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。

2026-08-15
日期2026-08-15
已评分—
均分—
最高—

Daily Papers — 2026-08-15

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

1. Prototype-Rectified Iterative Self-supervised Manifold Denoising under Severe Acoustic Shift

Authors: Ashish Anand Shukla, Rini Smita Thakur, Aryan Das, Vinod K. Kurmi

Categories: cs.SD, cs.LG | Accepted as a full paper at ACM CIKM 2026 Score: 8.11/10 (Obj:9 Id:6 Ind:8 Comp:8 Eff:9 Nov:8 Rep:9)

  • Strength: On UrbanSound8K, PRISM lifts accuracy from the 58.77% zero-shot CLAP baseline to 71.71% (+12.94 pp) and beats the oracle-prompt ContextDA by 9.41 pp using a training-free, 0.0009 ms single matrix-vector multiply per samp。
  • Weakness: The paper motivates against gradient-based TTA but reports no TENT/TPT accuracy numbers on its benchmarks, and its headline CAR fix trades overall US8K accuracy (71.23 vs 71.71) for a worst-class gain.
  • Full review: Claude Code 全文七公理审稿

Audio-Text Foundation Models (ATMs) fail catastrophically under severe acoustic noise, yet existing adaptation strategies either rely on gradient-based Test-Time Adaptation (TTA), which reinforces noise rather than signal, or on prompt tuning that requires privileged noise annotations unavailable at inference. We address these failures with PRISM (Prototype-Rectified Iterative Self-supervised Manifold Denoising), a training-free, source-free TTA framework grounded in the Affine Noise Hypothesis: severe acoustic noise induces a low-rank affine shift in the multimodal latent space, with more than 90% of distortion energy confined to the leading 60 principal components. PRISM estimates and reverses this distortion from an unlabeled target batch using frozen text prototypes as geometric anchors via three closed-form geometric corrections compiled into a single static projection matrix by Affine Bias Regression. At inference, adaptation reduces to one matrix-vector multiplication in 0.0009 ms, making it substantially faster than gradient-based TTA while requiring no additional training. On UrbanSound8K, PRISM improves over the zero-shot baseline by 12.94 percentage points and surpasses an oracle-assisted TTA baseline by 9.41 percentage points, despite never observing its privileged augmented noise prompts. We further identify the Polyphonic Trap, a principled failure mode of subspace deflation for broadband classes, and resolve it via Confidence-Aware Regression (CAR), recovering up to 8.16 percentage points for the worst-affected class.

2. Zipf’s Law of Abbreviation in a Logographic Script: Coding-Theoretic Bounds on Chinese Character Stroke Counts

Authors: Mustafa Ergen

Categories: eess.SP, eess.AS | 13 pages, 4 figures, 3 tables. The analysis, figures and manuscript were produced end-to-end with Claude (Anthropic) in a single session; all numerical results were independently recomputed from the raw data Score: 7.42/10 (Obj:9 Id:9 Ind:8 Comp:7 Eff:7 Nov:8 Rep:3)

  • Strength: 中文笔画成本首次落入 Petrini et al. 报告的词长最优性 62–67% 带（Ω=0.668 主语料 / 0.609 复现），且 5 元 Huffman 绝对界 4.34 笔画与熵界 4.28 表明观测系统 1.66 倍于最优编码。
  • Weakness: 代码与派生数据集仅 “available on request” 而未见公开仓库，Ω 对库存深度敏感（截断至 1,000 字种时从 0.668 降至 0.377），全文数值无法独立复现。
  • Full review: Claude Code 全文七公理审稿

Zipf’s law of abbreviation – the tendency of frequent forms to be short – is one of the best-supported regularities in language, and recent work has moved from demonstrating it to measuring how far lexicons are compressed relative to principled baselines. That programme has so far addressed word lengths in alphabetic and syllabic scripts. We transfer it to a logographic script, taking the stroke as the unit of articulatory cost and the Chinese character as the coded form. Combining a stroke-order database covering all 20,902 characters of the CJK basic block with two independent frequency corpora (258.9M and 193.3M tokens), we find that the mean character type costs 12.71 strokes but the mean character token in running text only 7.22. Using the dually normalised optimality score of Petrini et al. (2026), the simplified inventory reaches Omega = 0.668, with the replication corpus at 0.609 – inside and just below the 62-67% band those authors report for word lengths across 20 languages and 8 scripts, suggesting a compression ceiling largely independent of script type and cost unit. A logographic script also makes absolute coding bounds computable, since strokes come from a closed five-element taxonomy: the exact 5-ary Huffman optimum is 4.34 strokes and the entropy bound 4.28, so the observed system is 1.66x above optimal coding. This gap is not slack but structure. The Kraft sum of 5^(-l_i) is 2.05 on the frequency list and 5.03 on the full inventory, so stroke strings are provably not uniquely decodable in one dimension; characters are disambiguated by the two-dimensional arrangement of strokes, not their sequence, and the forgone compression buys componential transparency. Finally, treating the mid-twentieth-century simplification reform as a controlled compression event, we find it raised optimality from 0.555 to 0.668, with savings concentrated in the 1,000 commonest characters.

