Daily Papers — 2026-06-26

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

1. Grammar-Guided Hierarchical Parsing for Long-form Audio Activity Recognition

Authors: Peng Zhang, Qingyu Luo, Philip J. B. Jackson, Wenwu Wang

Categories: cs.SD, eess.AS | Accepted to Interspeech 2026 Score: 8.2/10 (Obj:9 Id:8 Ind:9 Comp:8 Eff:8 Nov:8)

• Strength: 将神经事件检测与符号语法解析优雅结合，无需高层标签即可实现跨层级推理，显著降低了监督需求并提升时序一致性。
• Weakness: 高度依赖预定义语法规则的完备性及底层事件检测的准确性，在高度嘈杂或分布外事件下解析鲁棒性存疑。

中文摘要： 长形式音频具有内在的层级结构：细粒度事件构成子活动，子活动再组成更高层级的活动，而以往工作往往对各层级分别建模，导致跨层级不一致并需要多层级监督。本文将该问题形式化为基于事件级证据的层级解析：在给定带类别后验的检测事件片段条件下，推断出与顺序一致的”活动-子活动-事件”解析树。作者提出层级活动文法（Hierarchical Activity Grammar）来约束解析过程，从而仅依赖事件级监督即可完成端到端的层级结构推断。该方法有效缓解了跨层级不一致问题，并在长形式音频活动识别任务上取得了更优的性能。

Long-form audio exhibits an inherent hierarchy: fine-grained events form sub-activities, which in turn constitute higher-level activities. Prior work often models these levels separately, leading to cross-level inconsistencies and requiring supervision at multiple levels. We formulate the problem as hierarchical parsing from event-level evidence: given detected event segments with class posteriors, we infer an order-consistent Act-Sub-Event parse tree. We propose Hierarchical Activity Grammar, e

2. What Was That Again? Certified Robustness for Automatic Speech Recognition

Authors: Andrew C. Cullen, Neil Marchant, Jiani Xie, Paul Montague, Benjamin I. P. Rubinstein

Categories: cs.LG, cs.AI, cs.CR, cs.SD | 17 pages Score: 8.0/10 (Obj:9 Id:7 Ind:8 Comp:8 Eff:8 Nov:8)

• Strength: 提出基于E值锦标赛的双重门控认证框架，优雅地避开了昂贵的序列对齐，在高噪声下认证召回率远超基线且大幅降低WER。
• Weakness: 数学证明与实际ASR安全的语义鸿沟仍存，E值机制在极端对抗攻击下的鲁棒性边界需进一步验证。

中文摘要： 自动语音识别（ASR）系统对对抗性扰动和良性扰动均极为敏感，但在已部署系统中检测此类行为却因缺乏真实转录的先验知识而极具挑战。本文提出一种受认证机制启发的方法，无需依赖真实转录即可在部署场景中识别并缓解扰动带来的负面影响。该方法通过降低词错误率（WER）、提升召回率，并削弱置信度与WER之间的Spearman相关性来增强系统鲁棒性。实验表明，该机制能显著改善ASR系统在扰动条件下的可靠性与可信度。这一工作为无参考转录场景下的ASR鲁棒性评估与保障提供了新思路。

Automatic Speech Recognition systems are notoriously both sensitive to adversarial and benign perturbations. While this has been repeatedly demonstrated using reference datasets, detecting such behaviors in deployed systems is incredibly challenging, due to the absence of oracle knowledge of the true transcription. We demonstrate that employing a certification-inspired mechanism can significantly decrease WER, increase recall, and decrease the Spearman correlation between confidence and WER. We

3. HPRO: Hierarchical Progressive Reward Optimization via Preference Extraction for Emotional Text-to-Speech

Authors: Sihang Nie, Xiaofen Xing, Rui Xing, Haoming Li, Ruitong Xiao et al.

Categories: eess.AS, cs.CL, cs.SD | 7 pages, 3 figures, 3 tables; Preprint Score: 7.9/10 (Obj:8 Id:8 Ind:7 Comp:8 Eff:8 Nov:8)

• Strength: 精准识别了情感TTS中偏好优化的两大结构性错配（信息冲突与尺度鸿沟），并通过解耦编解码器与层次化渐进对齐给出了极其对症的解决方案。
• Weakness: 方法的有效性高度依赖HD-Emo codec对内容与风格的完美解耦，若解耦不彻底仍可能面临奖励作弊的风险。

