每日 arXiv

音频 · 语音 · 音乐 · 声学

每日新爬取的 arXiv 论文，按评分排序。中文摘要帮你快速判断是否值得深读。

2026-08-09
日期2026-08-09
已评分—
均分—
最高—

Daily Papers — 2026-08-09

13 papers on audio, speech, music, and acoustics.

1. Steering dense music retrieval with open-vocabulary concept discovery

Authors: Julien Guinot, Alain Riou, Elio Quinton, György Fazekas

Categories: cs.SD, eess.AS | Accepted to ISMIR 2026 Score: 7.16/10 (Obj:8 Id:6 Ind:6 Comp:8 Eff:6 Nov:8 Rep:9)

  • Strength: 将 SAE concept attribution 从独立神经元排序重构为稀疏逆问题，在 cardinality-matched 比较下 AUROC 从~50%提升至~65-80%，bundle recall 改善10-70%，且方法 training-free、~20ms/inversion。
  • Weakness: Baseline 仅 DTN cosine probing 一种，评测仅覆盖 top-20 concepts / 单数据集 / 无用户研究，Mahalanobis 距离同时用作正则项和诊断指标存在轻微循环论证。
  • Full review: Claude Code 全文七公理审稿

Controllable music retrieval lets users find music that is, for example, more ambient, less distorted, or without guitar while preserving the other semantic content of an original seed query. Sparse autoencoders (SAEs) are a promising interface for this kind of concept-level control, but a key problem remains: given a free-form text concept, which sparse features should be edited? In shared multimodal embedding spaces, standard attribution methods often select neurons that match the concept’s wording but not the audio examples that express it. This leads to weak or unstable edits: relevant features are missed when concepts are distributed across neurons, while others are selected due to text alignment rather than audio-side structure. We address this with a lightweight, training-free method that recovers a sparse set of audio features whose decoded representation reconstructs the target concept while remaining consistent with audio-space geometry. This reframes concept attribution as a sparse inversion problem rather than a text-side neuron-ranking heuristic. The method requires neither paired audio-text supervision nor SAE retraining. We evaluate this approach in steerable music retrieval and show that the recovered supports align more closely with concept-bearing audio examples and achieve a stronger trade-off between edit strength and preservation than alignment baselines, enabling more precise concept amplification and suppression with reduced drift on preservation metrics.

2. A Unifying Perspective on Audio Generative Modeling: Latent Representations and Modeling Strategies

Authors: Dongchao Yang

Categories: eess.AS Score: 6.84/10 (Obj:8 Id:5 Ind:8 Comp:8 Eff:5 Nov:6 Rep:5)

  • Strength: 提出依赖范围 (dependency horizon) 与条件模糊性 (conditional ambiguity) 两个诊断维度，成功将离散/连续、语义/声学等多组二分法解耦为统一框架，并通过7个代表性系统的系统比较展示了该框架的解释力。
  • Weakness: 全文无任何实验验证——框架提出的诊断测量（上下文消融曲线、条件模糊性代理）和受控评估协议（第7节）均停留在建议层面，未提供哪怕一个受控实验证明框架能产生正确的比较结论。
  • Full review: Claude Code 全文七公理审稿

Every audio generative system makes two coupled decisions: what representation to generate, and how to model its distribution. This paper organizes audio generative modeling around this coupling. For representation design, we compare discrete, continuous, and hybrid latents through four objectives: representation burden, distortion, empirical modelability, and streaming compatibility. For distribution modeling, rather than treating a latent’s difficulty as an intrinsic scalar, we use two diagnostic dimensions: dependency horizon, how far useful context extends, and conditional ambiguity, how much uncertainty remains after conditioning. These dimensions refine the common semantic-versus-acoustic intuition: variables with a long dependency horizon should receive global modeling capacity. Conditionally ambiguous detail may be delegated to a local or iterative generator. Applied to representative systems, this view shows that RVQ’s residual order gives ordered capacity but not ordered semantics, that AudioLM’s semantic-versus-acoustic cascade is one explicit placement of this boundary rather than a universal template, and that autoregression, iterative refinement, and hybrid designs differ chiefly in how they trade dependency horizon against critical-path generation cost. The distinction between discrete and continuous latents describes the output interface; dependency horizon, conditional ambiguity, and streaming determine how that interface should be modeled. Rather than cataloguing individual systems, we provide an evaluation and design framework for comparing representation-model pairs.

