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。

2026-08-28
日期2026-08-28
已评分—
均分—
最高—

Daily Papers — 2026-08-28

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

1. Exploring the Design Space of Representation Learning for Audio Transformations

Authors: Sungho Lee, Marco Martínez-Ramírez, Junghyun Koo, Wei-Hsiang Liao, Kyogu Lee et al.

Categories: cs.SD, eess.AS | Accepted to ISMIR 2026 Score: 7.90/10 (Obj:9 Id:9 Ind:8 Comp:9 Eff:8 Nov:7 Rep:5)

  • Strength: 统一框架下穷举三种对比目标（处理一致性/描述对齐/前向预测等变）的 7 种组合，检索超越最强基线 AFx-Rep 达 15.9pp（61.1% vs 45.2%），并给出双嵌入互补的可操作设计准则：距离型任务用变换嵌入（源泄漏 67.2% vs Fx-Encoder 家族 89%），探针型任务用处理后音频嵌入（链估计 IoU 53.8）。
  • Weakness: 关键的受控比较依赖自建处理器库与 1.5M 预渲染子批训练管线，但承诺的代码仓库 SonyResearch/RLAT 截至 2026-09-05 仍为 404 无法核实；参数估计仅与基线持平（MAE×100 约 17.5 vs 19.1），且检索与风格迁移均无人类主观评测佐证。
  • Full review: Claude Code 全文七公理审稿

Neural audio representation learning has enabled a range of content-oriented applications, but the resulting features remain limited for tasks involving audio processing. Furthermore, it is not obvious what processing-aware representations should capture: the processing itself, abstracted away from source content, or the processed audio that retains it. Existing approaches implicitly commit to one or the other and also differ in their models, data, and evaluation, obscuring which design choices drive their behavior. We address both questions within a unified framework of three objectives: processing consistency, description alignment, and equivariance via forward prediction. We compare all combinations of the objectives under a controlled setup and reveal their relative strengths and interactions. Our framework produces both a transformation embedding and a processed-audio embedding, and we find that the two play complementary roles: distance-based tasks favor the former, while probe-based tasks favor the latter. Combined with improvements in network architecture and training pipeline, our representations outperform prior baselines across retrieval, probe-based evaluation, and style transfer.

2. Phoneme- and Word-Level Metrics Using Self-Supervised Speech Representations for Forced Alignment Evaluation

Authors: V. S. D. S. Mahesh Akavarapu, Michael Daniel, Gerhard Jäger

Categories: cs.CL | Accepted at EMNLP-2026 (Findings) Score: 7.84/10 (Obj:9 Id:10 Ind:6 Comp:8 Eff:16 Nov:16 Rep:9)

  • Strength: 提出首个基于 SSL 表征的 reference-free 对齐评测指标 PCMI/WACS，与 gold AAS 相关达 r=−0.78（p<0.001），跨 85 种 FLEURS 语言 + 45 种 DoReCo 语言 + Archi/Rutul 验证，代码/模型/数据全开源（Apache-2.0，EMNLP-2026 Findings）。
  • Weakness: 对齐器与评测指标共享 MMS backbone 的循环偏置未正面讨论，gold 标注由第二作者完成且每语言仅约 7 分钟测试音频，指标对静音吸收欠惩罚（AAS 88.5→46.3 ms 时 PCMI 仅 0.306→0.319），标注者间一致性上限缺失使 r≈0.78 缺少绝对参照。
  • Full review: Claude Code 全文七公理审稿

Forced alignment evaluation typically requires manually annotated timestamps, limiting large-scale and multilingual analysis. We introduce two corpus-level metrics based on self-supervised (SSL) speech representations for reference-free forced alignment evaluation: Phoneme-Cluster Mutual Information (PCMI) and Word Acoustic Consistency Score (WACS). PCMI measures agreement between aligned phoneme labels and clusters induced from SSL-speech representations, while WACS measures consistency of repeated word realizations using dynamic time warping similarity between word representation sequences. Using both random and systematic perturbations, we show that PCMI and WACS degrade consistently under alignment perturbations. We further analyze the metrics across multiple alignment systems on 85 languages from FLEURS, validate them against manually annotated alignments from 45 languages in DoReCo, and evaluate them on two phonologically complex low-resource languages. The metrics effectively separate high- and low-quality alignments and correlate strongly with timestamp-based alignment quality measures. Our results demonstrate that SSL-speech representations enable scalable, reference-free forced alignment evaluation. The metrics are available as an open-source Python package at https://github.com/mahesh-ak/forced-aligner-metrics.

