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。

2026-07-22
日期2026-07-22
已评分—
均分—
最高—

Daily Papers — 2026-07-22

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

1. SimulS2ST-Omni: Data-Efficient Streaming Speech-to-Speech Translation via Explicit Trajectory Supervision

Authors: Rongshen He, Xinyu Liang, Dekun Chen, Jiaqi Li, Mingjie Chen et al.

Categories: cs.SD | 29 pages, 19 figures, 16 tables Score: 7.16/10 (Obj:9 Id:8 Ind:7 Comp:6 Eff:8 Nov:6 Rep:6)

  • Strength: 在匹配骨干网络的严格控制下，证明 Thinker–Talker 双流架构在低配对数据流式 S2ST 中显著优于统一解码器（+4.5 BLEU，远超 3-seed 噪声），且仅用 ~2k 小时配对数据即可与闭源 SOTA LiveInterpret 2.0 竞争，人工评测在 Zh→En 上显著优于后者（4.64 vs 4.01, p=0.003）。
  • Weakness: 核心架构（Thinker–Talker）直接采用自 Qwen2.5-Omni 非本文原创，联合文本-码轨迹监督虽为真实扩展但增量有限，且实验仅覆盖中英一种语言对，代码与模型 checkpoint 尚未发布。
  • Full review: Claude Code 全文七公理审稿

Long-form streaming speech-to-speech translation (S2ST) requires incremental, unbounded translation under strict latency constraints. Existing methods typically suffer from sentence-bounded supervision or demand massive paired-S2ST supervision. We introduce a training recipe enabling a speech language model for sentence-level and long-form streaming S2ST using only $\sim$2k hours of paired cross-lingual S2ST data, layered atop auxiliary supervision. Anchored by auxiliary multitask training, our approach remains robust even when the paired-S2ST budget itself is reduced by 90\%. Our core contribution, joint text-code trajectory supervision, schedules target text and acoustic semantic codes as a unified commitment path, eliminating the need for separate, unstable speech-side emission controllers. Furthermore, our two-stream Thinker–Talker factorization significantly outperforms unified-decoder baselines by decoupling linguistic reasoning from dense acoustic prediction to mitigate modality interference. Finally, our system achieves highly competitive quality-latency trade-offs on RealSI and ACL60/60-dev, matching state-of-the-art, closed-source S2ST systems such as LiveInterpret~2.0 on ASR-BLEU.

2. Sound Probabilistic Safety Bounds for Large Language Models

Authors: Mahdi Nazeri, Anne-Kathrin Schmuck, Sadegh Soudjani, Alessandro Abate

Categories: cs.CL, cs.AI | The Initial version of this manuscript has been available on OpenReview, see https://openreview.net/forum?id=papImkPLf5 Score: 6.80/10 (Obj:9 Id:5 Ind:9 Comp:6 Eff:5 Nov:8 Rep:5)

  • Strength: 首次为LLM有害输出概率提供形式化可靠下界，利用部分生成树扩展保证数学正确性，在7个实验中一致获得非平凡下界（10^{-34}至0.284），而Monte Carlo和Clopper-Pearson在同等预算下均为零。
  • Weakness: 未与最强同期经验方法（Angell et al.重要性采样、Zhao et al.序列蒙特卡洛）进行实验比较，且全部7个实验仅使用最简词法安全oracle，未验证更复杂分类器oracle下的有效性。
  • Full review: Claude Code 全文七公理审稿

We propose a novel framework for computing rigorous bounds on the probability that a large language model (LLM) generates harmful output to a given prompt. We study a new application of the Clopper-Pearson confidence intervals to obtain probably approximately correct (PAC) bounds for this problem. As our main technical contribution, we propose an algorithm that leverages features in the latent space to prioritize exploring branches in the auto-regressive generation tree that are more likely to produce harmful outputs. Our approach in particular enables the efficient computation of useful lower bounds, even in scenarios where the true harm probability is extremely small, and crucially, the obtained lower bounds are sound, i.e., formally proven to be less than the actual harmfulness probability: our experimental results demonstrate the effectiveness of our method by computing non-trivial lower bounds on state-of-the-art LLMs. This study newly enables the evaluation and statistical certification of LLMs.