3. VoxZip: Semantic-Anchored Temporal KV Cache Compression for Long-Context Audio Inference

Authors: Wenxu Jia, Dongjie Fu, Xize Cheng, Fangming Feng, Linjun Li et al.

Categories: cs.AI, cs.SD Score: 6.53/10 (Obj:8 Id:6 Ind:7 Comp:6 Eff:7 Nov:5 Rep:8)

  • Strength: Train-free 两阶段框架在 Qwen3-Omni 上实现 20x KV 压缩仍保持 91.41% 长音频性能，端到端吞吐提升 1.93x、峰值内存降低 3.34x。
  • Weakness: 仅在单一模型家族 Qwen3-Omni 上验证，ASR 文本注入的增强效应与压缩效应未分离，且同期工作 AudioKV 已独立提出 ASR-guided audio KV compression 的核心 idea。
  • Full review: Claude Code 全文七公理审稿

Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks. Despite this progress, their long-context inference remains severely bottlenecked by prohibitive KV cache memory demands. Existing text-centric compression methods struggle here, often disrupting speech continuity or discarding crucial semantic cues. To address this, we propose VoxZip, a train-free, two-stage semantic-anchored KV cache compression framework. The first stage uses automatic speech recognition (ASR) transcriptions as explicit semantic anchors to temporally align, compress, and fuse audio tokens, significantly reducing the initial KV cache while elevating token information density. To further improve the compression ratio, the second stage employs a dynamic filtering strategy based on temporally decayed accumulated attention to evict non-essential tokens while mitigating early-token bias. Comprehensive evaluations on Qwen3-Omni across six diverse audio benchmarks demonstrate the superiority of our approach. VoxZip excels in long-audio reasoning and consistently maintains high-fidelity perception on short-form tasks. Notably, it sustains over 90\% of the uncompressed baseline performance even under an aggressive 20x KV cache compression in long-context scenarios. Furthermore, at a 4x compression ratio, VoxZip yields a 1.9x increase in inference throughput alongside a 3.3x reduction in peak memory overhead. Code and models will be available at https://github.com/MM-Speech/VoxZip.

4. AeroReformer2: Spoken-Query Referring Segmentation for Aerial Images

Authors: Rui Li, Chenxi Duan, Haoyang Yang

Categories: cs.CV Score: 6.50/10 (Obj:9 Id:6 Ind:5 Comp:5 Eff:8 Nov:6 Rep:9)

  • Strength: 首次定义并构建了遥感图像全句语音查询引用分割基准 VoiceAeroRef（52,466 三元组，8 种口音-性别 TTS，9 条件硬测试集），AeroReformer2-Swin-B 在 clean/hard test 上分别达 62.09%/54.09% mIoU，全面超越所有音频适配 baseline（+5.38/+5.63 点 vs LSCF），且内存最低。
  • Weakness: 语音和噪声均为 TTS/人工合成而非真实野外采集（论文自认 “should collect synchronized speech from analysts”），benchmark 生态效度受限；方法为已有组件（bilateral seg + kernel linear attn + token gating + wav2vec 2.0）的工程集成，缺少 1/4 refinement head 和 dual-scale vs sing。
  • Full review: Claude Code 全文七公理审稿

