Icassp-2026

ICASSP 2026 - 语音合成 共 63 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 T-Cache: Fast Inference For Masked Generative Transformer-Ba 9.0分 前25% 🥈 Wavenext 2: Convnext-Based Fast Neural Vocoders with Residua 9.0分 前25% 🥉 VoXtream: Full-Stream Text-To-Speech With Extremely Low Late 8.5分 前25% 4. EMORL-TTS: Reinforcement Learning for Fine-Grained Emotion C 8.5分 前25% 5. No Verifiable Reward for Prosody: Toward Preference-Guided P 8.0分 前25% 6. Marco-Voice: A Unified Framework for Expressive Speech Synth 8.0分 前25% 7. Neuromamba: Adaptive Frequency Filtering with a Pyramid Mamb 8.0分 前25% 8. Group Relative Policy Optimization for Text-to-Speech with L 8.0分 前25% 9. Do You Hear What I Mean? Quantifying the Instruction-Percept 8.0分 前25% 10. OV-INSTRUCTTTS: Towards Open-Vocabulary Instruct Text-to-Spe 8.0分 前25% 11. HD-PPT: Hierarchical Decoding of Content- and Prompt-Prefere 8.0分 前25% 12. Emotion-Aligned Generation in Diffusion Text to Speech Model 8.0分 前25% 13. Measuring Prosody Diversity in Zero-Shot TTS: A New Metric, 8.0分 前25% 14. DAIEN-TTS: Disentangled Audio Infilling for Environment-Awar 8.0分 前25% 15. BridgeCode: A Dual Speech Representation Paradigm for Autore 8.0分 前25% 16. Continuous-Token Diffusion for Speaker-Referenced TTS in Mul 8.0分 前10% 17. Prosody-Guided Harmonic Attention for Phase-Coherent Neural 8.0分 前25% 18. Optimizing Speech Language Models for Acoustic Consistency 8.0分 前25% 19. NCF-TTS: Enhancing Flow Matching Based Text-To-Speech with N 8.0分 前25% 20. ARCHI-TTS: A Flow-Matching-Based Text-to-Speech Model with S 8.0分 前25% 21. EMG-to-Speech with Fewer Channels 7.5分 前25% 22. VividTalker: A Modular Framework for Expressive 3D Talking A 7.5分 前25% 23. Real-Time Streaming MEL Vocoding with Generative Flow Matchi 7.5分 前25% 24. From Hallucination to Articulation: Language Model-Driven Lo 7.5分 前25% 25. SynParaSpeech: Automated Synthesis of Paralinguistic Dataset 7.5分 前25% 26. Asynchrony-Aware Decoupled Multimodal Control for Cued Speec 7.5分 前10% 27. DMP-TTS: Disentangled Multi-Modal Prompting for Controllable 7.5分 前25% 28. RRPO: Robust Reward Policy Optimization for LLM-Based Emotio 7.5分 前25% 29. Syncspeech: Efficient and Low-Latency Text-to-Speech Based o 7.5分 前25% 30. Principled Coarse-Grained Acceptance For Speculative Decodin 7.5分 前25% 31. SPADE: Structured Pruning and Adaptive Distillation for Effi 7.5分 前25% 32. Entropy-Guided GRVQ for Ultra-Low Bitrate Neural Speech Code 7.5分 前25% 33. Discrete Diffusion for Generative Modeling of Text-Aligned S 7.5分 前25% 34. Emotional Dimension Control in Language Model-Based Text-To- 7.5分 前25% 35. Beyond Global Emotion: Fine-Grained Emotional Speech Synthes 7.5分 前25% 36. QFOCUS: Controllable Synthesis for Automated Speech Stress E 7.5分 前50% 37. Synthetic yet Striking? Assessing Vocal Charisma in TTS via 7.5分 前25% 38. TMD-TTS: A Unified Tibetan Multi-Dialect Text-to-Speech Fram 7.5分 前25% 39. Deep Dubbing: End-to-End Auto-Audiobook