中文摘要： 针对大语言模型驱动的文本转语音（TTS）在监督微调下趋于统计平均韵律、情感表现力受限的问题，本文指出现有偏好驱动优化方法存在信息冲突与奖励套牢两类结构性错配：内容与情感共享隐空间会产生对抗梯度，而单一奖励信号易诱导模型走捷径。为此提出 HPRO——基于偏好提取的分层渐进式奖励优化框架，通过分层解耦内容与情感的表征空间以消除梯度冲突，并借助渐进式奖励机制引导模型从基础自然度向情感表达力逐步过渡，规避奖励套牢陷阱。实验表明，HPRO 在情感表现力与可控性上显著优于标准监督微调及现有偏好优化方法，为情感 TTS 提供了新的优化范式。

Recently, Large Language Model (LLM)-based Text-to-Speech (TTS) models have achieved remarkable naturalness. However, the standard Supervised Fine-Tuning paradigm often converges to statistically averaged prosody, limiting emotional expressiveness. While preference-driven optimization offers a promising alternative, existing approaches suffer from two structural mismatches: information conflict, where content and emotion in a shared latent space produce conflicting gradients, leading to reward h

4. HybridCodec: Modeling Discrete and Continuous Representations for Efficient Speech Language Models

Authors: Artem Ploujnikov, Francesco Verdini, Samir Sadok, Mirco Ravanelli

Categories: cs.LG, cs.AI | Accepted Score: 7.5/10 (Obj:8 Id:7 Ind:8 Comp:6 Eff:8 Nov:7)

• Strength: 有效解决了纯离散语音编码的量化损失问题，通过连续残差显著提升了说话人特征保留率并减少了自回归步数，效果与效率双增。
• Weakness: 混合离散-连续范式及AR+NAR的架构是对已有语音多码本/残差生成模式的增量演进，压缩洞察与范式突破有限。

中文摘要： 这篇论文针对离散音频表征在构建多模态文本-音频系统及将音频能力集成到大语言模型（LLMs）时，因离散化过程导致信息损失并在下游任务中性能下降的问题，提出了一种混合编解码框架HybridCodec。该方法将时间压缩的离散token与降维后的连续残差表征相结合，以离散token捕捉语音的核心语义结构，同时用连续残差补充离散化过程中丢失的细粒度信息，从而在保留离散表征易于与语言模型对接的优势的同时减少信息损失。该框架由一个混合离散-连续编解码器构成，旨在为语音语言模型提供兼顾效率与表达力的音频表征方案。通过这种混合建模方式，HybridCodec期望能在下游语音任务上缓解纯离散表征带来的性能退化，实现更高效的语音语言模型构建。

Discrete audio representations have become increasingly popular for building multimodal text-audio systems and integrating audio capabilities into Large Language Models (LLMs). However, numerous studies report performance degradation on various downstream tasks due to information loss during discretization. To address this, we propose a novel approach combining temporally compressed discrete tokens with dimensionality-reduced continuous residuals. Our framework consists of a hybridized discrete-

5. From General-Purpose Audio Tagging to Spatially Grounded Sound Event Localization and Detection

Authors: Stefano Giacomelli, Stefano Damiano, Claudia Rinaldi, Fabio Graziosi, Toon van Waterschoot

Categories: cs.SD, cs.AI | Technical Report (KU Leuven - UnivAQ) Score: 6.4/10 (Obj:8 Id:6 Ind:8 Comp:4 Eff:7 Nov:5)

• Strength: 将预训练音频 tagging 模型有效迁移至 SELD 任务，工程闭环完整且具备实际效果。
• Weakness: 核心方法多为已有模块（预训练骨干、FOA、track-wise SED等）的组合堆叠，缺乏本质创新与理论洞察。

中文摘要： 本文研究如何将预训练的通用音频标注（GP-AT）模型扩展到具有空间定位能力的声事件定位与检测（SELD）任务。提出的 AT2SELD 框架将预训练的 AT 主干网络与紧凑的一阶 Ambisonics（FOA）空间处理相结合，并采用逐轨 SED 与笛卡尔 DOA 估计、置换感知监督及校准机制。该工作刻画了语义音频先验如何在数据与计算受限条件下支撑面向定位的场景分析。研究旨在揭示通用音频语义先验对空间感知的迁移价值，为低资源场景下的 SELD 提供可行路径。

This report investigates the extension of pretrained General-Purpose Audio Tagging (GP-AT) models toward spatially grounded Sound Event Localization and Detection (SELD). The proposed AT2SELD framework couples a pretrained AT backbone with compact First-Order Ambisonics (FOA) spatial processing, track-wise SED and Cartesian DOA estimation, permutation aware supervision, and calibration. It characterizes how semantic audio priors support localization-aware scene analysis under data, computation,

6. Do Speech Emphasis Models Generalize across Languages and Emotions?

Authors: Megan Wei, Deepali Aneja, Jiaqi Su, Yunyun Wang, Haonan Chen et al.