Spoken language offers a natural, hands-free interface for specifying an arbitrary target in dense remote-sensing imagery, yet existing referring remote-sensing image segmentation benchmarks accept only written expressions. To bridge this gap, we introduce \dataset, a spoken-query benchmark derived from RISBench that adds accent- and voice-diverse speech while preserving the original image, mask, and data splits. Its hard evaluation sets combine rotor, wind, and mixed interference with three signal-to-noise levels. We also propose \model, an efficient bilateral network that combines a boundary-preserving visual path with token-preserving speech encoding, kernel linear cross-modal attention, and a resolution refinement head. The design conditions visual features at two scales without materializing a dense speech–visual affinity matrix, then restores fine boundaries using high-resolution visual features. On the clean test split, \model with Swin-Base achieves 62.09\% mean intersection over union (mIoU) and 68.22\% overall intersection over union (oIoU), outperforming the strongest audio-adapted remote-sensing baseline by 5.38 and 2.08 percentage points, respectively. It retains the best hard-set mIoU at 54.09\%. To the best of our knowledge, this is the first benchmark and model study of full-sentence spoken-query referring segmentation for remote-sensing imagery. The code will be made publicly available.

5. Beyond Reconstruction: Full-Context Generative DiT for Music Generation

Authors: Yunjia Li, Menglin Wu, Junyu Dai, Xinyue Fan, Xiangang Li et al.

Categories: eess.AS Score: 6.42/10 (Obj:8 Id:8 Ind:5 Comp:6 Eff:8 Nov:6 Rep:5)

  • Strength: EMDC（错误率匹配+KNN采样的codec corruption训练策略）在合成corruption下将ViSQOL提升0.77，且在固定LM token的盲听中获得69.7%非平局胜率；完整系统在Artificial Analysis Leaderboard独立盲测中排名第3（Elo 1129, N=2289）。
  • Weakness: 代码、训练数据、tokenizer、上游LM和VAE均未开源，完整系统无法独立复现；完整系统比较缺少与同类学术hybrid系统（MeLoDy/Seed-Music/InspireMusic/SongBloom）的直接端到端比较。
  • Full review: Claude Code 全文七公理审稿

Hybrid music generators combine the long-range planning of an autoregressive language model with the fidelity of a diffusion- or flow-based acoustic renderer. Yet renderers are trained with clean, target-derived codec tokens but deployed with imperfect language-model predictions, creating codecinterface exposure bias. Rather than treating rendering as a simple reconstruction task,we formulate it as full-context generation from an imperfect discrete plan.We introduce FullDiT, a conditional DiT that fuses eight frame-aligned RVQ streams with independently encoded captions and lyrics and uses non-causal self-attention over the complete acoustic latent sequence. During training, Error-Matched Distractor Conditioning (EMDC) matches per-codebook replacement rates to teacher-forced top-1 error rates and samples near-miss tokens from cosine-KNN neighborhoods without changing the acoustic target. At inference, four-way classifier-free guidance (4-CFG) independently scales codec, lyric, and caption guidance increments. Matched ablations show that EMDC improves ViSQOL by 0.77 under synthetic corruption and is clearly preferred in non-tied comparisons with fixed languagemodel tokens. Further ablations show gains from full-song context and renderer-side text conditioning. The complete system outperforms five commercial systems on 15 of 18 automatic metrics and ranks among the top three on the Artificial Analysis Music with Vocals Leaderboard. The demo page is available at https://selinacloudl.github.io/fulldit-demo/.

6. Visual Token Codec: Unleashing Spatial Redundancy for ViT Feature Coding

Authors: Donghui Feng, Fengxi Zhang, Changsheng Gao, Wenhan Yang, Qi Wang et al.

