Daily Papers — 2026-08-01

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

1. Experience-Calibrated Contrastive Decoding for Mitigating Hallucinations in LM-Based Text-to-Speech

Authors: Chenlin Liu, Minghui Fang, Zhonghao Bi, Zekai Su, Rong Wang et al.

Categories: eess.AS, cs.LG, eess.SP | Work in progress Score: 7.40/10 (Obj:9 Id:8 Ind:6 Comp:8 Eff:9 Nov:6 Rep:5)

Strength: ECCD 是一种 training-free 的 decoding-time 方法，在 4 个 LM-based TTS 模型上将 WER/CER 降低高达 55.6%（所有 SeedTTS-Eval 设置和 25 个多语言设置中的 24 个），CMOS 提升 +0.644，且组件消融清晰证明了每个设计选择的必要性。
Weakness: 该方法是标准对比解码组件（专家锚定、正向修正、自适应校准）的组合，这些组件在文本/视觉领域已有先例，且缺少代码发布、“Work in progress”标记，其因果分析依赖于近似的 ASR 边界标注。
Full review: Claude Code 全文七公理审稿

Language model-based text-to-speech (LM-based TTS) remains vulnerable to speech hallucinations that deviate from the target text. Existing mitigation mainly relies on architectural changes or additional training, while decoding-time control remains underexplored. We present a conditional information view that distinguishes text-derived alignment information from experience information supplied by acoustic context and learned speech regularities. We hypothesize that an important class of hallucinations begins when alignment support is insufficiently reflected in the selected token at a vulnerable transition. Using predictions from the same speech LM with and without text conditions, we propose Experience-Calibrated Contrastive Decoding (ECCD), a training-free method that strengthens alignment support while preserving useful experience information. ECCD preserves the original expert distribution, applies only positive alignment enhancement, and calibrates its strength using set-level experience compatibility. Across four models, ECCD reduces WER/CER by up to 55.6% in all SeedTTS-Eval settings and 24 of 25 multilingual CV3-Eval settings. A listening test yields a CMOS gain of $+0.644$ while retaining strong speaker similarity. Further analysis shows that alignment influence and decision-level gain vary within linguistic units and are lower at first-error boundaries than at matched correct boundaries. Overall, these extensive experiments and analyses identify conditional information control as a promising decoding-time direction for mitigating speech hallucination.

Authors: Ruidong Zhang, Jiacheng Liu, François Guimbretière, Cheng Zhang

Categories: cs.SD, cs.HC, cs.LG Score: 6.80/10 (Obj:9 Id:5 Ind:5 Comp:8 Eff:6 Nov:9 Rep:9)

Strength: 首个开放词汇三模态可穿戴 SSI 数据集（34h/18k utterances, 5,356 unique words, 39/39 phoneme coverage），填补了非侵入式 SSI 开放词汇数据空白，baseline 26.3% WER 证明任务可解且 scaling 未饱和。
Weakness: 单说话人自采集自评测限制了 WER 数值的外部可信度，三模态中 video/audio 模态的训练价值无实验验证，baseline 缺乏架构消融和最简 baseline 对比。
Full review: Claude Code 全文七公理审稿

Wearable silent speech interfaces (SSIs) are limited to small, closed vocabularies. Approaches achieving larger vocabularies require obtrusive hardware such as facial electrodes. We present SoniSpeech, the first large-scale, open-vocabulary, trimodal dataset for wearable SSI using acoustic-sensing eyewear. It contains 34 hours across 18,000 utterances with three synchronized modalities: ultrasound echo profiles, voiced audio, and frontal video, in both voiced and silent modes. The corpus draws from the SODA dialogue dataset, providing contemporary conversational English with 5,356 unique words and full phoneme coverage. A CTC-based ResNet-34 baseline achieves 26.3% word error rate (WER) on open-vocabulary silent speech recognition, the first benchmark for this task. Dataset is available at https://doi.org/10.7298/xjjr-9m85