3. Multirate State Space Models for End-to-End Processing of Pulse Density Modulated Speech Signals

Authors: Ludovic Boulanger, Sean U. N. Wood

Categories: cs.SD, eess.AS Score: 7.40/10 (Obj:9 Id:8 Ind:6 Comp:8 Eff:7 Nov:8 Rep:5)

  • Strength: 单一 16 kHz PCM 训练的 SSM 编码器在 128 kHz–2 MHz 全部 PDM 采样率下保持 >92% KWS 准确率（降幅 <4%），2 MHz 下增强 PESQ 3.08/STOI 0.93 与 PCM SOTA 持平，SSM 输出可免抗混叠下采样 65,536 倍且增强延迟仅 4.5 ms（满足助听器 <6 ms）。
  • Weakness: 全部 PDM 结果来自软件合成 ΣΔ 数据，无真实麦克风/嵌入式硬件功耗验证，未与先前 PDM 方法做同条件直接对照，且声明开源的代码仓库（NECOTIS/ssm-speech-processing）实查为空、仅含 24 字节 README。
  • Full review: Claude Code 全文七公理审稿

Deep neural networks (DNNs) based on state-space models (SSMs) are increasingly applied to speech processing, but typically operate on pulse-code-modulated (PCM) audio. This constrains deployment on low-power, always-on edge devices, which commonly use single-bit pulse-density-modulated (PDM) micro-electromechanical (MEMS) microphones for their noise robustness, low cost, and variable sampling rates that enable low-power operation. In fact, converting PDM to PCM requires low-pass filtering and decimation, imposing costly overhead on resource-constrained hardware. While prior works have attempted to process PDM signals directly, they require long training times and generalize poorly across sampling rates. In this paper, we show that the SSM has two key properties that remediate these issues: its continuous-time parametrization allows it to produce a consistent representation of the input audio signal, regardless of the modulation strategy and sampling rate, and its long-term memory enables this representation to be aggressively downsampled without needing any anti-aliasing operations. We then propose a novel end-to-end PDM speech processing architecture that uses an SSM to encode the input audio signal into a modulation- and sampling-rate-invariant latent representation. We show that our proposed architecture achieves robust speech classification and enhancement gains at low-power sampling-rates (512 kHz) and similar performance to state-of-the-art algorithms operating on PCM data when tested on standard PDM sampling-rates of 2 MHz. Moreover, we show that the SSM’s output can be downsampled by more than 65,000 times, thus significantly reducing the number of processing timesteps in downstream layers.

4. MuSP-Bench: Advanced Multimodal Benchmarking of Music Understanding across Score and Performance

Authors: Milan Liessens Dujardin, Song-Ze Yu, Kevin Miao

Categories: cs.MM, cs.SD Score: 7.30/10 (Obj:9 Id:5 Ind:8 Comp:8 Eff:8 Nov:8 Rep:5)

  • Strength: 首个系统覆盖乐谱-演奏跨模态推理的人工编写音乐基准（490 题、18+6 部真实作品、四维标注），量化揭示了前沿模型演奏音频理解崩溃（最强 35.7%、最差 0.9%）与 S&P 联合推理在音频输入下骤降（30.6%→6.9%）的明确短板。
  • Weakness: 460 题开放式答案的确定性归一化判分规则未公开未校验、无人类对照组天花板、无标注者间一致性或 Limitations 章节，评测 harness 与闭源模型推理参数未发布导致第三方复现受阻。
  • Full review: Claude Code 全文七公理审稿

Musicians commonly communicate music through scores and performances. Scores encode musical intent, while performances realize it in sound. To investigate whether models can meaningfully engage with both modalities, we introduce MuSP-Bench, a human-authored benchmark of 490 questions targeting understanding across Musical Scores and Performances. The benchmark distinguishes itself by spanning score-based, performance-based, interpretive, and long-horizon reasoning across classical piano and orchestral works. We evaluate frontier multimodal large language models under multiple input conditions. Our results show that these models struggle substantially to understand scores, while facing even greater challenges when reasoning about performance audio. The benchmark is available at https://musp.vaclis.net/.