Categories: cs.CV | 13 pages, 9 figures Score: 6.42/10 (Obj:9 Id:6 Ind:9 Comp:5 Eff:8 Nov:5 Rep:6)

  • Strength: VTC 通过分离 global/patch token 并在 patch grid 上使用 SCCTX 熵模型，在 DINOv2 和 SAM3 的五项下游任务上实现 15.7×–37.4× bitrate reduction（at 90% bypass performance），编码延迟从 LaMoFC-VTM 的 580s 降至 0.091s。
  • Weakness: 方法各组件（factorized prior、SCCTX、checkerboard、VBR）均直接借用已有图像压缩工作，缺乏 feature-matching loss 的独立消融和 SCCTX 子组件贡献分析，且未与 DT-UFC [24] 进行实验对比。
  • Full review: Claude Code 全文七公理审稿

Distributed deployment of large vision foundation models often partitions a ViT backbone and exchanges intermediate token features between computing nodes, making efficient feature compression critical under bandwidth and computation constraints. Existing ViT feature codecs typically flatten heterogeneous global and patch tokens into an L x C pseudo image, causing entropy models to mainly capture sequence-axis dependencies while overlooking the native two-dimensional patch-grid structure. In this paper, we show that ViT patch tokens retain strong local spatial correlations on the original grid. To exploit this structural prior, we propose the Visual Token Codec (VTC), a dual-path learned codec that separates global and patch tokens into dedicated coding paths. Global tokens are compressed with a lightweight factorized prior, whereas patch tokens are encoded on the patch-token grid using a spatial-channel context entropy model. To support intermediate-layer compression and practical rate adaptation, VTC further incorporates feature-matching supervision after subsequent ViT blocks and variable-rate modules within a single codec. Experiments on DINOv2 and SAM3 show that VTC consistently outperforms representative ViT feature coding baselines on classification, segmentation, and detection tasks. At 90% of uncompressed-feature performance, VTC reduces bitrate by 15.7x-37.4x across these tasks. We further provide intermediate-layer rate-utility analyses for practical transmission- and storage-oriented deployment scenarios.

7. Deferred Audio Pruning with Local Audio-Visual Dynamics for Omni-LLMs

Authors: Kyeongyoon Lee, Hongyeob Kim, Youngeun Kim, Sungeun Hong

Categories: cs.AI, cs.MM, cs.SD Score: 6.20/10 (Obj:8 Id:6 Ind:8 Comp:5 Eff:8 Nov:5 Rep:6)

  • Strength: A-PACK 在 Qwen2.5-Omni-7B/3B 上四个 benchmark 平均保持 95.5-98.8% 精度的同时减少 prefill FLOPs 78%、提升 decoding throughput 2.21×，baseline 覆盖全面且效果广泛。
  • Weakness: 方法由 6+ 个已有组件（CKA、DA-FPS、query attention pruning 等）组合而成，15 个超参数带来较高任意性，且未与同期 OmniDrop/OmniSIFT 等直接比较，新颖性增量有限。
  • Full review: Claude Code 全文七公理审稿

Omni-modal LLMs jointly process audio, video, and text, but long multimodal sequences incur substantial prefill and KV-cache costs. Existing omni-modal compression methods primarily focus on pre-LLM token reduction, leaving modality-specific compression across the LLM boundary underexplored. We propose A-PACK, a two-stage framework that defers audio pruning until query-conditioned multimodal interactions emerge. Our analysis shows that audio exhibits higher task-relevant information density and representational diversity per token than video. We further find that local audio-visual dynamics provide a more effective cue for visual selection than token-wise matching. We therefore preserve audio and compress video with local dynamics before the LLM, then progressively prune low-relevance audio and visual tokens and their KV-cache entries inside the LLM. Across four benchmarks on Qwen2.5-Omni-7B/3B, A-PACK achieves the strongest average performance among the evaluated prior methods while reducing prefill FLOPs by up to 78% and improving decoding throughput by up to 2.21x.