3. AnyBand: Unified Multi-Bandwidth Speech Extension via Frequency-Aware In-Context Spectral Infilling

Authors: Junchuan Zhao, Minh Duc Vu, Bowen Zhang, Ye Wang

Categories: cs.SD | 16 pages, 11 figures, 7 tables Score: 6.30/10 (Obj:8 Id:6 Ind:8 Comp:5 Eff:7 Nov:5 Rep:6)

Strength: 将多带宽 BWE 统一为上下文频谱填充是一个真实的公式创新，在标准和不规则截止频率下，在 VCTK 和 EARS 上始终实现最低的 LSD/HF-LSD，并在严重带宽受限的输入中表现最强。
Weakness: 该方法堆叠了 5 个以上大多数为已有的组件（F5-TTS 主干、流匹配、对抗性判别器、F0 条件、课程学习、CFG），但消融实验显示个别判别器的边际贡献不一致，且缺少与并发统一 BWE 工作 (UniverSR, CodecFlow) 的比较，其中 CodecFlow 与 AnyBand 共享作者。
Full review: Claude Code 全文七公理审稿

Bandwidth extension (BWE) aims to recover missing high-frequency content from band-limited speech. Existing methods often formulate BWE as a fixed or predefined bandwidth conversion problem, potentially requiring cutoff-specific models or retraining when the input bandwidth changes. This assumption limits their applicability to practical scenarios where speech may arrive with diverse cutoff frequencies. We propose AnyBand, a unified BWE framework that recasts bandwidth extension as in-context spectral infilling. Motivated by prompt-based zero-shot speech generation, AnyBand conditions high-frequency generation on the observed low-frequency spectrum, using the available band as a frequency-domain prompt that conveys content, speaker, prosodic, and spectral-envelope cues. This formulation enables a single model to perform cutoff-conditioned generation over a continuous range of input bandwidths. AnyBand is trained with missing-band conditional flow matching and an Easy-to-Balanced cutoff curriculum over continuously sampled cutoff frequencies. To better exploit the spectral prompt, we introduce a frequency-aware Diffusion Transformer that models cross-frequency interactions and long-range temporal dependencies, followed by a physically motivated multi-view adversarial refinement stage to enhance spectral realism, envelope coherence, and harmonic consistency. Experiments on multiple datasets and bandwidth settings show that AnyBand consistently improves spectral reconstruction over existing baselines while achieving competitive perceptual quality across both standard and irregular input cutoffs. Audio samples are available.

4. Anomalous Sound Detection Meets Noise-Aware Self-Supervised Learning

Authors: Takuya Fujimura, Gordon Wichern, Yoshiki Masuyama, Christoph Boeddeker, Kohei Saijo et al.

Categories: eess.AS Score: 6.05/10 (Obj:8 Id:5 Ind:6 Comp:5 Eff:8 Nov:4 Rep:6)

Strength: NA-SSL 框架在 DCASE 2026 Challenge Task 2 中以 70.24% official score 获得第一名（51 个团队、168 份提交），超越第二名 4.78 分，且在三个 SSL backbone（BEATs/EAT/Dasheng）上一致有效。
Weakness: 核心方法（NA-SSL 框架、NA 层设计、MSE distillation）完全来自同一团队此前发表的 NABEATs [15]（IWAENC 2026），本文仅做了 backbone 扩展和场景适配，新颖性有限；NA-SSL 提升高度依赖 valveEmu 机器类型，部分机器类型反而下降，泛化性未讨论；缺少简单双通道降噪 baseline 对比和代码开源链接。
Full review: Claude Code 全文七公理审稿

In this paper, we introduce noise-aware self-supervised learning (NA-SSL) models for noise-aware anomalous sound detection (NA-ASD). NA-ASD is an ASD task with two-channel audio recordings, where one microphone is located close to the target machine and the other is located farther away to capture noise. For this task, we simulate two-channel recordings using diverse audio datasets and train NA-SSL models to extract clean SSL representations of the close-microphone signal by using the far-microphone recording dominated by background noise as auxiliary information. The NA-SSL models are then used as frontends in the standard ASD framework. Our experimental evaluation on the DCASE 2026 Challenge Task 2 development dataset demonstrates the effectiveness of the NA-SSL framework across three base SSL models (BEATs, EAT, and Dasheng), both with and without discriminative fine-tuning. Furthermore, the challenge results proved the effectiveness of the proposed approach, where the NA-BEATs system won the challenge by a large margin, achieving an official score of 70.24%, while the second-place system achieved 65.46%.

