音频生成论文趋势 2023–2026

主要结论

• 音频生成论文从 2023 年的 32 篇增长到 2026 年的 136 篇，三年间扩大到约 4.3 倍。
• ICLR 是发音频生成论文最多的会议，2023–2026 年共 135 篇，占四个会议总量的 38%。
• 交叉方向里涨得最快的是应用 · computer vision：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 16 篇。
• 在「语音与音频」的 5 个主题里，它的 2026 年论文数排第 3；同组增长最快的视听学习是 1.8 倍，音频生成是 4.3 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Unified Audio Language Model with Text-Aligned Factorized Audio Tokenization
• ICML 2026PHALAR: Phasors for Learned Musical Audio Representations
• ICLR 2026AC-Foley: Reference-Audio-Guided Video-to-Audio Synthesis with Acoustic Transfer
• CVPR 2026AudioStory Generating Long-Form Narrative Audio with Large Language Models
• NeurIPS 2025🎧MOSPA: Human Motion Generation Driven by Spatial Audio
• ICML 2025Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities
• ICLR 2025ARLON: Boosting Diffusion Transformers with Autoregressive Models for Long Video Generation
• CVPR 2025AudCast: Audio-Driven Human Video Generation by Cascaded Diffusion Transformers
• NeurIPS 2024Vocal Call Locator Benchmark (VCL) for localizing rodent vocalizations from multi-channel audio
• ICML 2024An Independence-promoting Loss for Music Generation with Language Models
• ICLR 2024Error Norm Truncation: Robust Training in the Presence of Data Noise for Text Generation Models
• CVPR 2024FlowVQTalker: High-Quality Emotional Talking Face Generation through Normalizing Flow and Quantization
• NeurIPS 2023High-Fidelity Audio Compression with Improved RVQGAN
• ICML 2023Make-An-Audio: Text-To-Audio Generation with Prompt-Enhanced Diffusion Models
• ICLR 2023AudioGen: Textually Guided Audio Generation
• CVPR 2023Conditional Generation of Audio From Video via Foley Analogies
• NeurIPS 2026AudioCALM: Continuous Autoregressive Language Modeling for Universal Audio Generation
• ICML 2026SonicMaster: Towards Controllable All-in-One Music Restoration and Mastering
• ICLR 2026AudioX: A Unified Framework for Anything-to-Audio Generation
• CVPR 2026Avatar Forcing Real-Time Interactive Head Avatar Generation for Natural Conversation