System with Text-to- 7.5分 前25% 40. Erasing Your Voice Before it’s Heard: Training-Free Speaker 7.5分 前25% 41. InstructAudio: Unified Speech and Music Generation with Natu 7.5分 前25% 42. GLA-GRAD++: An Improved Griffin-Lim Guided Diffusion Model f 7.5分 前25% 43. Int-MeanFlow: Few-Step Speech Generation with Integral Veloc 7.5分 前25% 44. Training Flow Matching Models with Reliable Labels via Self- 7.5分 前25% 45. Hierarchical Discrete Flow Matching For Multi-Codebook Codec 7.5分 前25% 46. Frame-Stacked Local Transformers for Efficient Multi-Codeboo 7.5分 前25% 47. Direct Preference Optimization For Speech Autoregressive Dif 7.5分 前25% 48. MirrorTalk: Forging Personalized Avatars Via Disentangled St 7.0分 前25% 49. Residual Tokens Enhance Masked Autoencoders for Speech Model 7.0分 前50% 50. SP-MCQA: Evaluating Intelligibility of TTS Beyond the Word L 7.0分 前50% 51. SPAM: Style Prompt Adherence Metric for Prompt-Based TTS 7.0分 前50% 52. Gelina: Unified Speech and Gesture Synthesis Via Interleaved 7.0分 前50% 53. Retrieval-Based Speculative Decoding For Autoregressive Spee 7.0分 前50% 54. T-Mimi: A Transformer-Based Mimi Decoder for Real-Time On-Ph 7.0分 前50% 55. Wave-Trainer-Fit: Neural Vocoder With Trainable Prior And Fi 7.0分 前25% 56. EmoShift: Lightweight Activation Steering for Enhanced Emoti 7.0分 前50% 57. Task Vector in TTS: Toward Emotionally Expressive Dialectal 7.0分 前50% 58. Quantifying Speaker Embedding Phonological Rule Interactions 7.0分 前25% 59. PFluxTTS: Hybrid Flow-Matching TTS with Robust Cross-Lingual 7.0分 前50% 60. LP-CFM: Perceptual Invariance-Aware Conditional Flow Matchin 7.0分 前25% 61. SFM-TTS: Lightweight and Rapid Speech Synthesis with Flexibl 7.0分 前25% 62. MELA-TTS: Joint Transformer-Diffusion Model with Representat 7.0分 前25% 63. Combining Multi-Order Attention and Multi-Resolution Discrim 6.5分 前50% 📋 论文详情 🥇 T-Cache: Fast Inference For Masked Generative Transformer-Based TTS Via Prompt-Aware Feature Caching 🔥 9.0/10 | 前25% | #语音合成 | #实时处理 | #零样本 #语音大模型 ...

ICASSP 2026 - 语音增强 #对抗防御 共 1 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Adversarial Defense via Generative Speech Enhancement Module 7.5分 前25% 📋 论文详情 🥇 Adversarial Defense via Generative Speech Enhancement Module ✅ 7.5/10 | 前25% | #语音增强 #对抗防御 | #语音增强 #数据增强 | #语音增强 #对抗防御 👥 作者与机构 第一作者：未说明 通讯作者：未说明 作者列表：Chi-Tao Chen（国立中央大学资讯工程学系），Chun-Shien Lu（中央研究院资讯科技研究所），Jia-Ching Wang（国立中央大学资讯工程学系） 💡 毒舌点评 本文巧妙地将对抗防御问题转化为语音增强任务，使用一个轻量级（2M参数）且高效的生成模型（MP-SENet）实现了在多个数据集和攻击类型下的出色防御效果，推理速度远超基于扩散模型的竞品。然而，其核心防御机制（高斯噪声注入+增强）在理论上可能不够“坚固”，面对精心设计的自适应攻击时（如论文表5），性能仍有显著下降，且在SC09这一基准上并未超越最强的对比方法AudioPure。 🔗 开源详情 代码：提供了官方GitHub仓库链接：apoman123/SpeechEnhancementDefense。 模型权重：论文中提及使用了在DNS Challenge上预训练的MP-SENet模型，但未明确是否公开其微调后的防御专用权重。 数据集：使用了公开数据集：SC09（Google Speech Commands子集）， VCTK， QKWS， DNS-Challenge。 Demo：未提及。 复现材料：给出了关键的训练数据增强细节（噪声dBFS范围及最优值）、损失函数公式与权重、攻击参数设置。但优化器、学习率等训练配置未说明。 引用的开源项目：依赖了公开模型MP-SENet，并引用了多个基线方法和攻击方法的开源实现（如DefenseGAN, AudioPure, PGD攻击代码等）。 📌 核心摘要 ...