5. VoiceCodeBench: Evaluating Exact Structured-Token Recovery in Automatic Speech Recognition

Authors: Tyler Baumgartner, Brandon Tai, Lisa Kaelin-Martin, Candice Fan, Luc Debaupte et al.

Categories: cs.CL | 11 pages, 1 figure, 9 tables Score: 7.21/10 (Obj:9 Id:7 Ind:7 Comp:8 Eff:7 Nov:7 Rep:6)

  • Strength: 12 个商业 ASR 系统上量化了 WER 的信息边界（WER 与 CTEM/TSR 的 Spearman 相关均为 -0.73），最强系统 TSR 仅 68.7%，59.6% 实体失败集中于 URL/命令/邮箱/文件路径/邮政地址五类；300 段真人录音、1,482 个审计实体的数据集已在 HuggingFace 公开（MIT）。
  • Weakness: 基线偏闭源 API 且与既有实体敏感度量（Semantic Distance 等）无对比，未接入真实下游系统验证 CTEM/TSR 与实际解析失败的相关性，GitHub 未检索到官方评估代码且评分依赖会漂移的 GPT-5.5 验证器（人工审计仅覆盖 1.1% 决策）。
  • Full review: Claude Code 全文七公理审稿

Automatic speech recognition (ASR) systems are commonly evaluated with word error rate (WER), yet many voice workflows depend on exact written values for identifiers, paths, and measured quantities. A transcript can appear fluent and achieve low WER while corrupting a value that a downstream system must parse, store, or execute. We introduce VoiceCodeBench, a benchmark for evaluating exact structured-token recovery in English ASR. It contains 300 human-recorded workplace segments spanning eight workflow domains and 1,482 audited target entities across 26 entity types, each with a canonical written form recoverable from the audio. Under a raw-audio-only protocol, systems receive audio bytes without additional context or metadata. Alongside WER, we evaluate Canonical Token/Entity Match (CTEM), Task Success Rate (TSR), and per-type exact recovery. Across 12 baseline ASR systems, lower WER generally corresponded to better structured-token recovery but did not fully determine it: Spearman correlations were -0.73 for both WER versus CTEM and WER versus TSR. The strongest baseline by TSR reached only 68.7%, leaving nearly one third of recordings with at least one unrecovered workflow-critical value. These results show that entity-sensitive metrics are needed to assess whether ASR output preserves exact values that production systems must parse, route, store, compare, or execute.

6. Auditing Generative Audio Calls for Known-Task Audio-LLM Evaluation

Authors: Mengzhe Geng

Categories: cs.SD, cs.CL, eess.AS Score: 7.05/10 (Obj:8 Id:8 Ind:6 Comp:8 Eff:7 Nov:7 Rep:5)

  • Strength: 受控调用审计设计严谨，VocalSound holdout 上 no-call 选择器 0.921 与带生成调用的选择器 0.925 配对差仅 0.004（95% CI [−0.025, 0.033]），并用 WavLM 探针 0.854（全程 2.5s、零调用）与 Holm 校正的 8 组配对检验支撑”已知闭集任务上生成式调用无显著边际价值”的结论。
  • Weakness: 主结论仅基于 240 条 holdout 且 ESD 外部验证结论模式不一致（stack 融合 Δ=0.075 为正、agree 反转为负），生成模型仅 3 个，全文无公开代码仓库与冻结输出，selective prediction/级联推理相关文献未做系统对照。
  • Full review: Claude Code 全文七公理审稿