3. Ultra-Compact CNN Architectures for Tropical Bird Audio Detection on Microcontrollers

Authors: Muhammad Mun’im Ahmad Zabidi, Mohd Yamani Idna Idris, Norisma Idris

Categories: cs.SD | 25 page, 6 figures Score: 6.74/10 (Obj:8 Id:8 Ind:6 Comp:8 Eff:8 Nov:5 Rep:8)

  • Strength: DrongoNet-Micro (919 params, 6.26 kB) 在 SEABAD 上实现了 0.9810 AUC 和 98.3% 召回率，在判别能力上匹配重训练的 TinyChirp 基准（ΔAUC -0.05 pp, p=0.59），同时参数减少 28 倍，并通过设备上测量证实了具体的部署增益（SD 卡寿命从 28→45 天，1.8% 占空比）。
  • Weakness: 架构新颖性有限 — DrongoNet 是对 TinyChirp CNN-Mel 的增量消融，使用了已建立的组件（GAP, focal loss, INT8），且部署经济学依赖于投影（而非测量）的 AudioMoth 功率模型；电池寿命和测试集阈值调优的独立性因数据集作者重叠和投影而非测量的设备外推而存在不足。
  • Full review: Claude Code 全文七公理审稿

Passive acoustic monitoring of tropical biodiversity is bottlenecked by the storage and battery cost of continuously recording soundscapes in which bird vocalisations typically occupy less than 10% of the audio. Autonomous recording units built on low-power microcontrollers (typically ARM Cortex-M with $\leq$256 kB of RAM) address this by triggering only on likely-positive segments, but the on-device options are unsatisfying: coarse frequency-energy triggers such as Goertzel filters flood SD cards with false positives at $\sim$71% precision, whereas neural detectors developed for temperate single-species tasks are either too large to deploy or transfer poorly to species-rich tropical settings. We present DrongoNet, a family of three INT8 CNN detectors sized for this envelope and validated on a 50,000-clip, 1,677-species Southeast Asian tropical dataset (SEABAD). The headline model, DrongoNet-Micro (919 parameters, 6.26 kB, 0.9810 AUC, 98.3\% mean recall at τ = 0.35), is a drop-in replacement for the Goertzel trigger used in commodity field recorders: at α = 0.10 tropical prevalence it captures 8 pp more bird vocalisations than Goertzel and extends a 32 GB card from $\sim$28 to $\sim$45 days of monitoring. DrongoNet-Nano (5.09 kB) bounds the ultra-low-flash extreme; DrongoNet-Edge (33.06 kB, 0.9991 AUC) targets Linux SBCs. On SEABAD, Micro matches a retrained TinyChirp CNN-Mel baseline within 0.1 pp AUC at 28$\times$ fewer parameters, confirming that the family is deployment-agnostic across mel-spectrogram bird corpora but requires per-environment retraining. Full INT8 quantisation costs $<$0.12% AUC across all three variants.

4. Audio-Zero: Label-Free Self-Evolution for Fine-Grained Audio Reasoning

Authors: Siqian Tong, Xuan Li, Chaozhuo Li, Baolong Bi, Yiwei Wang et al.

Categories: cs.SD, cs.AI Score: 6.53/10 (Obj:8 Id:6 Ind:6 Comp:5 Eff:9 Nov:5 Rep:6)

  • Strength: 仅用 2k 无标注音频对比对，通过自博弈游戏+GRPO 在两个 LALM 上全面超越 label-dependent 强 baseline（Qwen2.5-Omni 上 MMAU +8.0%、MMAR +10.0%、TREA +4.0%），且无 general audio 能力 trade-off。
  • Weakness: 核心机制与 Vision-Zero（ICLR 2026）高度同构，新颖性为跨域迁移+音频特定游戏设计；缺少最简 baseline（直接用 Audio-Alpaca 偏好标注做 DPO 的对比）、reward scorer 细节未完整公开、无代码发布声明。
  • Full review: Claude Code 全文七公理审稿

Large Audio Language models (LALMs) have made rapid progress on acoustic understanding, yet they still struggle with fine-grained audio reasoning (e.g., recognizing event order, repetitions and duration). Existing post-training methods heavily rely on expensive external labels or provide only coarse semantic signals. To bridge this gap, we introduce Audio-Zero, the first label-free self-evolution framework in the field of LALMs that improves fine-grained auditory perception and reasoning. Audio-Zero constructs an auditory self-play game from unlabeled audio contrast pairs: most players hear a reference audio, while one odd listener hears a subtle variant. The model first generates clues describing what it hears and then identifies the odd listener by reasoning over inconsistencies among clues. Since the odd listener is known by construction, the game provides verifiable rewards without any annotated answers. Experiments with Qwen2-Audio-7B-Instruct and Qwen2.5-Omni-7B on TREA, MMAU Test-mini and MMAR show that Audio-Zero improves fine-grained audio reasoning while preserving broad audio understanding. Evolutionary and diagnostic analyses further reveal that increasingly fine-grained auditory descriptions emerge naturally from game pressure.