5. DRONEAUDIONET: Noise Suppression for Drone Audition-based Search and Rescue

Authors: Chitralekha Gupta, Soundarya Ramesh, Yifei Luo, Suranga Nanayakkara

Categories: cs.SD, eess.AS | *first two authors are equal contributors Score: 5.95/10 (Obj:8 Id:6 Ind:8 Comp:8 Eff:5 Nov:5 Rep:8)

Strength: 将 AudioSep 重构为 drone-noise estimator，引入 learnable mask-scaling α 使 mask 突破 unit-circle，在 HV 类 -20~-10 dB 取得 F1 +10.6% 相对提升（p<0.05），并在 OOD DREGON 上保持 HV F1 0.69 vs 0.63。
Weakness: 未对比 2025 年已发表的 drone-specific 学习方法（rotor-conditioned models、semi-blind BSS、Spatial-U-Net），且在 -30~-20 dB 绝对 F1 仅 0.17，HNV/NH 类多处不及 AudioSep-FT，SI-SDR 改善微小，真实可用性证据不足。
Full review: Claude Code 全文七公理审稿

Microphones mounted on UAVs enable aerial acoustic scene analysis applications such as search-and-rescue, wildlife monitoring, and industrial inspection. However, drone rotor noise often dominates the mixture signal at SNRs well below -10 dB, making source recovery extremely challenging. Existing enhancement and source separation methods are typically designed for near-balanced mixtures and degrade substantially in drone audition settings. In this work, we propose DRONEAUDIONET, a drone noise suppression method that reframes a source separation model as a drone noise estimator. To better model drone-dominant mixtures, we introduce a learnable mask-scaling mechanism that allows mask magnitudes beyond unity, together with an additive residual correction term for improved drone estimation and source recovery. We train and evaluate our model on a publicly available drone audition dataset and test generalizability on an out-of-domain dataset with unseen drone hardware and flight modes. Results show that DRONEAUDIONET consistently improves downstream sound classification performance, with the largest gains observed for human vocal sounds. Our findings demonstrate the importance of drone-specific modeling for robust aerial acoustic perception and highlight the potential of source separation methods for real-world drone-assisted search-and-rescue.

6. Hidden-Domain Routing for All-Type Audio Deepfake Detection

Authors: Yifan Gao, Yao Tian, Hongbin Suo, Haonan Lu

Categories: cs.SD | Accepted by ACMMM 2026 Score: 5.79/10 (Obj:8 Id:5 Ind:8 Comp:5 Eff:8 Nov:4 Rep:5)

Strength: 系统在官方 AT-ADD Track2 blind 评测中取得 96.10% Macro-F1 排名第一，比最强官方 baseline FT-XLSR-AASIST 提升 16.63 个百分点，四类音频全面取胜且组件互补性有数据支撑。
Weakness: 方法本质是”先分类再检测”的 cascade 方案加现成模型拼装，缺少路由错误传播分析、融合规则消融和最简系统级 baseline 对比，且无代码开源声明。
Full review: Claude Code 全文七公理审稿

All-type audio deepfake detection requires authenticity decisions across speech, environmental sound, singing voice, and music, while the audio type is unavailable at inference time. In AT-ADD Track2, this setting creates a hidden audio-domain condition: the binary real/fake label is shared across domains, but representation structure and detector-score behavior vary with audio type. We present a closed-condition routed system that first recovers the hidden audio domain and then interprets detector scores within the selected branch. The AudioType-BEATs-6s Router estimates audio type from a 6-second window; speech inputs are handled by the Speech-XLSR Expert, while sound, singing, and music rely on EAT-based general-audio experts with branch-local score interpretation. Development-set representation analysis, router-family comparisons, and component results show audio-domain separation and complementary detector strengths across audio types. On the official AT-ADD Track2 final evaluation, the system achieves 96.10% Track2 Macro-F1 and ranks first on the final leaderboard, with type-wise Macro-F1 scores of 88.07%, 98.18%, 99.07%, and 99.08% for speech, sound, singing, and music, respectively. These results support recovering the hidden audio domain before interpreting detector scores in all-type audio deepfake detection.