Speech and audio LLMs are often evaluated by asking whether a waveform prompt beats an automatic speech recognition (ASR) transcript. For known closed-set tasks, that comparison conflates two factors: access to acoustic evidence and the need to call a generative audio model. We evaluate this distinction as a controlled call-decision problem. For each example, a policy chooses among keeping a transcript label, using encoder evidence from Contrastive Language-Audio Pretraining (CLAP), Audio Spectrogram Transformer (AST), or WavLM, and calling Qwen2-Audio, Qwen2.5-Omni, or MOSS-Audio; the decisive ablation removes all generative actions while keeping the selector and development protocol fixed. On VocalSound, transcripts reach 0.296 accuracy, so waveform information is needed. Yet supervised CLAP and WavLM controls reach 0.850 and 0.854 with no generative audio calls. A selector with generative actions reaches 0.925 accuracy using 12.5% calls, compared with 0.921 for the matched no-call selector (paired difference 0.004; 95% CI [-0.025,0.033]). Agreement and stacking features improve weaker selectors but do not beat the strongest no-call control. For known-task endpoint claims, the relevant quantity is the marginal value of the generative call after transcript and encoder evidence have already been used.

7. Is Prosody Lost in Translation? Fine-Grained Cross-Lingual Prosody Similarity Across Languages

Authors: Haopeng Xie, Ismail Rasim Ulgen, Sofia Son, Berrak Sisman, Philipp Koehn

Categories: cs.SD, cs.CL, eess.AS | Accepted to EMNLP 2026 Findings Score: 6.80/10 (Obj:9 Id:8 Ind:6 Comp:8 Eff:6 Nov:7 Rep:4)

  • Strength: 首个基于专业配音数据的词级跨语言韵律分析（三语言对约 110 小时），音高相关 0.229–0.246、时长相关 0.884–0.891 均显著超过打乱基线（p < 10⁻⁸⁰），并定位名词区域为韵律保留最强锚点（能量效应 +0.0355–0.0402）。
  • Weakness: 发现止步于相关性观察、无任何 S2ST 系统级验证实验；核心配音数据因版权不可公开且截至审稿日无官方代码仓库，关键数字不可复现，三语言对覆盖使跨类型学泛化无法检验。
  • Full review: Claude Code 全文七公理审稿

Prosody plays an important role in speech translation, conveying information such as emphasis, emotion, and intent beyond lexical content. However, despite recent progress in expressive speech-to-speech translation (S2ST), little is known about how prosodic patterns are similar/different across languages. Understanding these cross-lingual similarities and differences is crucial for effectively incorporating prosody into expressive S2ST systems. In this work, we present the first fine-grained cross-lingual analysis of prosody using multilingual dubbing data across English-German, English-Spanish, and English-French language pairs. We analyze the similarity of pitch, energy, and temporal feature patterns between source and target speech and investigate the linguistic and alignment-related factors affecting this similarity. Our analysis reveals inherent cross-lingual correlations in prosodic structure between certain languages. The findings provide important insights into the transferability of prosody across languages and offer empirical guidance for future expressive speech-to-speech translation systems.

8. A Shaky Voice Is Not Always a Dodge: Benchmarking Textual and Vocal Evasion Detection in Earnings Calls

Authors: Mirae Kim, Seonghun Jeong, Youngjun Kwak

Categories: cs.CL, cs.SD | EMNLP 2026 Findings Score: 6.70/10 (Obj:8 Id:6 Ind:8 Comp:8 Eff:6 Nov:8 Rep:3)

  • Strength: 首个财报电话会文本+声音双标签回避基准（505 QA 对 / 60 会，文本 κ 0.866、跨模态不一致率 32.1%），并以 speaker-centering 实验证明 audio-LLM 依赖绝对声学量——centering 后强模型相关从 0.195 跌至 0.032，声音侧最佳模型 unconfident F1 仅 39.2 远低于文本侧 GPT-5 的 87.4。
  • Weakness: 505 对规模小且最佳修复手段收益微弱（speaker-normalized prompting 仅 +1.0~+4.7 F1），音频数据受商业许可约束、GitHub 无代码发布（搜索 0 结果），关键实验结果无法独立复现。
  • Full review: Claude Code 全文七公理审稿

Existing approaches to evasion detection in earnings calls focus on textual transcripts, treating evasion as a single-dimensional phenomenon. We argue that evasion in spoken communication is inherently multidimensional: beyond what executives say, how they say it carries independent and complementary information. To study these dimensions jointly, we introduce DualEvasion, a benchmark for evasion detection across text and audio in earnings call Q&A. The benchmark contains 505 annotated question-answer pairs from 60 earnings calls, each with two independent labels: textual evasion (direct vs. evasive) and vocal cues operationalized as speaker confidence (confident vs. unconfident). Our experiments show that state-of-the-art multimodal models struggle to detect vocal confidence, particularly on unconfident responses. Our analysis suggests these models interpret acoustic cues in isolation rather than relative to each speaker’s baseline. Providing speaker-level references yields modest improvements, but a substantial gap with human performance remains.

