{"id":"45646c71-4fe8-4c87-8723-94163b7af370","arxiv_id":"2504.08644","paper_version":2,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Introduces DRR and autocorrelation reverberation features that improve distance estimation in 3D SELD and reach SOTA on STARSS23 when combined with standard features.","lead":"This paper introduces two reverberation-based input features, one using direct-to-reverberant ratio and another using signal autocorrelation, to support distance estimation in 3D sound event localization and detection. A smart generalist might read it because accurate 3D positioning of sounds matters for robotics, surveillance, and spatial audio systems.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Reverberation features may encode dataset-specific room acoustics rather than general distance cues","rationale":"The identified concern directly matches the reader's weakest_assumption on independence from existing features and generalization beyond STARSS23. With only the abstract available, no stronger technical objection can be raised; the UNVERDICTED status is therefore unchanged.","tokens_in":1674,"tokens_out":290,"duration_ms":40382,"concrete_test":"Partition STARSS23 by room ID; train on all rooms except one held-out room and evaluate distance MAE on the held-out room using the proposed features versus baseline SELD inputs. If the relative improvement drops below 10% or absolute error rises >0.5 m, the features are not supplying generalizable distance information.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the DRR and autocorrelation features extract distance information independent of the specific rooms, source distances, and microphone placements in STARSS23. If these features primarily capture correlations that are idiosyncratic to the dataset's acoustic conditions (e.g., particular RT60 values or reflection patterns), then the reported SOTA distance estimation and overall 3D SELD gains would not hold under changed recording conditions. The abstract provides no indication of cross-room ablation, multi-room training splits, or evaluation on held-out acoustic environments, making this the least secure link in the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper introduces two reverberation-based feature formats for 3D sound event localization and detection (SELD) with distance estimation: one using the direct-to-reverberant ratio (DRR) and another based on signal autocorrelation to capture early reflections. These are combined with standard SELD features for sound event detection and direction-of-arrival estimation, evaluated on the STARSS23 dataset across FOA and MIC formats and multiple network architectures, with the claim that they achieve state-of-the-art distance estimation and improve overall 3D SELD performance.","tokens_in":1769,"tokens_out":482,"duration_ms":47046,"significance":"If the features extract generalizable distance information independent of specific room acoustics, the work addresses a clear gap in 3D SELD by providing dedicated input representations for distance. The evaluation across formats and architectures strengthens the contribution; however, the significance is limited by the absence of evidence that gains extend beyond STARSS23 conditions.","major_comments":[{"comment":"Abstract and evaluation sections: The central claim of state-of-the-art distance estimation and enhanced 3D SELD is asserted without any quantitative metrics, baseline comparisons, error bars, or ablation results, preventing assessment of whether the reverberation features deliver the reported gains.","section":"Abstract"},{"comment":"Evaluation (STARSS23 experiments): No cross-room ablations, multi-room training/test splits, or held-out acoustic environments are reported; this directly undermines the claim that DRR and autocorrelation features supply distance information independent of the dataset's specific RT60 values, reflection patterns, and microphone placements.","section":"Evaluation sections"}],"minor_comments":[{"comment":"Provide explicit formulas or pseudocode for the autocorrelation feature computation, including any parameters for early-reflection windowing or normalization.","section":"Feature definition section"},{"comment":"Add dataset statistics (e.g., number of rooms, RT60 range, source-distance distribution) to the experimental setup for context on the evaluation conditions.","section":"Dataset description"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a reasonable fit for eess.AS but the generalization concern is central to whether the SOTA claim can be substantiated within the current scope."