语音识别论文趋势 2023–2026

主要结论

• 语音识别论文从 2023 年的 75 篇降到 2026 年的 50 篇，热度在回落。
• CVPR 是发语音识别论文最多的会议，2023–2026 年共 114 篇，占四个会议总量的 50%。
• 交叉方向里涨得最快的是应用 · speech and audio：同时属于两个主题的论文从 2023 年的 13 篇增至 2026 年的 15 篇。
• 在「语音与音频」的 5 个主题里，它的 2026 年论文数排第 5；同组增长最快的音频生成是 4.3 倍，语音识别是 0.7 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 2026CS-Dialogue: A 104-Hour Dataset of Spontaneous Mandarin-English Code-Switching Dialogues for Speech Recognition
• ICML 2026Shortcut-Resistant CAM Distillation for Long-Tailed Recognition
• ICLR 2026CTC-DRO: Robust Optimization for Reducing Language Disparities in Speech Recognition
• CVPR 2026CLEX Complementary Label Exchange Learning for Noisy Facial Expression Recognition
• NeurIPS 2025EuroSpeech: A Multilingual Speech Corpus
• ICML 2025OWLS: Scaling Laws for Multilingual Speech Recognition and Translation Models
• ICLR 2025Speech Robust Bench: A Robustness Benchmark For Speech Recognition
• CVPR 2025CryptoFace: End-to-End Encrypted Face Recognition
• NeurIPS 2024[Re] CUDA: Curriculum of Data Augmentation for Long‐tailed Recognition
• ICML 2024Learning Label Shift Correction for Test-Agnostic Long-Tailed Recognition
• ICLR 2024It's Never Too Late: Fusing Acoustic Information into Large Language Models for Automatic Speech Recognition
• CVPR 2024From Coarse to Fine-Grained Open-Set Recognition
• NeurIPS 2023Leave No Stone Unturned: Mine Extra Knowledge for Imbalanced Facial Expression Recognition
• ICML 2023Robust Speech Recognition via Large-Scale Weak Supervision
• ICLR 2023Continuous pseudo-labeling from the start
• CVPR 2023BioNet: A Biologically-Inspired Network for Face Recognition
• NeurIPS 2026PIGRAM: An Interpretable Patch--Motif Interaction Grammar for Protein--Nucleic-Acid Recognition
• ICML 2026Context-free Recognition with Transformers
• ICLR 2026Confident and Adaptive Generative Speech Recognition via Conformal Risk Control
• CVPR 2026DialogueVPR Towards Conversational Visual Place Recognition

全部语音识别论文

共 16 篇，第 1 / 1 页

1. NeurIPS 2026BrainWhisperer: Leveraging Whisper for Speech Decoding in Neuroprosthetics
2. NeurIPS 2026CS-Dialogue: A 104-Hour Dataset of Spontaneous Mandarin-English Code-Switching Dialogues for Speech Recognition
3. NeurIPS 2026Eliciting Zero-Shot Named Entity Recognition in Large Language Models via Instruction Semantic Elaboration
4. NeurIPS 2026Handwritten Text Recognition Lives in the High-Pixel Variance Subspace
5. NeurIPS 2026Is Text All You Need? Text as a Universal Information Bottleneck for Speech LLMs
6. NeurIPS 2026Isharah-Selfie: Continuous Sign Language Recognition Dataset for One-handed Signing
7. NeurIPS 2026OLA-Place: Cross-Modal Place Recognition without Global Descriptors
8. NeurIPS 2026PIGRAM: An Interpretable Patch--Motif Interaction Grammar for Protein--Nucleic-Acid Recognition
9. NeurIPS 2026PuppetGait: Generalizing Gait Recognition via 3D Body-Aligned LVM Features
10. NeurIPS 2026SFPR: Structural Fingerprinting for LiDAR-to-OpenStreetMap Place Recognition
11. NeurIPS 2026SMASH: Probing Speech Recognition Robustness via Semantically Targeted Bit Flips
12. NeurIPS 2026Speech Tokenizers are Vulnerable: Transferable Semantic Attack and Robust Tokenizer
13. NeurIPS 2026TrAction: Action Recognition with Sparse Trajectories
14. NeurIPS 2026VSRo-200: A Romanian Visual Speech Recognition Dataset for Studying Supervision and Multimodal Robustness
15. NeurIPS 2026WAXAL: A Large-Scale Multilingual African Language Speech Corpus
16. NeurIPS 2026When a Window Is Not an Action: Selective Phase-Script Deliberation for Sliding-Window Human Activity Recognition

常见问题

2023–2026 年四大会议一共有多少篇语音识别论文？

共 230 篇，按年份分别是 2023 年 75 篇、2024 年 55 篇、2025 年 50 篇、2026 年 50 篇。

NeurIPS 2026 有多少篇语音识别论文？

NeurIPS 2026 有 16 篇语音识别论文，上一届是 22 篇。

ICLR 2026 有多少篇语音识别论文？

ICLR 2026 有 7 篇语音识别论文，上一届是 7 篇。

ICML 2026 有多少篇语音识别论文？

ICML 2026 有 8 篇语音识别论文，上一届是 5 篇。

CVPR 2026 有多少篇语音识别论文？

CVPR 2026 有 19 篇语音识别论文，上一届是 16 篇。

哪个会议的语音识别论文最多？

CVPR，2023–2026 年共 114 篇。