9. Effects of HRTF Augmentation on Predicted Spatial Release from Masking in Music

Authors: Jack Webb, Christophe Lesimple, Volker Kuehnel, Lorenzo Picinali

Categories: eess.AS | Accepted for publication in the Proceedings of Forum Acusticum 2026 Score: 6.42/10 (Obj:8 Id:6 Ind:7 Comp:8 Eff:7 Nov:5 Rep:5)

  • Strength: 首次用解构对照（ILD/ILD+ITD）把 HRTF 空间线索增强的音乐 SRM 预测增益归因到频依赖 ILD 放大，N0 下 HRTF₄.₀ 比个体 HRTF 平均 +3.08 dB，且诚实报告听损下缩水至 1.05 dB、WDRC 无法恢复（Δ=+0.01 dB）。
  • Weakness: 全文无任何行为验证，两个听觉模型在 ITD 贡献上分歧达 30 个百分点（+59% vs +29%），且无代码/数据开放，25/405 HRTF 评测池的个体变异性未报告。
  • Full review: Claude Code 全文七公理审稿

Separating individual musical instruments within a complex mixture of sounds poses a persistent challenge for listeners with hearing loss. Although spatial separation of sources improves speech recognition in this population, the potential benefits of spatial cue enhancement for music perception remain largely unexplored. This paper introduces a method to increase spatial cue salience through the augmentation of individual head-related transfer functions (HRTFs). Auditory model analyses indicate that augmented HRTFs may enhance the separability of musical instruments relative to individual HRTFs. Predicted benefits persist when moderate sensorineural hearing loss is modelled, though they are substantially reduced. Simulated hearing aid processing does not restore these benefits to normal-hearing levels.

10. No Detectable Change in Side-Level WER from Prompt-Level Context: A Preregistered Ablation on a Production Oral-History Corpus

Authors: Theodore O. Cochran, Stephanie Dodson, Keith Nore

Categories: cs.CL, cs.AI, cs.SD | 31 pages, 1 figure. Preregistered on OSF (https://osf.io/ns49b, DOI 10.17605/OSF.IO/NS49B); release materials and dated provider-documentation snapshots in the study component (https://osf.io/9nsvr) Score: 6.40/10 (Obj:8 Id:6 Ind:5 Comp:8 Eff:6 Nov:7 Rep:5)

  • Strength: 预注册+哈希冻结+实现审计的零结果评测：19 面/10.6 小时劣化磁带上 full 提示对 gpt-4o-transcribe 的 side-level WER 配对差仅 +0.6 点（[-1.1,+1.0]，p=0.57），四项预注册假设全不显著，并量化出配置选择杠杆（gpt-4o 比 gemini-2.5-flash 好 17.8 WER 点，p<0.0001）与列表 token 错误 −1.7 点但仅折合 0.08 WER 点的。
  • Weakness: 每格单次转录的重跑变异（配对 ΔWER 范围 [-10.2,+25.0]）大于被检验效应量，19 面聚类于 8 位受访者（一个家庭占 9 面）致有效 n 更低，Gemini 侧重跑区间 [-25.5,+14.6] 无信息量，私有语料+无 GitHub 代码仓库+商业 API 漂移使第三方复现近乎不可行。
  • Full review: Claude Code 全文七公理审稿