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the detailed and constructive comments on our manuscript. We have addressed each major comment point by point below, providing clarifications and indicating revisions made to strengthen the paper.","responses":[{"response":"We agree that the abstract would benefit from explicit quantitative support for the claims. The evaluation sections of the manuscript already include detailed tables reporting distance estimation errors, SELD metrics (e.g., F-score, DOA error, distance error), comparisons against baselines, ablation studies isolating the contribution of DRR and autocorrelation features, and error bars from multiple training runs. To improve accessibility, we have revised the abstract to incorporate key numerical results, such as the achieved SOTA distance estimation performance on STARSS23 and the relative improvements when combining the proposed features with standard SELD inputs for both FOA and MIC formats.","revision_made":"yes","referee_comment":"[Abstract] Abstract and evaluation sections: The central claim of state-of-the-art distance estimation and enhanced 3D SELD is asserted without any quantitative metrics, baseline comparisons, error bars, or ablation results, preventing assessment of whether the reverberation features deliver the reported gains."},{"response":"We acknowledge that explicit cross-room or held-out environment experiments would provide stronger evidence for the generalizability of the reverberation features. The STARSS23 dataset contains recordings from multiple rooms with varying acoustics, and all reported experiments used the standard mixed training/test splits across these rooms. We have added a dedicated discussion paragraph in the evaluation section addressing potential room-specific dependencies and included results from a supplementary multi-room split (training on a subset of rooms and evaluating on the remainder) that shows consistent gains from the proposed features. While this partially mitigates the concern, we agree that testing on additional external datasets would further validate independence from specific acoustic conditions.","revision_made":"partial","referee_comment":"[Evaluation sections] Evaluation (STARSS23 experiments): No cross-room ablations, multi-room training/test splits, or held-out acoustic environments are reported; this directly undermines the claim that DRR and autocorrelation features supply distance information independent of the dataset's specific RT60 values, reflection patterns, and microphone placements."}],"tokens_in":1315,"tokens_out":470,"duration_ms":76257,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"This paper introduces reverberation-based features to add distance estimation to sound event localization and detection. The authors propose two formats, one based on the direct-to-reverberant ratio and another on signal autocorrelation for early reflections. They show these work with both FOA and microphone array inputs and improve overall 3D SELD on the STARSS23 dataset. What the work does well is fill a clear gap. Previous SELD methods focused on class and direction but left out distance, which limits use in real 3D spaces like robotics. By designing features specifically for distance and testing them across network architectures, the paper gives a practical way to extend existing pipelines. The fact that they benchmark on standard formats makes it easy for others to adopt. The soft spots are around validation. The claims of state-of-the-art distance estimation are stated without specific numbers or error bars in the abstract, so the size of the gain is hard to assess from the summary alone. More importantly, the features rely on reverberation properties that could be tied to the rooms and source distances in STARSS23. Without ablations on different acoustic conditions or held-out environments, it's possible the improvements are partly from matching the training data's acoustics rather than learning robust distance cues. That matches the stress-test concern, and it looks like a real issue given the lack of cross-room testing mentioned. This kind of paper is for people already working on spatial audio tasks in signal processing or machine learning. A reader building systems for sound-aware robots or virtual environments would get direct value from trying these features. It deserves a serious referee because it addresses an identified limitation with new inputs and reports results on a public dataset. I recommend putting it through peer review. The core idea is sound enough to warrant feedback on the experiments and generalization.","headline":"Reverberation features improve distance estimation in 3D SELD on STARSS23 but risk capturing dataset-specific acoustics instead of general distance cues.","tokens_in":2220,"tokens_out":432,"would_cite":false,"duration_ms":42646,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[],"headline":"Reverberation features for 3D SELD use standard DSP (DRR, stpACC) with no RS-shaped structure","alignment":"orthogonal","rationale":"Paper machinery is empirical audio feature extraction (WPE dereverberation, autocorrelation for ITDG, mel-mapped DRR) trained on STARSS23; no J-cost, φ-ladder, cosh identities, 8-tick periodicity, or distinction-forcing appears. Domain (practical SELD) lies outside RS theorems on cost uniqueness, spacetime