8. Multilingual Emotion Neurons in Large Audio-Language Models

Authors: Xiutian Zhao, Philipp Koehn, Björn Schuller, Berrak Sisman

Categories: cs.CL, eess.AS Score: 6.20/10 (Obj:8 Id:6 Ind:8 Comp:8 Eff:6 Nov:6 Rep:5)

  • Strength: 首次在4个LALM和12种语言上提供了LALM多语言情感编码的因果神经元级机制说明，显示CR-Fusion恢复了30-59%的语言不变情感成分，并在3/4模型的留出语言上优于最强单语基线达2-4.5 pp ESS。
  • Weakness: 缺少random-selection baseline定量对比、CR-Fusion使用8×50 instances vs monolingual 50 instances的预算不公平比较未讨论、Qwen2.5-Omni-7B为明确失败案例（deactivation ESS +0.58）、每语言单语料使”跨语言”结论实为”跨语料库”结论、代码未开源。
  • Full review: Claude Code 全文七公理审稿

Emotion is central to human communication, and its expression varies across languages. Large audio-language models (LALMs) achieve strong performance on multilingual speech tasks, yet it remains unclear whether they encode emotion through language-specific correlations or language-agnostic representations. We present the first neuron-level interpretability study of this question. We define Multilingual Emotion Neurons (MLENs) as functional units exhibiting stable emotional selectivity and aligned causal effects across languages, and introduce Consistency-Regularized Fusion (CR-Fusion) to identify them. Across four modern LALMs and 12 typologically diverse languages, emotion-sensitive neurons identified independently per language show minimal overlap, and additional monolingual identification data saturates quickly without isolating more transferable units, motivating identification from pooled cross-lingual evidence. Causal interventions demonstrate that MLENs identified by CR-Fusion provide more precise and transferable affective control than monolingual neuron sets in both zero-shot and low-resource settings. Leave-one-out ablations further reveal asymmetric transfer: individual identification languages, including low-resource ones, contribute non-redundant evidence, while several low-resource languages benefit most from the resulting cross-lingual transfer. Together, our findings provide the first causal, neuron-level account of how LALMs encode emotion across languages, and establish multilingual neuron identification as an effective mechanism for understanding cross-lingual affective behavior.

9. From Speech to Interaction: Analyzing Multimodal Systems in Cocktail-Party Scenarios

Authors: Thai-Binh Nguyen, Zhaolin Li, Jan Niehues, Alexander Waibel

Categories: cs.CL | Accepted at ICMI 2026 Score: 6.10/10 (Obj:8 Id:5 Ind:8 Comp:6 Eff:6 Nov:5 Rep:6)

  • Strength: 对 12 个 CHiME-9 MCoRec 系统构建统一 taxonomy，发现 overlap 不充分解释性能差异（cross-speaker confusion 31-53% vs ~14%），最佳系统达 57% 相对错误率降低。
  • Weakness: 缺乏组件级 ablation，核心反直觉发现仅基于 7 个 session 的定性分析，各系统详细实现依赖未发布的系统论文，且所有比较有意义的洞察（强 AVSR > 前端分离、LLM 聚类 > 时序聚类）均不反直觉。
  • Full review: Claude Code 全文七公理审稿

Humans have the remarkable ability to engage in spontaneous informal conversations and selectively attend to individual speakers while filtering out competing speech from nearby conversations. This “cocktail party” scenario still presents severe challenges to speech recognition systems. The CHiME-9 MCoRec task provides a testbed where systems must recognize groups of speakers and transcribe each of their conversations from audio-visual input. In this work, we analyze a diverse set of systems, representing different design directions for addressing the cocktail-party scenario, where the best system achieves up to 57% relative error reduction. We identify three main strategies: (1) explicit or implicit audio-visual target speech separation, (2) improved audio-visual speech recognition for each target speaker, and (3) the use of large language models to group speakers into conversations and enhance conversational consistency. Our analysis shows that these directions address complementary failure modes of the cocktail-party problem, and that high speech overlap alone does not explain performance differences, challenging the common assumption that overlap is the primary source of difficulty in cocktail-party recognition.

10. BAMU: Bitstream-Aware Marginal-Utility Allocation for Frozen Pretrained Neural Speech Codecs

Authors: Mingyu Zhao, Zijian Lin, Yutang Feng, Jiatao Chen, Fan Wang et al.