5. Layer-Wise Decision Fusion for Fake Audio Detection Using XLS-R

Authors: Yixuan Xiao, Ngoc Thang Vu

Categories: cs.SD | Accepted to Interspeech 2025 Score: 6.10/10 (Obj:8 Id:6 Ind:8 Comp:6 Eff:6 Nov:5 Rep:8)

  • Strength: Layer-wise decision fusion 在 ITW 跨域评测上达到 EER 6.90%，较 feature fusion baseline (10.97%) 相对提升 37%，并通过层权重 heatmap 分析揭示了 FF 模型的 feature collapse 现象。
  • Weakness: SOTA 声称不够稳健——与 2B 参数模型的 7.20% 差距仅 0.30% (在 std 范围内)，且仅在 ASV+ITW 两个数据集上验证；核心 idea (per-layer decision fusion) 与已有 MoE per-layer fusion for FAD (arXiv:2409.11909) 高度相似，新颖性增量有限。
  • Full review: Claude Code 全文七公理审稿

Recent fake audio detection methods often leverage large speech models to achieve robust speech representations. These models are typically very deep, providing multiple layer-wise representations. However, current works often rely solely on single layer representation or feature fusion to extract one utterance-level representation for decision making. These methods risk underutilizing rich information from multiple layers and might induce feature collapse. We propose a novel layer-wise decision fusion method that applies fusion after per-layer decision making and achieves the best cross-dataset performance on In-the-Wild dataset (EER 6.90%) compared to other strong baselines. Our model design also makes the model more transparent, allowing us to conduct detailed analysis to reveal the underlying mechanism of decision making.

6. Multimodal Speaker Verification as a Threat to Speaker Anonymization

Authors: Ashi Garg, Cristina Aggazzotti, Leibny Paola García-Perera, Nicholas Andrews

Categories: eess.AS Score: 6.05/10 (Obj:8 Id:5 Ind:8 Comp:5 Eff:5 Nov:5 Rep:9)

  • Strength: 系统证明多话语聚合（帧级优于话语级）与多模态融合（音频+文本+韵律）可显著降低匿名化语音的 EER（A-A N=15 下从 37.59% 降至 22.63%，相对改进 39.8%），揭示现有孤立话语评测协议高估了隐私保护。
  • Weakness: 未引用且未比较 VoxATtack (2507.12081) 和 “You Are What You Say” (2506.09521) 两项直接相关工作，且仅在 Fisher 单一数据集和 Stream-Voice-Anon 单一匿名化系统上评测，泛化性证据不足。
  • Full review: Claude Code 全文七公理审稿

Most automatic speaker verification (ASV) systems operate on individual utterances, despite real-world interactions typically consisting of multiple utterances. As speech accumulates, increasingly rich speaker information becomes available through acoustic, prosodic, and linguistic cues, potentially challenging speaker anonymization methods that primarily target vocal characteristics. We investigate ASV in a multi-utterance, multimodal setting and examine whether aggregating information across anonymized speech impacts privacy. We first study audio-only aggregation across multiple anonymized utterances and observe consistent performance improvements as more speech becomes available. We then incorporate prosodic and linguistic information, showing that multimodal systems outperform unimodal approaches. Finally, we compare aggregation strategies and find that frame-level aggregation yields the lowest EERs. Even with only five anonymized utterances, combining audio and text reduces EER by over 15% relative to audio-only aggregation, demonstrating that substantial speaker-discriminative information remains accessible despite anonymization.

7. Efficient Chain-of-Modality Reasoning via Progressive Compression for Spoken Language Models

Authors: Pengchao Feng, Chao-Hong Tan, Qian Chen, Wen Wang, Xiangang Li et al.

