inspection of the small source-data tables and their relationship to the 2022 receiver-response analysis.

Static review of the two deposited R scripts for data linkage, analysis steps, dependencies and reproducibility boundaries.

A bounded test of whether a previously familiar heterospecific recruitment call can occupy the recruitment position in a novel ordered sequence.

Potential tabular reproduction asset for published playback outcomes and acoustic analyses; not assumed to contain raw audio.

Competing formal explanations for receiver-response evidence in Japanese-tit ABC–D sequences.

Bounded playback comparison of great-tit behavior toward Carolina-wren FME–D call sequences and D-only calls.

Primary methods reference for controlled resource-transfer measurements through common endophyte networks.

Potential data for checking the paper's small controlled isotope-transfer results after rights and structure review.

Potential statistical-analysis reproduction after code, dependency and rights review.

Primary-source method comparator for researcher-observed context, note-type categories and parrot vocal variation.

Methods and annotation-schema comparator for how observed behavioral context relates to parrot note types and acoustic measurements.

Self-assess readiness of a bounded open software component for possible Digital Public Goods Alliance review.

Methods reference for when citizen-science smartphone audio can support a defined bioacoustic question and how device variability changes acoustic measurements.

Potential reference for studying the provenance and acoustic usability of citizen-science nightingale recordings.

Cross-taxon field-method reference for model-specific microphone variability, detection limits and reference-device checks.

Potential reference for inspecting cross-device wildlife recordings after exact file, rights and privacy review.

Primary methods reference for PteroSet's time-frequency annotation, bird-presence detection baseline and project-held-out evaluation.

Candidate for a later reproducibility check of time-frequency bird detection; not a behavior or communication benchmark.

Potential reproducibility reference for converting Raven annotations, segmenting time-lapse audio and building project-level splits.

Possible baseline comparison artifact only after its model and weight lineage is independently reviewed.

Methods comparator for dog-held-out context classification and transfer from human-speech representations; not evidence of call meaning or cross-species transfer.

Historical methods reference for induced canine vocalizations and stimulus-associated labels; not admitted to Talk2Nature training.

Potential reproducibility reference only if a specific code release and dependencies are lawfully obtained and reviewed.

Primary methods reference for multi-channel measurement and analysis of slow electrical activity in oyster-fungus-colonised substrate.

Potential source data for reviewing the paper's electrophysiology analysis; not an animal-behavior dataset or communication benchmark.

Understand established bioacoustic observation, annotation and classification workflows before adding advanced research features to Talk2Nature.

Possible reference implementation for researcher-side acoustic annotation and detection workflows; no current integration task selected.

Potential strong-label detection/classification comparator for bird soundscape methods; not a communication or receiver-response benchmark.

Methods comparator for continuous bioacoustic event labeling, individual-held-out evaluation and pretraining-data boundaries.

Metadata-only method reference for event labeling, animal-held-out splits and benchmark data-containment decisions.

Potential comparison for multimodal, owner-contributed canine observations and project-level open-data plans; not a released dataset or validated communication model.

Primary method and hypothesis reference for extracellular voltage-spike analysis in four fungal species; not a decoded-language result.

Potential raw-signal resource for a rights-cleared, expert-reviewed replication of fungal electrical measurement and signal-processing methods.

Methodological and linguistic critique of interpreting fungal extracellular voltage patterns as a language.

Methods and evidence review for fungal electrophysiology, extracellular measurement artifacts and study-design controls.

Mobile bioacoustics implementation and training-evaluation methods comparator for four cattle vocalization classes.

Mobile inference model files listed by the CowRumIA README; no Talk2Nature use approved.

Four-class cattle vocalization dataset described as stress/calm by beef/dairy category; candidate for metadata-level study-design review only.

Bounded methods comparator for how playback/social acoustic context can alter trained parrot vocal output; not a decoding or consumer-app benchmark.

Computer-vision reference for detecting and measuring a known, structured animal signal against a feeder-location outcome.

Historical implementation reference for camera-based waggle-run detection, orientation estimation and mapping.

Potential reference sample for inspecting detector duration/orientation labels after resolving data rights and provenance.

Engineering comparator for long-duration colony tracking and realistic camera/compute requirements.

Methods comparator for dog-vocalization demographics, identity, activity/environment, emotion, detection and source separation; not evidence of call meaning.

Open-source project-structure comparator for a challenge landing page only; no dog-audio training or evaluation code was identified in the reviewed repository tree.

Context-label classification methods reference. Predicting activity/environment labels from bark audio is not a validated welfare measure, receiver response or animal-language translation.

Published alternative interpretation of Japanese-tit receiver responses, call order and claims about Core-Merge or language syntax.

Longitudinal change in learned contact-call dialects; a cultural-evolution and conservation comparator, not call-meaning decoding.

Longitudinal dialect and call-type measurement reference; deposited tables do not include the raw audio needed for an audio model.

Reference implementation for spectrographic cross-correlation, unsupervised random-forest similarity and acoustic-space comparisons across dialects and years.