ICASSP 2026 - 语音增强 共 75 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 A Lightweight Fourier-Based Network for Binaural Speech Enha 8.5分 前25% 🥈 DiTSE: High-Fidelity Generative Speech Enhancement via Laten 8.5分 前10% 🥉 Towards Lightweight Adaptation of Speech Enhancement Models 8.5分 前25% 4. FastEnhancer: Speed-Optimized Streaming Neural Speech Enhanc 8.5分 前25% 5. DisContSE: Single-Step Diffusion Speech Enhancement based on 8.5分 前10% 6. Sidon: Fast and Robust Open-Source Multilingual Speech Resto 8.5分 前25% 7. Spike-Driven Low-Power Speech Bandwidth Extension 8.0分 前25% 8. MAGE: A Coarse-to-Fine Speech Enhancer with Masked Generativ 8.0分 前25% 9. Deep Learning-Based Joint Optimization of Adaptive Feedback 8.0分 前25% 10. HyFlowSE: Hybrid End-To-End Flow-Matching Speech Enhancement 8.0分 前25% 11. HCGAN: Harmonic-Coupled Generative Adversarial Network for S 8.0分 前50% 12. Exploring Resolution-Wise Shared Attention in Hybrid Mamba-U 8.0分 前25% 13. Mixture To Beamformed Mixture: Leveraging Beamformed Mixture 8.0分 前25% 14. Modeling Strategies For Speech Enhancement in The Latent Spa 8.0分 前50% 15. LAFUFU: Latent Acoustic Features For Ultra-Fast Utterance Re 8.0分 前25% 16. Influence of Clean Speech Characteristics on Speech Enhancem 8.0分 前25% 17. LipsAM: Lipschitz-Continuous Amplitude Modifier for Audio Si 7.5分 前25% 18. MSANET: Multi-Scale Semantic Aggregation Network for Brain-A 7.5分 前25% 19. Bone-Conduction Guided Multimodal Speech Enhancement with Co 7.5分 前25% 20. The 3rd Clarity Prediction Challenge: A Machine Learning Cha 7.5分 前25% 21. Two-Stage Language Model Framework for Acoustic Echo Cancell 7.5分 前25% 22. E2E-AEC: Implementing An End-To-End Neural Network Learning 7.5分 前25% 23. SpatialNet-Echo: Real-Time Acoustic Echo Cancellation via In 7.5分 前25% 24. A Stabilized Hybrid Active Noise Control Algorithm of GFANC 7.5分 前25% 25. Enhancing Speech Intelligibility Prediction for Hearing Aids 7.5分 前25% 26. H-nnPBFDAF: Hierarchical Neural Network Partitioned Block Fr 7.5分 前25% 27. Joint Deep Secondary Path Estimation and Adaptive Control fo 7.5分 前25% 28. Enhancing Noise Robustness for Neural Speech Codecs Through 7.5分 前25% 29. Low-Bandwidth High-Fidelity Speech Transmission with Generat 7.5分 前25% 30. From Diet to Free Lunch: Estimating Auxiliary Signal Propert 7.5分 前25% 31. Beamforming Using Virtual Microphones for Hearing Aid Applic 7.5分 前50% 32. I-DCCRN-VAE: An Improved Deep Representation Learning Framew 7.5分 前25% 33. Do We Need EMA for Diffusion-Based Speech Enhancement? Towar 7.5分 前50% 34. Hair Noise Analysis and Mitigation for Smart Glasses Audio C 7.5分 前25% 35. Are Modern Speech Enhancement Systems Vulnerable to Adversar 7.5分 前25% 36. UJCodec: An End-to-end Unet-Style Codec for Joint Speech Com 7.5分 前25% 37. Spatial Covariance Matrix Reconstruction for Speech Enhancem 7.5分 前25% 38. Training-Free Inference-Time Scaling for Audio Source Separa 7.5分 前25% 39. Forward