7. Beyond Prompt Adherence: Auditing Attribute-Level Voice Control in Speech Generation

Authors: Xianhao Zhou, Jianghao Wu

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

Strength: 通过 5,940 个配对输出系统性揭示三个语音生成系统中 54.5%–95.8% 的 target-responsive 输出同时存在 substantial off-target 声学变化，且变化模式因系统而异
Weakness: VoDER-Cal 的联合成功率仅 14.3%，相对 Target-only 的 binary 改善不显著（+0.19pp, 95% CI 包含 0），核心增益来自候选池扩大而非重排序规则
Full review: Claude Code 全文七公理审稿

Natural-language descriptions have become a flexible interface for controlling generated speech. Existing evaluations largely assess whether an output matches a prompt, but prompt matching alone does not reveal whether characteristics outside the intended change remain stable. We examine this distinction through a controlled paired audit of three speech-generation systems: CosyVoice3, VoxCPM2, and Fish-Speech-S2. The evaluation contains 5,940 outputs spanning six reference speakers, ten texts, three random seeds, and eleven conditions. Using acoustic, prosodic, content, and speaker measurements, we find that responses in the expected target direction are frequently accompanied by changes outside descriptor-specific signal-level target sets. This pattern remains among outputs whose target response exceeds baseline seed variation, and the accompanying changes differ substantially across systems. We further introduce VoDER-Cal, a training-free candidate selector that retains sufficiently strong target responses while favoring smaller off-target deviations. A three-candidate pool raises the joint success rate from 4.8% under single-sample direct generation to approximately 14% for all candidate-selection policies. Within the matched three-candidate budget, VoDER-Cal reduces held-out off-target deviation from 0.344 under target-only selection to 0.276 and improves listener-rated preservation. Preservation-sensitive evaluation therefore complements prompt-adherence evaluation, while candidate reranking offers a practical inference-time improvement. Code, configuration files, and analysis scripts are available at https://github.com/intelland/VoDER

8. UOT-IR: Structured Routing of High-Polyphony Symbolic Music into Fixed-Budget Representations

Authors: Ziyue Kang, Nan Nan, Chenhao Lin, Xiaohong Guan

Categories: cs.SD, cs.AI, cs.LG | 8 pages, 2 figures Score: 5.60/10 (Obj:6 Id:5 Ind:8 Comp:5 Eff:5 Nov:7 Rep:5)

Strength: 将固定预算符号音乐压缩重构为结构化路由问题并用 UOT 求解，在 template standardization 上 SC=14.7165、BC=0.3406 大幅优于所有 baseline，问题表述和 OT 迁移均是真实创新。
Weakness: Adaptive preservation 中 Note-F1=0.9120 vs Greedy 0.9113 差距仅 0.0007 缺乏显著性检验，方法含 5 阶段/4 cost 组件/TTA/Viterbi 多个工程模块但组件必要性分析浅，仅在单一数据集 SymphonyNet 评测且代码未开源。
Full review: Claude Code 全文七公理审稿

High-polyphony symbolic music is increasingly used in generation, analysis, and arrangement, yet many downstream tasks require bounded representations with fixed tracks or slots. Converting richly orchestrated scores into compact forms is therefore necessary, but existing approaches relying on heuristic simplification or generic representation-space reduction often fail to preserve structural roles, orchestration compatibility, and playability under strict budgets. To address the issue, this study reformulates the compression problem as a fixed-budget structured routing problem and proposes Unbalanced Optimal Transport for Information Routing (UOT-IR), a training-free framework based on constrained unbalanced optimal transport. UOT-IR combines an orchestration prior, adaptive marginal relaxation, temporal decoding, and playability-aware projection to produce compact and musically coherent bounded representations. This work further studies two practical settings under the same slot budget: template standardization, which maps each input to a predefined bounded template, and adaptive preservation, which retains representative content without assuming an external template. Experiments on the SymphonyNet corpus show that UOT-IR delivers strong overall performance across both settings, including the best Note-F1 in adaptive preservation (0.9120), together with the lowest structural cost (14.7165) and bad structural confusion rate (0.3406) in template standardization. This work establishes a principled paradigm for fixed-budget symbolic music compression, offering a practical path toward compact, structured, and musically coherent symbolic representations.