Categories: cs.CL, cs.SD Score: 5.89/10 (Obj:8 Id:6 Ind:6 Comp:5 Eff:6 Nov:5 Rep:6)

  • Strength: ECoM Reasoning 在 CoM 架构上实现 text token 3.2x 压缩（96.56→30.21 tokens）和首语音延迟 3.1x 加速（4.90s→1.60s），同时平均准确率小幅提升（57.74%→60.66%），三阶段 progressive training 优于直接训练（Table 5）。
  • Weakness: 摘要声称 “40% of text tokens” 与实际数据（31.3%）不符；+3% 准确率提升无显著性检验且在 SVAMP 上下降；仅验证 1.5B 模型和 4 个小学级数学基准，泛化性未验证。
  • Full review: Claude Code 全文七公理审稿

Spoken language models (SLMs) enable natural human-computer interaction, but their reasoning ability still lags behind that of text-based large language models, especially on spoken mathematical question answering tasks. One important reason is that SLMs reason over purely verbalized mathematical expressions, which are harder to interpret than symbolic text. However, directly transferring text-based reasoning to SLMs is nontrivial due to architectural constraints and the additional computational requirements. To address this challenge, we propose Efficient Chain-of-Modality Reasoning (ECoM Reasoning), the first framework to introduce compressed reasoning into SLMs. By compressing the textual component so that it jointly serves as speech guidance and reasoning representation, ECoM Reasoning improves reasoning accuracy while using a smaller token budget than the standard Chain-of-Modality (CoM) architecture, which generates intermediate text before speech. To train this capability, we further propose Progressive Compression, a curriculum-based strategy that gradually trains the model from full-form reasoning to compressed reasoning. Experiments on spoken mathematical question answering benchmarks show that ECoM Reasoning improves accuracy by 21% over standard CoM without explicit reasoning, and by 3% over CoM with full reasoning traces while using only 40% of the text tokens, demonstrating that it enhances SLM reasoning while remaining inference-efficient.

8. RPPNet: Perceptually-Grouped Rhythm-Pitch Primitives for Long-Term Structure Melody Generation via Boundary-Aware Modeling

Authors: Tieyao Zhang, Yuke Liu, Jiaxing Yu, Xinda Wu, Kejun Zhang et al.

Categories: cs.SD, cs.AI Score: 5.74/10 (Obj:6 Id:5 Ind:5 Comp:5 Eff:5 Nov:8 Rep:8)

  • Strength: 首次将音乐心理学感知分组规则（GPR/听觉惯性/相似性）操作化为变长 RPP 结构单元用于旋律生成，消融实验（Random-Grouped, PPL 2.21 vs 2.36, p<0.001 主观全维度）证明感知语义而非变长本身带来提升。
  • Weakness: 主结论依赖 15 人小样本主观评测，客观指标无法跨 baseline 比较，baseline 偏旧（Museformer/MELONS 均为 2022），验证范围限于 32-bar 单声部旋律，与标题”Long-Term Structure”的广度不匹配。
  • Full review: Claude Code 全文七公理审稿

Existing symbolic music generation models typically use bars as the basic structural unit. However, human perception of musical phrases often does not align with notated bar lines, leading to long-term structural fragmentation. This paper proposes RPPNet-a two-stage deep learning architecture with variable structural boundaries. It first generates variable-length Rhythm-Pitch Primitive (RPP) sequences, where each RPP encodes note count, rhythm, and contour; then decodes the RPP sequences into concrete notes. The grouping of RPPs is automatically derived from acoustic cues, auditory inertia, and similarity perception based on music psychology. Experiments show that melodies generated by RPPNet are superior in both long-term structure and musicality, with significant improvements across all subjective evaluation dimensions. Ablation studies confirm that the performance gain stems from the structural correctness of the psychological representation, rather than from model capacity. This work offers an interdisciplinary perspective for music generation, integrating music theory, computational modeling, and music psychology.

9. Scalable Keyword Spotting via Modular Network Expansion

Authors: Viktor Khaymonenko, Dzmitry Saladukha, Aliaksei Rak, Alexander Rostov

Categories: cs.SD | Accepted to Interspeech 2026 Score: 5.70/10 (Obj:7 Id:6 Ind:8 Comp:5 Eff:6 Nov:5 Rep:6)

  • Strength: 通过冻结整个 base 网络（含 BN 统计量）并附加 ≤10k 参数的扩展分支，在 GSC v2 上将新关键词 FRR 从 6.46% 降至 4.37%，同时由构造保证旧关键词零回归，且 MACs 低于 Adapters/LoRA baseline（16.34M vs 18.45M/20.52M）。
  • Weakness: 仅在 GSC v2 的 5 个词对上验证，未引用也未比较两篇直接同任务竞争工作（AnalyticKWS [ACL 2025] 和 Kim et al. [Interspeech 2024]），代码未开源，方法本质是 PNN 在 KWS 的特化应用，标题声称 “Scalable” 但多阶段扩展留给 future work。
  • Full review: Claude Code 全文七公理审稿

Keyword spotting (KWS) models on embedded devices often need to add new keywords after deployment, but updates are difficult when original training data are unavailable and regressions on existing triggers are unacceptable. At a fixed operating point, our method reduces average new-keyword false reject rate (FRR) from 6.46 to 4.37 versus a parameter-matched separate-model baseline and outperforms parameter-efficient tuning baselines (adapters, LoRA), while using fewer multiply-accumulate operations (MACs) under the same added-parameter budget ($\leq$10k): 16.34M vs 18.45M/20.52M. We achieve this via parameter-capped modular expansion: the base network, including batch-normalization statistics and the core classifier, is frozen, and only a lightweight expansion branch with a separate new-keyword head is trained, preserving core logits, shipped outputs, and thresholds for existing keywords.