Convolutive Prediction for Frame Online Monaural Spe 7.5分 前50% 40. MeanFlowSE: One-Step Generative Speech Enhancement via Condi 7.5分 前10% 41. FlowSE-GRPO: Training Flow Matching Speech Enhancement via O 7.5分 前25% 42. Aligning Generative Speech Enhancement with Perceptual Feedb 7.5分 前25% 43. PG-SE: Predictive Acceleration and Correction for Generative 7.5分 前25% 44. Dynamically Slimmable Speech Enhancement Network with Metric 7.5分 前25% 45. Lightweight Phoneme-Conditioned Bandwidth Extension for Body 7.5分 前25% 46. Fast-ULCNet: A Fast and Ultra Low Complexity Network for Sin 7.5分 前25% 47. ParaGSE: Parallel Generative Speech Enhancement with Group-V 7.5分 前25% 48. High-Fidelity Speech Enhancement Via Discrete Audio Tokens 7.5分 前25% 49. DISSR: Disentangling Speech Representation for Degradation-P 7.5分 前25% 50. Ranking The Impact of Contextual Specialization in Neural Sp 7.5分 前25% 51. BSMP-SENet:Band-Split Magnitude-Phase Network for Speech Enh 7.0分 前25% 52. DECAF: Dynamic Envelope Context-Aware Fusion for Speech-Enve 7.0分 前25% 53. DAT-CFTNet: Speech Enhancement for Cochlear Implant Recipien 7.0分 前50% 54. Acoustic Teleportation Via Disentangled Neural Audio Codec R 7.0分 前25% 55. Reference Microphone Selection for Guided Source Separation 7.0分 前50% 56. Low-Latency Audio Front-End Region-of-Interest Beamforming f 7.0分 前25% 57. AmbiDrop: Array-Agnostic Speech Enhancement Using Ambisonics 7.0分 前50% 58. Joint Multichannel Acoustic Feedback Cancellation and Speake 7.0分 前25% 59. Gdiffuse: Diffusion-Based Speech Enhancement with Noise Mode 7.0分 前25% 60. An Efficient Neural Network for Modeling Human Auditory Neur 7.0分 前25% 61. Shortcut Flow Matching for Speech Enhancement: Step-Invarian 7.0分 前25% 62. Generalizability of Predictive and Generative Speech Enhance 7.0分 前50% 63. Mambaformer: State-Space Augmented Self-Attention with Downu 7.0分 前25% 64. Auditory-Inspired Transformer for Binaural Speech Enhancemen 7.0分 前25% 65. A State-Dependent Markov Diffusion Process for Generative Sp 6.5分 前25% 66. Confidence-Based Filtering for Speech Dataset Curation with 6.5分 前50% 67. Sampling-Rate-Agnostic Speech Super-Resolution Based on Gaus 6.5分 前25% 68. Low-Frequency Harmonic Control for Speech Intelligibility in 6.5分 前50% 69. What the student learns in knowledge distillation: A subspac 6.5分 前50% 70. MeanSE: Efficient Generative Speech Enhancement with Mean Fl 6.5分 前25% 71. On The Design of Efficient Neural Methods for Geometry-Agnos 6.5分 前50% 72. Position-Invariant Fine-Tuning Of Speech Enhancement Models 6.5分 前50% 73. Stereophonic Acoustic Echo Cancellation Using an Improved Af 6.0分 前50% 74. Towards Real-Time Generative Speech Restoration with Flow-Ma 6.0分 前50% 75. Is Phase Really Needed for Weakly-Supervised Dereverberation 6.0分 前50% 📋 论文详情 🥇 A Lightweight Fourier-Based Network for Binaural Speech Enhancement with Spatial Cue Preservation 🔥 8.5/10 | 前25% | #语音增强 | #深度学习 | #轻量级模型 #空间音频 ...