3. Semantic Space of Parts of Speech

Authors: Jiří Milička, Ivan Kraus, Arnold Stanovský, Anna Vysloužilová, Barbora Štěpánková et al.

Categories: cs.CL Score: 5.68/10 (Obj:8 Id:4 Ind:5 Comp:6 Eff:4 Nov:6 Rep:9)

  • Strength: 论文把 5 种语言各 10,000 个高频词形的 100 维 word2vec 嵌入压缩到 3 维瓶颈空间呈现词类模糊边界，并在 OSF 完整开源源码与每语言交互式三维图谱（每份约 5MB，链接均可访问）。
  • Weakness: 全文无任何基线对比与量化指标，分类器准确率/F1 全部推迟到 OSF 支持材料，词类模糊性的三维结构未经独立金标准验证，效用宣称全靠定性描述。
  • Full review: Claude Code 全文七公理审稿

Parts of speech categorization is understood in the European linguistic tradition as crisp categorization, which is also reflected in corpus linguistics, where each disambiguated token is assigned exactly one POS. However, the assigned categories are largely determined by arbitrary decisions distilled into annotation manuals. Since some words stand between parts of speech in their semantics or typical syntax, and some parts of speech are closer to each other than others, POS categorization seems inherently fuzzy. We analyze this fuzziness using word2vec embeddings, training a neural network to reduce their high dimensionality to three dimensions relevant for determining parts of speech. This creates a three-dimensional space onto which we map several thousand words, revealing which are prototypical and which lie on the boundaries, and visualizing relationships between parts of speech. The study uses Universal Dependencies POS tags for French, Czech, Finnish, Russian, and English.

4. AudioTQ: A Data-Oblivious 6-Bit CPU Audio Codec via Randomized Hadamard Rotation and Lloyd-Max Quantization

Authors: Sahil Gangurde

Categories: cs.SD, cs.AI, cs.CR | 8 pages, 1 figure, 1 table. Code is available at https://github.com/lostmartian/audioTQ Score: 4.80/10 (Obj:8 Id:4 Ind:5 Comp:5 Eff:4 Nov:4 Rep:6)

  • Strength: 提出首个将 LLM 旋转量化迁移到音频时域的编码器，以 FWHT + 6-bit Lloyd-Max + 1-bit QJL 在纯 Python 单线程下实现 3.91x（74.4%，相对 32-bit 浮点）压缩与 30.24 dB SQNR，并给出稀疏输入下 SQNR 崩塌至 1.31 dB 的失效模式推导。
  • Weakness: 无任何实测 baseline（唯一对比为均匀量化解析公式，MP3/AAC/Opus/μ-law 均未评测），实际位率约 361 kbps 对 30 dB SQNR 的率-失真点不具竞争力，SQNR 低于文中自算的 7-bit 均匀量化（31.8 dB）与高斯 Lloyd-Max 理论极限，且论文未链接自己已公开的代码仓库。
  • Full review: Claude Code 全文七公理审稿

Lossy audio compression algorithms traditionally rely on psychoacoustic modeling and frequency-domain representations (e.g., MP3, AAC, and Opus) to discard information that is imperceptible to the human auditory system. While highly effective, these approaches are computationally complex and domain-specific. In this paper, we present the design and mathematical formulation of AudioTQ, a data-oblivious lossy audio codec that operates directly in the time domain. Inspired by Large Language Model (LLM) weight quantization techniques (specifically the TurboQuant framework), AudioTQ uniformizes volatile time-domain amplitudes into a predictable standard normal distribution using an orthonormal, randomized Fast Walsh-Hadamard Transform (FWHT) rotation. This enables coordinate-wise scalar quantization using an offline-trained, MSE-optimal 6-bit Lloyd-Max quantizer, augmented by a 1-bit Quantized Joint Least-Squares (QJL) residual correction layer. The resulting 7-bit virtual indices are packed into native 8-bit containers, aligning with standard CPU register boundaries to ensure real-time single-threaded execution without hardware parallel accelerators. We detail the bitwise reconstruction of 24-bit studio stems, analyze the butterfly network of the FWHT, derive the mathematical failure modes under sparse inputs, and present benchmarks showing up to 74.4% physical size reduction alongside a Signal-to-Quantization-Noise Ratio (SQNR) of ~30 dB.