Supplying context at inference time to a large multimodal model is an inexpensive lever for adapting speech transcription to a domain, and earlier results on smaller models reported large gains. This work tested that mechanism where it ships, in the prompt-conditioning layer of a production oral-history transcription tool, on a sample from its own production corpus. Full prompt-level context did not detectably change side-level word error rate (WER), and none of the four preregistered hypotheses was supported. The design was a within-item paired ablation, preregistered with the analysis code frozen by hash before the confirmatory batch was scored; two disclosed gpt-4o pilot sides had been scored earlier, during scorer development. Nineteen cassette sides, about 10.6 hours of degraded 1970s-80s interview audio, were reprocessed through the production code path under three prompt arms, crossed with two deployed commercial configurations, gpt-4o-transcribe and gemini-2.5-flash, and scored against operator-corrected verbatim references. For gpt-4o-transcribe the median paired difference between the full-context and no-context arms was +0.6 WER points, with a side-resampled interval of [-1.1, +1.0]; the Gemini estimates were too unstable to support a comparable negative inference. A post-hoc rerun found run-to-run pipeline variability larger than the confirmatory differences, so effects of that size cannot be resolved from one transcription per cell. An implementation audit verified the manipulation was live, and sequence-alignment analysis found a small improvement on complete context-listed phrases, too small to materially change side-level WER, and for Gemini coexisting with worsened unlisted-token error. Evaluating context mechanisms therefore requires sequence-aligned term-level, insertion, and speaker-label measures alongside aggregate accuracy.

11. Low-Power End-to-End Cochlear Implant Speech Denoising with Spiking Neural Networks

Authors: Ludovic Boulanger, Sean U. N. Wood

Categories: cs.SD, cs.NE, eess.AS Score: 6.30/10 (Obj:8 Id:5 Ind:5 Comp:8 Eff:6 Nov:6 Rep:7)

  • Strength: 首个将 SNN 用于端到端 CI 降噪+编码的工作：在同容量预算（437k vs 552k 参数）下以 VSTOI 56 vs 57 (%)、SNRi 6.0 vs 6.1 dB 的近持平性能实现能耗 2461→372 μJ/s（>6× 降低），平均脉冲率 ~250 spikes/s 证实稀疏性，代码与数据公开。
  • Weakness: 能耗结论依赖未含内存传输成本的 CMOS 45 nm 理论估算（而文献指出内存占能耗大头），SNN 版做 HPO 而基线未调参且三项架构改动未消融，无真实 CI 用户主观听感实验，6× 数字与性能归因均未闭合。
  • Full review: Claude Code 全文七公理审稿

Cochlear implants (CI) restore hearing for individuals with severe to profound hearing loss. However, CI users often struggle to understand speech in noisy environments. Deep neural networks (DNN) have shown promise in enhancing speech for CI users, yet their high energy demands make them non-ideal for low-power CI processors. Spiking neural networks (SNN), on the other hand, offer comparable performance with significantly lower energy consumption. Hence, we propose a novel SNN inspired by the Deep ACE architecture that simultaneously performs speech enhancement and CI coding. Our model achieves competitive vocoded short-time objective intelligibility (VSTOI) and signal-to-noise ratio improvement (SNRi) scores compared to Deep ACE, while achieving more than a sixfold reduction in energy consumption.

12. When Robots Mishear Us: Mapping the Safety Risks of Voice-Controlled Embodied AI

Authors: Sihan Jia, Oliver Lemon

Categories: cs.AI, cs.CL, cs.RO Score: 6.30/10 (Obj:8 Id:5 Ind:5 Comp:8 Eff:6 Nov:7 Rep:5)

  • Strength: 首次系统刻画自然 ASR 错误对具身 AI 安全的影响：声学替换使 Qwen2.5-7B 在 POEX 上接受率从 55.33% 升至 60.00%，Severe 噪声（WER 80%）升至 67.33% 并产生最多 27 例安全翻转，且揭示纠错仅对声学类错误有效（AR 60.00→53.33）、对重度噪声无效。
  • Weakness: ASR 错误全靠 GPT-4 模拟生成而未经真实 ASR 系统输出分布验证，全文无统计检验或置信区间（136 条指令上 4-12 点差异），核心结论仅基于单一 7B 模型且未发布代码、提示词与错误样本集。
  • Full review: Claude Code 全文七公理审稿

We investigate whether automatic speech recognition (ASR) errors in user input can lead to unsafe outputs from Embodied AI (EAI) models. We find that ASR errors can lead to harmful instructions being accepted and executed by EAI models, thereby reducing safety. We simulate ASR errors and combine them with existing safety benchmarks (SafeAgentBench and POEX) to evaluate how different errors affect embodied AI safety. We find that some of them preserve semantic structure but increase harmful ambiguity, while others weaken the model refusal behaviour and allow unsafe plans to be generated and executed. We show that in some cases automatic correction of ASR errors can reduce the risk, but this is not always effective. Overall, we show that ASR errors lead to significant safety risks for embodied AI.