10. Learning the Arabic Dialect Continuum as a Continuous Space: A Regression Approach to Speaker Origin Prediction

Authors: Mohamed Aziz Khadraoui, Adel Ammar, Bilel Benjdira, Zahid Khan, Skander Turki et al.

Categories: cs.CL, cs.AI, cs.CY, cs.LG, cs.NE | Under review Score: 5.60/10 (Obj:8 Id:5 Ind:8 Comp:5 Eff:5 Nov:6 Rep:6)

  • Strength: 首次将阿拉伯语方言识别从离散分类重构为连续球面坐标回归，提出 leakage-free GroupKFold + city-masking zero-shot 评测协议，在 2,329 样本上达到 481.2 km 中位定位误差，Mantel test (r=0.131, p≈10⁻⁴) 为方言连续体假说提供了定量支持。
  • Weakness: 缺少任何 baseline 对比（包括最简质心 baseline 和离散分类 baseline）和组件级 ablation，无法判断连续回归是否真正优于离散分类+质心映射，也无法隔离各模块贡献。
  • Full review: Claude Code 全文七公理审稿

We present a regression-based approach to Arabic dialect geolocation that models dialectal variation as a continuous geographic space rather than discrete categories. Speaker origin is predicted as continuous latitude-longitude coordinates using a hierarchical neural architecture that fuses frame-level XLS-R-300M and Whisper-large-v3 encoder representations with phonotactic descriptors through a Transformer encoder and a learnable attention-pooled query. A spherical geodesic loss directly optimizes great-circle distance on Earth’s surface, avoiding distortions inherent to planar coordinate regression. Under a leakage-free 5-fold GroupKFold protocol grouped by source recording, our model attains a pooled median localization error of 481.2 km. Auxiliary country and city heads reach 64.5% and 45.2% accuracy, respectively. A permutation Mantel test on the learned latent space provides quantitative support for the Arabic dialect continuum hypothesis. To probe true generalization, we further introduce a city-masking protocol in which two cities per fold are removed from training but retained in validation. Under this zero-shot regime, the mean error rises to 1173.3 km, a 1.32x degradation relative to seen cities. Our findings establish continuous geographic modeling as a principled framework for Arabic dialect geolocation and quantify both its strengths and the substantial headroom that remains.

11. Improved Monitoring of Honey bee Colony Strength via Audio IoT Sensors, Modulation Tensorgrams and Recurrent Neural Networks

Authors: Mahsa Abdollahi, Yi Zhu, Heitor R. Guimarães, Nico Coallier, Ségolène Maucourt et al.

Categories: eess.AS Score: 5.50/10 (Obj:8 Id:5 Ind:5 Comp:4 Eff:6 Nov:4 Rep:8)

  • Strength: 在 3000+ 小时 UrBAN 数据集上验证 modulation spectrogram/tensorgram + CRDNN 在蜂群强度回归中显著优于 spectrogram/MFCC（random-split MAE 1.01 vs 3.5, r 0.97 vs 0.76），且模型轻量（4.5MB, 4.71ms）适合 IoT 边缘部署。
  • Weakness: modulation tensorgram 概念已在同一课题组 Zhu & Falk (CSL 2024) 中定义用于 COVID-19 检测但未被引用，”new representation” claim 不成立；推荐的 CRDNN-3D 在严格 hive-independent 设置下相比简单 modulation spectrogram + CNN-2D 增量不显著（MAE 3.31±1.36 vs 3.41±0.87，标准差重叠。
  • Full review: Claude Code 全文七公理审稿