Categories: eess.AS | 5 pages, 2 figures, 1 table Score: 5.79/10 (Obj:8 Id:6 Ind:8 Comp:5 Eff:6 Nov:5 Rep:6)

  • Strength: 在冻结 EnCodec 上实现全码率 PESQ 一致提升（0.086-0.151）和 MOS 3.449→3.780（+0.331, p<10⁻⁵），且码流可精确解码、码率预算严格不超标。
  • Weakness: 缺少与最直接同类工作 VRVQ 的数值对比，DAC 低码率退化（depth-3 PESQ -0.0256），方法为经典 bit-allocation + learned predictor + 工程码流的组合，组件级新颖性有限。
  • Full review: Claude Code 全文七公理审稿

Pretrained neural speech codecs typically use a fixed residual vector quantization (RVQ) depth for all frames, ignoring temporal variation in quantization difficulty. We propose BAMU, a bitstream-aware dynamic RVQ allocation framework for frozen pretrained codecs. A lightweight, rate-independent predictor estimates frame- and layer-wise marginal latent-distortion reductions, while a constrained allocator selects prefix-valid depths under an exact serialized-size budget. Experiments on EnCodec and DAC over LibriSpeech, together with VCTK evaluation, show consistent EnCodec gains and DAC improvements mainly at medium and high rates. A 30-listener study confirms a MOS improvement from 3.449 to 3.780 over matched fixed-depth coding.

11. CuteTTS: Efficient and High-Quality Speech Synthesis via Autoregressive Modeling of Continuous Latents

Authors: Yuqian Zhang, Yao Shi, Kexin Huang, Botian Jiang, Zhe Xu et al.

Categories: cs.SD, cs.AI, cs.CL | 20 pages, 6 figures, 10 tables. Technical report Score: 5.47/10 (Obj:6 Id:8 Ind:5 Comp:5 Eff:6 Nov:5 Rep:2)

  • Strength: 0.2B 参数下 LibriSpeech WER 2.16% / SIM 78.9%，distill 模型 latency 降低 23.3% / RTF 降低 40.8%，在 compact Cont. AR TTS 类别中具有竞争力；系统消融设计严谨，变量隔离清晰。
  • Weakness: 核心组件均为已有方法组合（DiTAR 框架 + Semantic-VAE + ECAPA + guided distillation），novelty 增量有限；代码和 550K 小时训练数据均未开源，主结果无法独立复现；distillation 效果仅与自身 base model 对比，未与其他 distillation 方法公平比较。
  • Full review: Claude Code 全文七公理审稿

Zero-shot text-to-speech (TTS) now supports interactive assistants, personalized media, and accessibility tools. All TTS systems require faithful linguistic rendering, consistent speaker identity, and low-latency response. Yet compact streaming systems must preserve sufficient acoustic detail in a predictable low-rate latent sequence, while iterative diffusion sampling and classifier-free guidance multiply inference cost at every autoregressive step. To strike a balance between high-fidelity synthesis and low-latency inference, we present CuteTTS, a compact continuous-autoregressive TTS system. It combines semantically aligned causal VAE latents with patch-level autoregression, explicit speaker conditioning, and a bidirectional flow-matching head. We further introduce guidance-step distillation, which absorbs classifier-free guidance and multiple solver steps into a single interval-conditioned student. Evaluations on LibriSpeech and Seed-TTS-Eval demonstrate competitive intelligibility and speaker similarity in zero-shot voice cloning, while distillation lowers first-audio latency by 23.3% and real-time factor by 40.8% relative to the base model with comparable objective and subjective quality. These results provide a practical path toward continuous-autoregressive TTS that reconciles high-fidelity generation with the latency demands of real-time interaction.