13. VocalAffectBench: Evaluating Vocal Emotion Recognition in AI Audio Models

Authors: Luc Debaupte, Tyler Baumgartner, Brandon Tai, Candice Fan, Bill Wang et al.

Categories: cs.CL, cs.SD | 6 pages, 6 tables. Benchmark, baseline predictions, and aggregate results are publicly available Score: 5.90/10 (Obj:8 Id:5 Ind:5 Comp:6 Eff:6 Nov:5 Rep:8)

  • Strength: 公开 MIT 许可 test-only 基准（280 条 7 类情感音频、2.32h、逐条预测与评分脚本全发布），量化出商用音频模型情感识别的系统性 neutral 偏置——neutral 召回 76.2% 而 fearful/surprised 仅 15.0%，最强基线 gemini-3.5-flash 也只达 44.3%（随机 14.3%）。
  • Weakness: 标签由单审核人单次核验、无评分者间信度、无人类表现锚点，且未引用 VoxEmo（arXiv 2603.08936）等直接竞品；摘要与全文数字不一致（273/46.5% vs 280/44.3%），闭源模型标签映射由作者单方决定，模型间排序缺统计可分性。
  • Full review: Claude Code 全文七公理审稿

Voice products increasingly need affective cues that are present in speech but absent from transcripts. We introduce VocalAffectBench, a public, test-only benchmark for evaluating whether AI audio models can identify expressed vocal emotion from raw audio. The benchmark contains 273 human-recorded English WAV clips from 51 speaker accounts totaling 1.95 hours across seven labels: angry, disgusted, fearful, happy, neutral, sad, and surprised, with 39 clips per class. All baselines are evaluated from audio alone, without transcripts or contextual metadata. Across six released baselines, average accuracy is 35.5%. The strongest baseline, gemini_3_5_flash, reaches 46.5% on the seven-way task, above the 14.3% random baseline but far from robust emotion recognition. A secondary valence-bucket analysis maps labels into positive, neutral, and negative classes, excluding surprised because its valence is ambiguous. Aggregate accuracy under this coarser view is 50.9%. Performance is highly uneven across classes. By recall, neutral is identified most reliably at 75.6% averaged across baselines, while surprised and fearful reach only 10.7% and 15.4%, respectively. These results show that the evaluated baselines can extract some affective signal from speech, but discrete expressed-emotion recognition remains fragile, especially for non-neutral emotions that are often most important in voice agent workflows.

14. Evoking Harmony via Convolution

Authors: Michael Gogins

Categories: cs.SD | Extended version of a paper to appear in the Proceedings of the 8th International Csound Conference(ICSC 2026), Trapani, 20–22 October 2026 Score: 5.90/10 (Obj:9 Id:6 Ind:5 Comp:8 Eff:5 Nov:6 Rep:3)

  • Strength: 将和弦音级唤起压缩为一次分区卷积的加窗正弦颗粒 IR，时频折衷量化至 1.68 半音/八度（T=30 ms）、能量泄漏 <3%、A 计权拟合 0.02 dB，并有 ICSC 2026 收录的完整伪代码。
  • Weakness: 全部证据为主观频谱图与作者听感，无感知评测或参数扫描；声称在线的代码与音频材料实测在 GitHub 材料目录中完全缺失，且仅 5 条引用（含 2 条自引）使新颖性无从核查。
  • Full review: Claude Code 全文七公理审稿

I show how to evoke the pitch-class content of a chord from an arbitrary source sound by convolving the source with an impulse response whose grains are one windowed sinusoid per pitch-class, across each octave of hearing range; while, at the same time, minimizing artifacts. A Csound user-defined opcode, chord_convolver, mixes a dry Dirac component into that response, and applies partitioned convolution once. I contrast the effect with a linear-frequency comb filter and with a generic constant-Q resonator bank, and I demonstrate musical use on a twilight field recording alongside the ruins of Chateau de Lagarde.