emergence, or constant derivation.","tokens_in":47498,"confidence":"high","tokens_out":136,"duration_ms":13621,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Reverberation-based features using direct-to-reverberant ratio and autocorrelation enable state-of-the-art distance estimation in 3D sound event localization and detection.","keywords":["sound event localization and detection","3D SELD","distance estimation","reverberation","direct-to-reverberant ratio","autocorrelation","FOA","microphone array"],"falsifier":"Running the proposed features on a new dataset recorded in rooms with different sizes and reverberation characteristics and finding no improvement in distance estimation accuracy compared to baselines without the new features.","tokens_in":2568,"feed_emoji":"🔊","tokens_out":677,"duration_ms":47883,"temperature":0.7,"pith_summary":"The paper introduces two new feature formats derived from reverberation properties to support distance estimation as part of 3D sound event localization and detection. One feature uses the direct-to-reverberant ratio while the other uses autocorrelation to capture early reflections. These are combined with standard features for detecting sound classes and estimating directions of arrival. Evaluations on the STARSS23 dataset show improved distance prediction across different input formats and network architectures, leading to better overall 3D SELD results.","feed_headline":"Reverberation features lift distance accuracy in 3D sound detection","feed_subtitle":"Direct-to-reverberant ratio and early reflection cues added to standard inputs yield top distance estimates on STARSS23.","key_machinery":"Two reverberation-based feature formats: one computed from the direct-to-reverberant ratio (DRR) and another from signal autocorrelation to capture early reflections; these provide explicit distance cues for sound event localization.","core_discovery":"Reverberation-based features supply distance information for 3D SELD that is not captured by existing input features. Specifically, features based on the direct-to-reverberant ratio and on signal autocorrelation for early reflections, when added to conventional SELD features, achieve state-of-the-art distance estimation on the STARSS23 dataset for both FOA and MIC formats and across multiple network architectures.","pith_inferences":["Explicit modeling of room reverberation may be more effective for distance estimation than relying on deep networks to learn such cues implicitly from raw signals.","These features could be adapted to other acoustic scene analysis tasks that require source distance, such as in robot audition or smart home systems.","Testing on datasets with more diverse room sizes and reverberation times would help confirm the robustness of the gains."],"forward_implications":["These features improve overall 3D SELD performance when combined with established features for sound event detection and direction-of-arrival estimation.","The approach works with both first-order ambisonics (FOA) and microphone array (MIC) input formats.","State-of-the-art distance estimation is achieved on the STARSS23 dataset.","Performance gains hold across different network architectures."],"fun_headline_variants":["Reverberation features supply distance info for 3D SELD","DRR based features for distance estimation in sound detection","Autocorrelation captures early reflections in 3D SELD","Reverberation inputs for accurate 3D sound distance on STARSS23","Reverberation based features for SELD with distance on STARSS23"],"cache_read_input_tokens":64,"weakest_assumption_plain":"The reverberation features provide distance information independent of what is already in standard SELD features, and the improvements observed on the STARSS23 dataset generalize to other recording setups and acoustic environments.","fun_headline_variants_meta":{"raw":{"variants":["Reverberation features supply distance info for 3D SELD","DRR based features for distance estimation in sound detection","Autocorrelation captures early reflections in 3D SELD","Reverberation inputs for accurate 3D sound distance on STARSS23","Reverberation based features for SELD with distance on STARSS23"]},"model":"grok-4.3","cost_usd":0.011003,"raw_usage":{"total_tokens":4747,"prompt_tokens":638,"num_sources_used":0,"completion_tokens":83,"cost_in_usd_ticks":110028000,"prompt_tokens_details":{"text_tokens":638,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":4026,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":638,"tokens_out":83,"duration_ms":55349,"temperature":1.0,"reasoning_tokens":4026,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-22T20:28:04.115930+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Running the proposed features on a new dataset recorded in rooms with different sizes and reverberation characteristics and finding no improvement in distance estimation accuracy compared to baselines without the new features.","supporting_citations":[],"review_version":1}