9. Scene2Sound: Auditory-Grounded Soundscape Generation for 3D Gaussian Worlds

Authors: Masaki Yoshida, Ren Togo, Takahiro Ogawa, Miki Haseyama

Categories: cs.CV, cs.MM, cs.SD | 14 pages. Project page: https://masaki-lmd.github.io/scene2sound/ Score: 5.50/10 (Obj:8 Id:5 Ind:5 Comp:4 Eff:5 Nov:5 Rep:8)

Strength: 定义了”为 3DGS 世界生成空间一致 soundscape”这一真实新任务，并提出 training-free pipeline + 2 个新空间一致性指标，在自构造指标和 user study 上验证了空间一致性优势。
Weakness: 方法是 6 个已有模块的工程组合（唯一新算法 Gaussian set matching 未证明相对简单替代的必要性），核心评测指标由作者自构造且仅 24 样本 FAD 可靠性存疑，SonoWorld re-impl. 在 FAD 上优于本文但公平性无法独立验证。
Full review: Claude Code 全文七公理审稿

3D Gaussian Splatting (3DGS) turns captured or generated imagery into photorealistic 3D world simulations that users can freely explore, yet these worlds remain silent. Because existing audio generation methods condition on a single image or viewpoint, their sound is tied to that observation and cannot stay consistent while a listener moves. We introduce the task of generating a spatially consistent soundscape for a given 3DGS world through auditory grounding, identifying which objects in the world should emit sound and anchoring each to a persistent 3D position, and present Scene2Sound, a training-free framework built on this grounding. From the input world alone, our pipeline selects viewpoints that jointly cover the scene, identifies sound-emitting objects with a vision-language model, and associates the multi-view detections into 3D instances through Gaussian set matching, which measures the overlap between the Gaussian sets that render each detection. Each source then receives generated audio that a standard object-based audio engine spatializes in real time at arbitrary listener poses. We further propose two spatial-consistency metrics, one testing whether rendered audio responds consistently to listener motion and one testing whether the claimed sources are supported by views held out from their placement. On a curated set of generated 3DGS worlds and on 3DGS scenes generated from real-world 360-degree captures, Scene2Sound preserves the audio quality of strong per-viewpoint baselines while remaining spatially consistent where per-viewpoint and single-panorama pipelines do not, and a user study confirms the perceptual benefit. Project page: https://masaki-lmd.github.io/scene2sound/.

10. Simulation-Based Plate-Reverb Parameter Estimation from a Single Impulse Response

Authors: Minhui Lu, Joshua D. Reiss

Categories: eess.AS, cs.LG | Accepted as a challenge paper at the 29th International Conference on Digital Audio Effects, Cambridge, MA, USA, 2026 Score: 5.50/10 (Obj:8 Id:5 Ind:5 Comp:8 Eff:5 Nov:3 Rep:8)

Strength: 方法极简（372维手工特征+ET/HGB平均），一步推理比默认PSO快210×（10.75s vs 2252.59s），在12条独立验证集上NMSE降低71.9%，全部基于公开模拟器可复现。
Weakness: 验证集仅36条且全部来自同一ModalPlate模拟器闭环，未与DiffSound等学习反演方法比较，PSO比较仅单次默认配置，方法为标准tree ensemble无机制创新，IR 0012出现18.56 dB严重失配。
Full review: Claude Code 全文七公理审稿

We present a simulation-trained, non-iterative estimator for Task A of the 1st DAFx Parameter Estimation Challenge. Each unnormalized plate-reverb impulse response is summarized by amplitude, spectral, and decay descriptors, and an ensemble of tree regressors estimates the six target parameters in one pass. Across two independent synthetic validation sets, the normalized models outperform the training-set mean and an earlier raw-regression baseline. On a shared set, the final ensemble also outperforms a single run of the official default PSO at substantially lower inference cost. Since the official labels are hidden, parameter accuracy is measured on simulator-matched data, and the released responses support only audio-side consistency checks. The estimator returns point estimates without uncertainty.