Honey bees (Apis mellifera) play a crucial role in agriculture and ecosystem stability as key pollinators of crops and wild plants. As such, monitoring hive strength remotely with Internet of Things (IoT) sensors has become a crucial task. Previously, handcrafted features extracted from the modulation spectrum of audio IoT devices were shown to improve acoustic monitoring of colony strength. In this paper, we hypothesize that important discriminative information is present in the temporal dynamics of the modulation spectrum, but this information is discarded with prior methods. As such, we explore the use of a new modulation tensorgram where the time dimension is kept. This new representation is used as input to a convolutional neural network (CNN) and a convolutional recurrent deep neural networks (CRDNN). Using the public UrBAN dataset, which contains more than 3,000 hours of beehive audio recordings, we show that the proposed method improves both accuracy and cross-hive generalizability over prior benchmark methods, and the results further suggest improved robustness to noisy in-the-wild recording conditions. We use saliency maps and gradient-weighted class activation maps for explainability and show the importance of the modulation spectral temporal dynamics for the task at hand. Overall, our results suggest that accurate, generalizable, and robust acoustic monitoring of honey bee colony strength is possible.

12. StellarTTS: Sparse Temporal Embedding for Low-Latency and Robust Speech Synthesis

Authors: Kaicheng Luo, Xuefei Gong, Yutao Sun, Jinling He, Yujie Hou et al.

Categories: cs.SD | Accepted by ASRU 2025 Score: 5.16/10 (Obj:8 Id:5 Ind:6 Comp:5 Eff:5 Nov:5 Rep:5)

  • Strength: 在 Seed-TTS test-hard 上 WER=7.47 优于所有 SOTA baseline（MaskGCT 10.27, F5-TTS 8.67），RTF=0.08 实现 4-9× 加速，sparse temporal embedding 的 central vs random ablation 有明确机制支持（WER 1.44 vs 2.34）。
  • Weakness: SIM-o 在 test-zh（0.712）低于四个 baseline 中三个，且未对 semantic distillation 做隔离 ablation；RTF 仅在 A100 测得未给移动端实测；代码/checkpoint 未承诺开源；三大组件均为已有方法改组合，组件级新颖性有限。
  • Full review: Claude Code 全文七公理审稿

The trade-off between robustness, latency, and prosody critically challenges text-to-speech (TTS) systems. Autoregressive models, despite fidelity, are slow and error-prone; non-autoregressive (NAR) alternatives, while fast, often sacrifice prosodic naturalness via rigid alignments. This paper introduces StellarTTS, a novel mobile-optimized NAR TTS framework based on a sparse temporal embedding strategy, enabling granular control of phoneme duration, pronunciation, and prosody. Furthermore, we propose a semantic-aware codec that facilitates efficient single-stage decoding. Conditioned on the sparse temporal embedding, our 83M-parameter lightweight masked generative transformer achieves a real-time factor (RTF) of 0.08. Experiments demonstrate that StellarTTS attains lower latency and stronger robustness compared to state-of-the-art TTS systems, while maintaining competitive performance in audio quality, prosodic naturalness, and speaker similarity.

13. Cumsum-Composable Phase Transport for Low-Cost Streaming Keyword Spotting

Authors: Mahesh Godavarti

Categories: cs.SD, cs.LG Score: 5.05/10 (Obj:5 Id:6 Ind:8 Comp:6 Eff:3 Nov:5 Rep:4)

  • Strength: 匹配的 cumsum 与扫描对比实验（Table 4）干净地隔离了时间基元，表明在单次种子运行下，cumsum+window (94.82%) 达到了与学习衰减扫描 (94.33%) 相当的准确率，同时在 Tesla T4 上训练速度提升 1.07 倍，延迟从 7.09ms 降至 5.01ms。
  • Weakness: 所有结果均为单次种子运行，没有代码发布，且主要效用比较仅针对一个非标准本地 CNN 基线（MelCNNMaxPool, 97.1%），而已发表的标准基线 BC-ResNet-2 (97.6%, 30K 参数) 和 DS-TC-ResNet (98.0%, 75K 参数) 均超过了论文的最佳准确率（97.3%）。
  • Full review: Claude Code 全文七公理审稿

State-space sequence models are attractive for streaming speech because they maintain compact recurrent state, but scan-style training kernels can have unfavorable constants for short audio tasks. We study cumsum-composable phase transport, a streaming-native temporal layer for keyword spotting. Each layer projects acoustic frames to complex channels, transports them by learned unitary rotations, accumulates a finite window using prefix differences, and applies a gated residual update. The same prefix representation gives exact batched training with ordinary cumulative sums and exact online inference with one prefix update per frame. Unitary transport is the key constraint: inverse rotations have norm one, keeping prefix terms well conditioned while memory is supplied by windows or block readouts. On Google Speech Commands v2 with 12 labels, mel+cumsum models retain competitive accuracy with compact baselines. The strongest single-seed run reaches 97.3\% test accuracy; a 51.6K-parameter tied model also reaches 97.3\%, and a 24.8K tied model reaches 96.8\% versus 97.1\% for a 25.6K MelCNNMaxPool baseline. In a matched cumsum-versus-scan benchmark, cumsum+window gives comparable accuracy, 94.82\% versus 94.33\%, while training 1.07x faster and reducing single-example latency from 7.09 ms to 5.01 ms on a Tesla T4. These results support cumsum phase transport as a simple low-cost temporal primitive for streaming keyword spotting.