ICASSP 2026 - 语音大模型 共 3 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Cross-Lingual Interleaving for Speech Language Models 7.5分 前25% 🥈 Cross-Modal Knowledge Distillation for Speech Large Language 7.0分 前25% 🥉 SpeechMapper: Speech-To-Text Embedding Projector for LLMs 7.0分 前25% 📋 论文详情 🥇 Cross-Lingual Interleaving for Speech Language Models ✅ 7.5/10 | 前25% | #语音大模型 | #预训练 #多语言 | #预训练 #多语言 👥 作者与机构 第一作者：Adel Moumen（Department of Engineering, University of Cambridge, UK） 通讯作者：未说明 作者列表：Adel Moumen（Department of Engineering, University of Cambridge, UK）、Guangzhi Sun（Department of Engineering, University of Cambridge, UK）、Philip C. Woodland（Department of Engineering, University of Cambridge, UK） 💡 毒舌点评 ...

ICASSP 2026 - 语音对话系统 共 10 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 DOMA: Leveraging Diffusion Language Models with Adaptive Pri 8.5分 前25% 🥈 PersonaPlex: Voice and Role Control for Full Duplex Conversa 8.5分 前25% 🥉 UTI-LLM: A Personalized Articulatory-Speech Therapy Assistan 7.5分 前25% 4. A Dataset of Robot-Patient and Doctor-Patient Medical Dialog 7.5分 前25% 5. Game-Time: Evaluating Temporal Dynamics in Spoken Language M 7.5分 前25% 6. The Role of Prosodic and Lexical Cues in Turn-Taking with Se 7.5分 前25% 7. Vocalnet-M2: Advancing Low-Latency Spoken Language Modeling 7.5分 前25% 8. Easy Turn: Integrating Acoustic and Linguistic Modalities fo 7.0分 前25% 9. Still Thinking or Stopped Talking? Dialogue Silence Intentio 6.5分 前25% 10. Enhancing Dialogue-Related Speech Tasks with Generated Spoke 6.5分 前25% 📋 论文详情 🥇 DOMA: Leveraging Diffusion Language Models with Adaptive Prior for Intent Classification and Slot Filling 🔥 8.5/10 | 前25% | #语音对话系统 | #扩散模型 | #意图识别 #槽填充 ...

ICASSP 2026 - 语音情感识别 共 49 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Context-Aware Dynamic Graph Learning for Multimodal Emotion 8.8分 前10% 🥈 Prompt-Guided Mixture-of-Experts for Robust Multimodal Senti 8.5分 前25% 🥉 Clue2Emo: A Brain-Inspired Framework for Open-Vocabulary Mul 8.5分 前25% 4. Attention-Weighted Centered Kernel Alignment for Knowledge D 8.0分 前25% 5. Staged Diffusion with Hybrid Mixture-of-Experts (MOE) for Mu 8.0分 前25% 6. DGSDNet: Dual-Graph Spectral Diffusion Network for Incomplet 8.0分 前25% 7. Graph-based Modality Alignment for Robustness in Conversatio 8.0分 前25% 8. Multimodal Self-Attention Network with Temporal Alignment fo 8.0分 前25% 9. It Is Personal: The Importance of Personalization for Recogn 8.0分 前25% 10. AMBER2: Dual Ambiguity-Aware Emotion Recognition Applied to 8.0分 前25% 11. MI-Fuse: Label Fusion for Unsupervised Domain Adaptation wit 8.0分 前25% 12. Speech Emotion Recognition based on Hierarchical Transformer 8.0分 前25% 13. Affect-Jigsaw: Integrating Core and Peripheral Emotions for 8.0分 前25% 14. When Audio Matters: A Lightweight, Hierarchical