Authors: Minhui Lu, Joshua D. Reiss

Categories: eess.AS, cs.LG | Accepted as a challenge paper at the 29th International Conference on Digital Audio Effects (DAFx 2026), Cambridge, MA, USA, September, 2026 Score: 5.16/10 (Obj:8 Id:5 Ind:5 Comp:8 Eff:5 Nov:5 Rep:8)

Strength: 将密集模态估计分解为”频段计数预测→稠密网格初始化→有界可微衰减/增益校准”三阶段，在两个合成验证集上将 local challenge-style error 相对官方默认 peak-picking baseline 降低约 66%，主要增益来自缓解模态计数严重低估（官方 ~68 模态 vs 参考 ~5240–6072）。
Weakness: 仅与未调参的官方默认 baseline 比较，未对比 ESPRIT/matrix-pencil 等经典强方法或 Yu et al. differentiable all-pole filters 等近期工作；验证集仅 8+12 个样本且参与超参选择，衰减/增益误差仍达 0.69–0.85（接近随机水平），未报告官方隐藏标签分数。
Full review: Claude Code 全文七公理审稿

Task B of the 1st DAFx Parameter Estimation Challenge requires estimating the frequencies, decay rates, gains, and number of modes in a dense plate-reverb impulse response. Weak and overlapping modes make sparse peak detection prone to severe undercounting. We train an ExtraTrees regressor on simulator-generated data to predict mode counts in four frequency bands. These counts define dense frequency grids, after which a differentiable all-pole resonator model refines decay and gain while keeping frequency fixed. On two separate synthetic validation sets, the system reduces a local challenge-style error by about 66% relative to the official default peak-picking baseline. The improvement is mainly associated with lower mode-count mismatch, while decay and gain remain the largest error sources. These findings support separating modal-density estimation from continuous parameter fitting.

12. REIMU: Efficient Heterogeneous Hierarchical Reasoning for SSL-Based Speech Deepfake Detection

Authors: Kwok-Ho Ng, Tingting Song, Bingwen Feng, Peiya Li

Categories: eess.AS, cs.AI, cs.SD Score: 4.74/10 (Obj:5 Id:5 Ind:9 Comp:5 Eff:2 Nov:4 Rep:9)

Strength: 系统性受控实验设计（4 SSL frontends × 4 架构配置 × 3 评测集），发现异构算子分配（MHSA high-level + GDN2 low-level）在同构 HRM 不改善的基础上提供了更优配置，且下游 backbone 参数减少 10.8%。
Weakness: 绝对性能与 SOTA 差距巨大（21DF 最佳 9.54% vs AASIST+wav2vec 2.0 的 2.85%），未与 AASIST 系列强 baseline 直接比较，最终消融中 SSL fine-tuning 和 data augmentation 的贡献远大于 backbone 架构差异，核心发现“异构>同构”仅在部分配置中成立。
Full review: Claude Code 全文七公理审稿

The increasing realism of speech generated by text-to-speech and voice conversion systems poses growing challenges to media integrity and voice authentication. Self-supervised learning (SSL) has substantially advanced speech deepfake detection, where downstream backbones conventionally process SSL representations through a single forward pass. This work investigates the practical effectiveness of recurrent hierarchical reasoning for this task. We term this controlled study REIMU and systematically compare conventional single-pass backbones, weight-shared recurrence, homogeneous HRM, and heterogeneous HRM across four Base-scale SSL frontends. We further examine heterogeneous high- and low-level modules that combine self-attention with linear attention. Experiments on the ASVspoof 2019 and 2021 evaluation sets show that recurrence and hierarchical decomposition do not inherently improve detection, whereas heterogeneous operator assignment provides a more competitive configuration. Notably, the heterogeneous design remains competitive while using 10.8\% fewer downstream parameters than the matched baseline, demonstrating its potential for parameter-efficient speech deepfake detection.