14. Nonlinear Bias-Compensated Adaptive Filter and Its Application for Time-Series Prediction

Authors: Yi Peng, Haiquan Zhao, Jinhui Hu

Categories: cs.LG, eess.AS Score: 4.95/10 (Obj:6 Id:5 Ind:8 Comp:4 Eff:5 Nov:4 Rep:5)

  • Strength: 将 RFF 映射、bias-compensated 项和 GA 损失函数组合为 RFFBCGA 算法，在非线性 EIV 模型下同时实现固定网络结构、输入噪声补偿和非高斯鲁棒性，在 sunspot 和 Chua’s circuit 两个数据集上收敛速度和稳态 MSE 均优于 BCKLMS。
  • Weakness: 方法本质是三个已有组件（RFF [2007/2012]、BC [2025]、GA [2019/2025]）的拼装，缺少关键消融 baseline（RFFKLMS、RFF+BC、RFF+GA），未隔离各组件贡献，且 BCKLMS 的字典选择方式可能不公平，无代码开源。
  • Full review: Claude Code 全文七公理审稿

Most existing nonlinear adaptive filtering algorithms only account for output noise, neglecting the fact that input noise is also prevalent in practice. Although the recently proposed bias-compensated kernel least mean square (BCKLMS) algorithm addresses input noise in the nonlinear errors-in-variables (EIV) model, it still suffers from two major limitations. First, the use of a fixed-size dictionary restricts network growth but also prevents it from fully capturing the characteristics of the input signal. Second, as an least mean square (LMS) based algorithm, it exhibits poor robustness in the presence of non-Gaussian noise in the output signal. To overcome these issues, this paper proposes the random Fourier bias-compensated filter under general adaptive function (RFFBCGA) algorithm. Within the random Fourier feature based bias-compensated (RFFBC) framework, the proposed algorithm not only maintains a fixed network structure and effectively mitigates input noise interference through the BC term, but also achieves improved characterization of the input signal. Moreover, by leveraging the flexible form of the general adaptive (GA) function, the algorithm’s robustness across various noise scenarios is further enhanced. Extensive simulations, including real-world time series prediction tasks, demonstrate the superiority of the proposed method.

15. A Diagnostic Evaluation Framework for AI-Generated Cover Songs Using Music-Theoretic and Acoustic Features

Authors: Yingxin Liang

Categories: cs.SD | 16 pages, 3 figures, 8 tables. Code and analysis materials: https://github.com/TiaaL/songecho-cover-metrics Score: 4.95/10 (Obj:6 Id:5 Ind:5 Comp:8 Eff:3 Nov:5 Rep:5)

  • Strength: D2/D3 分离的实证观察（6 个 D3≥4 但 D2≤2 的翻唱案例）提供了”调性稳定 ≠ 和声正确”的可靠证据，且对 9 个符号/声学特征的负面结果（Bonferroni 校正后全部不显著，规则试点 16 个比较全部 95% CI 跨零）诚实排除了低级特征作为自动评分替代的路径。
  • Weakness: benchmark 仅 30 样本/5 源曲/单标注者，音频因版权不可释放，未与任何学习型 baseline（MuQ-Eval, FAD）比较，所有自动化结论均未达统计显著，无法作为社区可用的证据基础设施。
  • Full review: Claude Code 全文七公理审稿

AI-generated covers often fail through local musical errors that a global quality score cannot locate: the vocal contour may remain recognizable while the accompaniment uses the wrong harmonic function, or the output may stay in key while the arrangement remains incomplete. We present a five-dimensional diagnostic framework covering melodic pitch, harmonic progression, key consistency, style consistency, and arrangement/production quality. The benchmark contains 30 covers generated from 5 source songs by 6 systems, with expert severity ratings and 9 symbolic or acoustic features. Harmonic progression and arrangement had the highest severe-error rates (53% and 47%), whereas key consistency was better preserved. Six covers combined acceptable key consistency with severe harmonic errors. Large-leap ratio had a nominal association with melodic ratings (Spearman rho = -0.429, uncorrected p = 0.018), but no feature correlation survived the nine-test multiplicity reference. An interpretable percentile-rule pilot likewise failed to outperform a fixed majority baseline reliably across 16 dimension-level comparisons. The results separate useful diagnostic evidence from dependable automatic scoring: low-level and symbolic summaries can expose particular symptoms, but they do not replace context-aware musical judgment.