12. Towards an LLM-based method for quantifying the sexual content in song lyrics

Authors: Ignacio M. Sticco

Categories: physics.soc-ph, cs.CL, cs.SD | 17 pages, 10 figures. Code, scoring prompt, and corpus: https://github.com/ignaciosticco/open-lyrics-scorer Score: 4.21/10 (Obj:6 Id:2 Ind:4 Comp:5 Eff:5 Nov:3 Rep:8)

  • Strength: 对 1,259 首 reggaeton 歌曲（该领域最大语料）的九维度 LLM 评分揭示 sexual explicitness 翻倍增长（0.15→0.31）且 Spotify flag 仅捕获 26.8% 的性显式歌曲，代码/语料/prompt 全部开源。
  • Weakness: 无任何 content-analysis baseline 对比实验（Zhang et al. 2024、Rospocher、Zamacola 等直接竞品均未在相同语料上比较），校准仅 10+5 首歌，方法新颖性被 Zhang et al. (2024) 等先前工作大量覆盖。
  • Full review: Claude Code 全文七公理审稿

Reggaeton is one of the most widely consumed music genres in the world, and its lyrics are commonly regarded as highly sexualized. This claim rests mostly on qualitative studies and on small-scale quantitative ones. This paper has two goals. First, we present a reproducible method that uses a large language model to quantify thematic content in song lyrics along several independent dimensions. The method is not restricted to sexual content. Second, we apply it to a corpus of 1,259 songs by 12 reggaeton artists released between 2002 and 2025. The analysis covers four topics: a dataset characterization, a per-artist comparison, an analysis of how the dimensions change over time, and a comparison between our sexual-explicitness score and Spotify’s own explicit flag. We release the data collection code, the scoring prompt, and the corpus, so that other researchers can replicate the approach or apply it to their own lyrics datasets.

13. Physics-Informed Learning for Robust Acoustic Localization with Calibrated Uncertainty

Authors: Jennifer N. Kampe, Changwoo J. Lee, Xin Shen, Ari Lehtiö, Sandro von Brandenburg et al.

Categories: stat.ML, cs.LG, cs.SD Score: 4.20/10 (Obj:8 Id:2 Ind:5 Comp:5 Eff:4 Nov:4 Rep:3)

  • Strength: Hybrid gate 框架在不恶化中位数误差的前提下，在森林模拟环境中将 p95 定位误差从 56.6m 降至 9.26m，并在 frozen-lake 实地数据上实现 GDOP-scaled conformal coverage 从 0.73 提升至 0.78（80% nominal）。
  • Weakness: 仅在一个简单环境（frozen lake, 3-mic array）进行了实地验证且改善有限（p95: 42.1→35.3m, p99 恶化），目标场景（forest）仅有模拟结果无实地验证，且未与 SRP/score-based 或任何学习类 baseline 比较，无任何 ablation 实验。
  • Full review: Claude Code 全文七公理审稿

Recent advances in Passive Acoustic Monitoring (PAM) offer an opportunity to obtain ecological spatial point-process data at unprecedented scale. However, realizing this opportunity necessitates the development of accurate and scalable localization methods. In real-world outdoor soundscapes, however, the assumptions underlying classical localization methods such as hyperbolic and score-based localization are routinely violated by multipath dominance, near-field effects, and complex propagation. Under these conditions, classical localization methods become brittle, with extreme errors possible even in small detection arrays. Rather than statistically replacing the underlying physics, we propose a method to refine it and increase robustness outside of ideal operating conditions: a learned model operating on physics-informed acoustic features corrects a fast hyperbolic solver where it produces implausible solutions, substantially reducing catastrophic worst-case errors while matching its median accuracy on field data. We further provide calibrated, geometry-aware uncertainty estimates suitable for propagation into downstream spatial models. Evaluating on distributed microphone arrays in real and simulated outdoor environments, we demonstrate that the proposed method yields robust, uncertainty-aware localization, providing a step toward scalable automated wildlife monitoring in complex acoustic environments.