Categories: cs.CL, cs.AI, cs.LG, cs.SD, eess.AS | Interspeech 2026 Score: 6.1/10 (Obj:9 Id:6 Ind:8 Comp:6 Eff:5 Nov:6)

• Strength: 构建了高质量的多语言多情感重音数据集(MMEE)，填补了该领域真实人类感知标注的空白；系统揭示了现有模型跨语言和跨情感的泛化规律。
• Weakness: 属于数据集与基准测试论文，缺乏方法层面的创新；未提出新的模型或机制来实质性地解决所暴露的跨语言泛化不足问题。

中文摘要： 现有的语音强调检测模型主要在单语言中性朗读语音上训练与评估，其跨语言和跨情感的泛化能力长期缺乏系统研究。为此，作者构建了MMEE语料库，包含7种语言、34种情感/风格类别下专业录制的10,000条富有表现力的语音（共14.13小时），每条样本由10位标注者提供三级感知标签。基于该语料库，作者对两种主流架构在单语及跨语言、跨情感条件下进行了基准测试，揭示了现有模型在语言与情感多样性面前泛化能力不足的问题。该工作为多语言、多情感场景下的语音强调检测提供了新的评测基准与数据资源。

Prosodic emphasis varies across languages, emotions, and speaking styles, yet existing emphasis detection models are largely trained and evaluated on monolingual neutral read speech. We introduce MMEE (Multilingual Multi-Emotion Emphasis), a corpus of 10,000 professionally recorded expressive utterances (14.13 hours) across 7 languages and 34 emotion/style categories, with three-level perceptual labels (10 annotations per sample). We benchmark two state-of-the-art architectures under monolingual

7. A Flexible Encoding Model for Non-Unique Note Alignments

Authors: Suhit Chiruthapudi, Adam Štefunko, Silvan Peter, Patricia Hu, Jan Hajič et al.

Categories: cs.SD, eess.AS | Published at the Music Encoding Conference (MEC), 2026 Score: 5.5/10 (Obj:8 Id:4 Ind:8 Comp:7 Eff:5 Nov:4)

• Strength: 解决了一个真实且被忽视的符号音乐对齐表示问题（如重复练习和数字低音即兴），格式设计向后兼容且最小化。
• Weakness: 本质上是数据格式的工程扩展而非ML方法创新，缺乏模型层面的因果归因和实证效果验证。

中文摘要： 这篇论文针对符号音乐对齐中现有格式只能表示唯一对应关系的局限，提出对Match文件格式进行最小化、向后兼容的扩展，以支持非唯一音符对齐。该方法的核心动机是覆盖排练中的练习重复段以及数字低音即兴实现等需要灵活编码的音乐实践。扩展方案在保持与既有格式兼容的同时，赋予了表示多重对应关系的能力。论文的主要贡献在于为这些此前难以被统一编码的对齐场景提供了一个通用且兼容的解决方案。

Symbolic music alignment links notes in a symbolic performance to their counterparts in a score. While existing alignment encoding formats provide unique correspondences between these notes, there are various musical practices and forms such as practice repetitions in rehearsal and improvised realizations in basso continuo that require a more flexible approach to encoding their alignments. In this paper, we propose a minimal, backward-compatible extension to the Match file format to support such

8. Screening Matters: A Comparative Study of Conventional and Crowdsourced Listening Tests

Authors: Anika Treffehn, Andrea Eichenseer, Emily Kratsch, Nicola Pia

Categories: eess.AS | accepted at Interspeech 2026 Score: 5.1/10 (Obj:8 Id:6 Ind:8 Comp:5 Eff:4 Nov:4)

• Strength: 针对音频评测领域真实存在的众包与实验室测试差异问题，进行了标准的统计对比分析，强调了筛选（Screening）的重要性。
• Weakness: 缺乏方法创新和效果提升，仅是对已有标准（P.800与P.808）的实证比较，未提出新方法或深刻洞察。

中文摘要： 这篇论文针对语音与音频编码技术的主观评测问题，对比研究了传统实验室环境下的P.800 DCR测试与众包环境下的P.808测试两种听音测试方法，核心问题在于众包方式虽具成本和效率优势，但其结果质量是否可与受控实验室测试相媲美。方法上，研究选取经典与神经语音编解码器作为评测对象，分别通过两种测试框架获取主观评分，并开展统计分析以考察二者的一致性与差异。研究通过对比不同编解码器在两种测试条件下的评分分布与统计特性，揭示了众包测试相对实验室测试的偏差特征与适用边界，为评测方法的选择提供了实证依据。

Subjective evaluation remains the most reliable way of testing speech and audio coding techniques. Crowdsourcing the listening task is a cost-efficient and fast way of conducting this evaluation, but the quality of the results tends to be inferior to that of conventional listening tests done in the controlled environment of a laboratory. In this paper, classical and neural speech codecs are evaluated to compare P.808 against P.800 DCR tests. A statistical analysis is conducted to investigate the