Fusion Model 8.0分 前25% 15. Behind the Scenes: Mechanistic Interpretability of Lora-Adap 7.5分 前25% 16. Encoding Emotion Through Self-Supervised Eye Movement Recons 7.5分 前25% 17. Inter-Dialog Contrastive Learning for Multimodal Emotion Rec 7.5分 前25% 18. ADH-VA: Adaptive Directed-Hypergraph Convolution with VA Con 7.5分 前10% 19. SURE: Synergistic Uncertainty-Aware Reasoning for Multimodal 7.5分 前25% 20. Tpeformer: Temporal Patch Embedding Transformer 7.5分 前25% 21. LETPAV: Lexicon-Enhanced Text with Progressive Audio-Visual 7.5分 前25% 22. Multimodal Variational Graph Network for Multimodal Sentimen 7.5分 前25% 23. Diffemotalk: Audio-Driven Facial Animation with Fine-Grained 7.5分 前25% 24. MECap-R1: Emotion-Aware Policy with Reinforcement Learning f 7.5分 前25% 25. FIDIC:Fine-Grained Conversational Emotion Recognition via In 7.5分 前25% 26. Whisper-QF: Leveraging Dual Cross-Attention Q-Former for Spe 7.5分 前25% 27. Temporal Graph Modeling for Speech Emotion Recognition Using 7.5分 前25% 28. Mixture-of-Experts Based Soft-Label Learning for Multi-Label 7.5分 前25% 29. Multi-Channel Speech Enhancement for Cocktail Party Speech E 7.5分 前25% 30. Evaluating Emotion Recognition in Spoken Language Models on 7.5分 前50% 31. InconVAD: A Two-Stage Dual-Tower Framework for Multimodal Em 7.5分 前25% 32. MSF-SER: Enriching Acoustic Modeling with Multi-Granularity 7.5分 前25% 33. Rationale-Guided Learning for Multimodal Emotion Recognition 7.0分 前25% 34. Bimodal Fusion Framework for Dynamic Facial Expression Recog 7.0分 前25% 35. Stress Prediction from Temporal Emotion Trajectories in Clin 7.0分 前25% 36. Emo-TTA: Improving Test-Time Adaptation of Audio-Language Mo 7.0分 前25% 37. Test Time Adaptation for Speech Emotion Recognition 7.0分 前25% 38. Plug-and-Play Emotion Graphs for Compositional Prompting in 7.0分 前25% 39. Reasoning Driven Captions to Assist Noise Robust Speech Emot 7.0分 前25% 40. EmoTri-RL: Emotion- and Cause-Aware Reinforcement Learning f 7.0分 前25% 41. Modeling Both Intra- And Inter-Utterance Variability for Con 6.5分 前25% 42. DDSR-Net: Robust Multimodal Sentiment Analysis via Dynamic M 6.5分 前50% 43. Scaling Ambiguity: Augmenting Human Annotation in Speech Emo 6.5分 前50% 44. Recovering Performance in Speech Emotion Recognition from Di 6.5分 前50% 45. B-GRPO: Unsupervised Speech Emotion Recognition Based on Bat 6.5分 前50% 46. Leveraging Large Speech Language Models as Evaluators for Ex 6.5分 前50% 47. Gen-SER: When the Generative Model Meets Speech Emotion Reco 6.5分 前50% 48. SmoothCLAP: Soft-Target Enhanced Contrastive Language-Audio 6.5分 前50% 49. Acoustic and Facial Markers of Perceived Conversational Succ 6.0分 前50% 📋 论文详情 🥇 Context-Aware Dynamic Graph Learning for Multimodal Emotion Recognition with Missing Modalities 🔥 8.8/10 | 前10% | #语音情感识别 | #多模态模型 | #大语言模型 #多任务学习 ...