全部音频生成论文

共 135 篇，第 2 / 3 页

51. ICLR 2026SyncTrack: Rhythmic Stability and Synchronization in Multi-Track Music Generation
52. ICLR 2026TaskCraft: Automated Generation of Agentic Tasks
53. ICLR 2026Text2Arch: A Dataset for Generating Scientific Architecture Diagrams from Natural Language Descriptions
54. ICLR 2026Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation
55. ICLR 2026Token-Based Audio Inpainting via Discrete Diffusion
56. ICLR 2026Topology of Reasoning: Retrieved Cell Complex-Augmented Generation for Textual Graph Question Answering
57. ICLR 2026Toward Complex-Valued Neural Networks for Waveform Generation
58. ICLR 2026UALM: Unified Audio Language Model for Understanding, Generation and Reasoning
59. ICLR 2026UniF$^2$ace: A $\underline{Uni}$fied $\underline{F}$ine-grained $\underline{Face}$ Understanding and Generation Model
60. ICLR 2026Unified Multi-Modal Interactive and Reactive 3D Motion Generation via Rectified Flow
61. ICLR 2026UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation
62. ICLR 2026Unveiling the Potential of Diffusion Large Language Model in Controllable Generation
63. ICLR 2026VibeVoice: Expressive Podcast Generation with Next-Token Diffusion
64. ICLR 2026YuE: Scaling Open Foundation Models for Long-Form Music Generation
65. ICLR 2025A Theory for Token-Level Harmonization in Retrieval-Augmented Generation
66. ICLR 2025Agents' Room: Narrative Generation through Multi-step Collaboration
67. ICLR 2025ARLON: Boosting Diffusion Transformers with Autoregressive Models for Long Video Generation
68. ICLR 2025Automated Proof Generation for Rust Code via Self-Evolution
69. ICLR 2025BirdSet: A Large-Scale Dataset for Audio Classification in Avian Bioacoustics
70. ICLR 2025Both Ears Wide Open: Towards Language-Driven Spatial Audio Generation
71. ICLR 2025Cached Multi-Lora Composition for Multi-Concept Image Generation
72. ICLR 2025Contrastive Learning from Synthetic Audio Doppelgängers
73. ICLR 2025CyberHost: A One-stage Diffusion Framework for Audio-driven Talking Body Generation
74. ICLR 2025Decoupling Layout from Glyph in Online Chinese Handwriting Generation
75. ICLR 2025Diff-2-in-1: Bridging Generation and Dense Perception with Diffusion Models
76. ICLR 2025Diversity-Rewarded CFG Distillation
77. ICLR 2025ECHOPulse: ECG Controlled Echocardio-gram Video Generation
78. ICLR 2025EcoFace: Audio-Visual Emotional Co-Disentanglement Speech-Driven 3D Talking Face Generation
79. ICLR 2025Hallo2: Long-Duration and High-Resolution Audio-Driven Portrait Image Animation
80. ICLR 2025Hydra-SGG: Hybrid Relation Assignment for One-stage Scene Graph Generation
81. ICLR 2025ImageFolder: Autoregressive Image Generation with Folded Tokens
82. ICLR 2025IPDreamer: Appearance-Controllable 3D Object Generation with Complex Image Prompts
83. ICLR 2025Loopy: Taming Audio-Driven Portrait Avatar with Long-Term Motion Dependency
84. ICLR 2025MDSGen: Fast and Efficient Masked Diffusion Temporal-Aware Transformers for Open-Domain Sound Generation
85. ICLR 2025MVTokenFlow: High-quality 4D Content Generation using Multiview Token Flow
86. ICLR 2025PeriodWave: Multi-Period Flow Matching for High-Fidelity Waveform Generation
87. ICLR 2025Personalized Representation from Personalized Generation
88. ICLR 2025PianoMotion10M: Dataset and Benchmark for Hand Motion Generation in Piano Performance
89. ICLR 2025Presto! Distilling Steps and Layers for Accelerating Music Generation
90. ICLR 2025Provence: efficient and robust context pruning for retrieval-augmented generation
91. ICLR 2025Ready-to-React: Online Reaction Policy for Two-Character Interaction Generation
92. ICLR 2025RFWave: Multi-band Rectified Flow for Audio Waveform Reconstruction
93. ICLR 2025SCOPE: A Self-supervised Framework for Improving Faithfulness in Conditional Text Generation
94. ICLR 2025SiReRAG: Indexing Similar and Related Information for Multihop Reasoning
95. ICLR 2025SoundCTM: Unifying Score-based and Consistency Models for Full-band Text-to-Sound Generation
96. ICLR 2025Sports-Traj: A Unified Trajectory Generation Model for Multi-Agent Movement in Sports
97. ICLR 2025SSLAM: Enhancing Self-Supervised Models with Audio Mixtures for Polyphonic Soundscapes
98. ICLR 2025TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation
99. ICLR 2025Talking Turns: Benchmarking Audio Foundation Models on Turn-Taking Dynamics
100. ICLR 2025TANGO: Co-Speech Gesture Video Reenactment with Hierarchical Audio Motion Embedding and Diffusion Interpolation

常见问题

2023–2026 年四大会议一共有多少篇音频生成论文？

共 357 篇，按年份分别是 2023 年 32 篇、2024 年 73 篇、2025 年 116 篇、2026 年 136 篇。

NeurIPS 2026 有多少篇音频生成论文？

NeurIPS 2026 有 21 篇音频生成论文，上一届是 32 篇。

ICLR 2026 有多少篇音频生成论文？

ICLR 2026 有 64 篇音频生成论文，上一届是 43 篇。

ICML 2026 有多少篇音频生成论文？

ICML 2026 有 25 篇音频生成论文，上一届是 10 篇。

CVPR 2026 有多少篇音频生成论文？

CVPR 2026 有 26 篇音频生成论文，上一届是 31 篇。

哪个会议的音频生成论文最多？

ICLR，2023–2026 年共 135 篇。