15. Predicting Turn-Taking Outcomes in Multi-Party Conversation: Interpretable Modelling of Speech and Gaze Dynamics with Interpersonal Closeness

Authors: Mark Dourado, Karim Haddad, Henrik G. Hassager, Stefania Serafin

Categories: cs.CL, cs.SD | Part of an industrial PhD collaboration between GN Group and Aalborg University Score: 5.89/10 (Obj:8 Id:6 Ind:6 Comp:8 Eff:6 Nov:5 Rep:3)

  • Strength: 在四人多边形对话（GaMMA 语料库）中用可解释弹性网逻辑回归预测话轮转换 gap/overlap，多模态 ROC AUC 0.76±0.04，相对仅注视基线提升 +0.18（95% CI [0.16, 0.19], p<0.001），且注视线索在 65 dB 噪声与变化声学条件下性能不降反稳，验证了注视作为抗噪补充话轮线索的价值。
  • Weakness: 代码因 IP 限制完全未公开，验证 0.76 AUC 需从 Zenodo 原始数据重写整条 pipeline；缺少与任何已有话轮结束预测方法的外部基线对比，gaze-only 基线 F1=0.66 靠全判 overlap 实现，框架/数据/操作化均沿自同组前作，新颖性限于场景扩展且 IOS 边际贡献未做单独消融。
  • Full review: Claude Code 全文七公理审稿

Smooth speaker transitions are fundamental to effective conversation and rely on an interlocutor’s ability to predict when to enter the conversation. This ability depends on accurately interpreting and expressing the verbal and non-verbal cues that signal when a speaker wishes to take or relinquish the floor. The process becomes even more complex in noisy, natural, multi-party settings, with multiple interlocutors available. This study models how gaze and speech, together with perceived interpersonal closeness, signal conversational floor changes in free four-person dialogue. Using the GaMMA corpus, we trained logistic regression models using interpretable, behaviourally motivated features extracted before each turn-taking event to classify floor-transfer outcomes as gaps or overlaps. Predictors included gaze features such as transition motifs and behavioural contrasts, entropy, gaze-based addressee identity, and mutual gaze, alongside speech features derived from speaker loudness, as well as perceived interpersonal closeness (IOS) between speakers. Results show that gaze features capture predictive structure, and that combining them with loudness improves performance (ROC AUC = 0.76 +- 0.04). Loudness reflected speaker control, while gaze dispersion and addressing indexed listener readiness and competitive entry. Performance remained robust across noise conditions, indicating that gaze provides a complementary, noise-resilient cue to turn-taking dynamics.

16. Klangfarbenakkord and Klangfarbenharmonien Metric Space Models for Music on Informational Geometry 1

Authors: Yusei Tamura, Shigekazu Ishihara, Ken Ito

Categories: cs.SD, cs.MM | 21 pages, 15 figures Score: 4.50/10 (Obj:6 Id:4 Ind:4 Comp:6 Eff:8 Nov:12 Rep:2)

  • Strength: 将归一化频谱视为概率密度、用一维 Wasserstein 距离（量纲 Hz）量化音色亲和度，W1=442 Hz 中分离出净音色形状变化 89 Hz，并以该指标客观确认了维也纳爱乐巴松传统指法对单簧管 Throat G 的融合补偿、协作设计出 13 个新双簧管指法。
  • Weakness: 全文无任何与 MFCC/KL/欧氏距离等替代度量的对比、无感知听测验证（论文自承”单听无法分辨”）、无代码无数据无录音协议，核心结果仅以无数值表格的图呈现，可复现性接近于零。
  • Full review: Claude Code 全文七公理审稿

This paper deals with the introduction of “geometric harmony”, a discipline that explicitly addresses the spectral characteristics of musical gamut. The framework of Western music, from Renaissance to the present, represents sound in terms of “pitch”-as is evident from its five-line staff notation system-and employs the fundamental frequency as its representative value, 440 Hz, etc. In this paper, by taking the timbres of specific individual instruments as elements and examining the Wasserstein distance between two voices, and Wasserstein deviations between three or more voices, we demonstrate that it is possible to expand the system whilst retaining the entire framework of conventional music theory. At the same time, as an example of practical utility in ensemble playing, we provide a detailed account of the two-voices affinity of the “Throat G” on clarinet, a note known for its fragility in ensemble contexts.