ICASSP 2026 - 语音摘要 共 1 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Semantic Anchor Transfer from Short to Long Speech in a Dist 7.5分 前25% 📋 论文详情 🥇 Semantic Anchor Transfer from Short to Long Speech in a Distillation-Based Summarization Framework ✅ 7.5/10 | 前25% | #语音摘要 | #知识蒸馏 | #端到端 #迁移学习 👥 作者与机构 第一作者：Xiang He (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心) 通讯作者：Liang He (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心；新疆大学智能科学与技术学院；清华大学电子工程系) 作者列表：Xiang He (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心)、Xuejian Zhao (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心)、Longwei Li (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心)、Liang He (新疆大学计算机科学与技术学院，新疆多模态信息技术工程研究中心；新疆大学智能科学与技术学院；清华大学电子工程系) 💡 毒舌点评 ...

ICASSP 2026 - 语音活动检测 共 5 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Lingometer: On-Device Personal Speech Word Counting System 8.0分 前25% 🥈 EEND-SAA: Enrollment-Less Main Speaker Voice Activity Detect 7.5分 前25% 🥉 Dual Data Scaling for Robust Two-Stage User-Defined Keyword 7.5分 前25% 4. EdgeSpot: Efficient and High-Performance Few-Shot Model for 7.5分 前25% 5. TVP-UNet: Threshold Variance Penalty U-Net for Voice Activit 7.0分 前25% 📋 论文详情 🥇 Lingometer: On-Device Personal Speech Word Counting System 🔥 8.0/10 | 前25% | #语音活动检测 | #端到端 | #低资源 #数据增强 ...

ICASSP 2026 - 语音理解 共 2 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Exploring Fine-Tuning Of Large Audio Language Models For Spo 8.0分 前25% 🥈 Scaling Spoken Language Models with Syllabic Speech Tokeniza 7.0分 前25% 📋 论文详情 🥇 Exploring Fine-Tuning Of Large Audio Language Models For Spoken Language Understanding Under Limited Speech Data 🔥 8.0/10 | 前25% | #语音理解 | #迁移学习 | #低资源 #多语言 👥 作者与机构 第一作者：Youngwon Choi (MAUM AI Inc., Republic of Korea) 通讯作者：Huu-Kim Nguyen (∗ 作者列表中标注星号，现单位为 Atmanity Inc., USA) 作者列表： Youngwon Choi (MAUM AI Inc., Republic of Korea) Jaeyoon Jung (MAUM AI Inc., Republic of Korea & Soongsil University, Republic of Korea) Hyeonyu Kim (MAUM AI Inc., Republic of Korea) Huu-Kim Nguyen (MAUM AI Inc., Republic of Korea → 现 Atmanity Inc., USA) Hwayeon Kim (MAUM AI Inc., Republic of Korea) 💡 毒舌点评 ...

ICASSP 2026 - 语音生成 共 1 篇论文 ← 返回 ICASSP 2026 总览 排名 论文 评分 分档 🥇 Why Do Speech Language Models Fail to Generate Semantically 7.0分 前25% 📋 论文详情 🥇 Why Do Speech Language Models Fail to Generate Semantically Coherent Outputs? A Modality Evolving Perspective ✅ 7.0/10 | 前25% | #语音生成 | #模型评估 | #语音大模型 #零样本 👥 作者与机构 第一作者：Hankun Wang（X-LANCE Lab， 上海交通大学计算机科学与技术学院） 通讯作者：Kai Yu（X-LANCE Lab， 上海交通大学计算机科学与技术学院） 作者列表：Hankun Wang（X-LANCE Lab， 上海交通大学）， Haoran Wang（X-LANCE Lab， 上海交通大学）， Yiwei Guo（X-LANCE Lab， 上海交通大学）， Zhihan Li（X-LANCE Lab， 上海交通大学）， Chenpeng Du（X-LANCE Lab， 上海交通大学）， Kai Yu（X-LANCE Lab， 上海交通大学） 💡 毒舌点评 ...