16. Black-Box Optimization for Identifying and Inverting Audio Dynamic Range Control Effects

Authors: Haoran Sun, Dominique Fourer, Hichem Maaref

Categories: cs.SD Score: 4.53/10 (Obj:6 Id:5 Ind:5 Comp:5 Eff:5 Nov:3 Rep:6)

  • Strength: 提出 derivative-free optimization 在 perceptual feature space 中做 DRC blind inversion，在 feature-space distance 上显著优于 MEE baseline（compression 0.29 vs 0.90）和 CleanUMamba（2.31），验证了 dynamic histogram features 对 DRC inversion 的有。
  • Weakness: 方法本质是已有组件拼装（inverse model [11] + features [20] + 标准 optimizer + MEE prior [4]），缺少与最直接竞争方法 DeepAFx/SPSA 的实验比较，且 feature-space 优化目标与评测指标重叠构成循环论证；参数 MSE 在 compression 上反而劣于 MEE baseline（0.36 vs 0.05），说明”优化有效”仅在 proxy space。
  • Full review: Claude Code 全文七公理审稿

Dynamic Range Compression (DRC) is a widely used nonlinear audio effect whose parameters are often unknown, making blind estimation and inversion challenging. In this work, we formulate DRC parameter estimation as a black-box optimization problem in a perceptually motivated feature space. Given an observed signal and a reference representation, we estimate the parameters that minimize the distance between feature descriptors of the reconstructed and reference signals. Unlike gradient-based approaches, the proposed method does not require differentiability of the DRC model or the feature extraction pipeline, enabling the use of nonlinear and histogram-based descriptors. Experimental results demonstrate that the proposed method achieves competitive performance in blind parameter estimation and dry signal recovery, outperforming or matching state-of-the-art models in terms of reconstruction quality.

17. Pushing the Frontier of Full-Song Generation: Hierarchical Autoregressive Planning Meets Flow-Matching Rendering

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

Categories: cs.SD, cs.AI, eess.AS Score: 4.16/10 (Obj:6 Id:5 Ind:3 Comp:5 Eff:5 Nov:4 Rep:2)

  • Strength: 完整端到端全曲生成系统，在 500-example 多语言 benchmark 上 18 维度中 15 项最高分，Artificial Analysis leaderboard Elo 1129 排名 2-3，FullDiT 消融设计合理（1.5B matched models + blinded expert test）。
  • Weakness: SongBench 同时用于训练 reward 筛选和最终评测（循环论证）；hybird-LM 零消融；instrumental 和 cover song 两任务无任何定量评测；无代码/数据/checkpoint 开源；各组件来自已有工作，系统级 novelty 有限。
  • Full review: Claude Code 全文七公理审稿

In this report, we present a unified song generation framework capable of producing high-quality full-length music from lyrics, text descriptions, and musical attributes. The proposed framework supports three tasks: Lyrics-to-Song Generation, which generates complete songs from text descriptions, lyrics, and musical attributes; Instrumental Music Generation, which creates music without vocals; and Cover Song Generation, which reinterprets existing songs with different styles while preserving their melodic content. Architecturally, our system consists of four main components: a semantic-aware tokenizer, hybird-LM, FullDiT, and a two-level melody module. The tokenizer encodes audio into 8-codebook RVQ tokens for efficient discrete music representation. Based on these tokens, hybird-LM performs hierarchical autoregressive audio-token modeling for full-song generation. To improve audio fidelity, FullDiT performs full-song flow matching in a continuous VAE latent space conditioned on codec tokens, lyrics, and text captions. For cover song generation, the melody module extracts and discretizes melody cues from reference audio to guide generation while preserving the original melodic content. Finally, we investigate DPO, GRPO, and OPD as reward-based post-training strategies for hybird-LM and apply flow-based GRPO to FullDiT to improve musicality and rendering quality. Experimental results on a multilingual automatic benchmark, complemented by the Artificial Analysis Music with Vocals leaderboard, show that the proposed framework achieves competitive performance in the evaluated settings.