← Back ◬ AI & Machine Learning Mar 30, 2026

Unlocking Strong Supervision: A Data-Centric Study of General-Purpose Audio Pre-Training Methods

arXiv AI Archived Mar 30, 2026 ✓ Full text saved

arXiv:2603.25767v1 Announce Type: cross Abstract: Current audio pre-training seeks to learn unified representations for broad audio understanding tasks, but it remains fragmented and is fundamentally bottlenecked by its reliance on weak, noisy, and scale-limited labels. Drawing lessons from vision's foundational pre-training blueprint, we argue that the audio field must first establish its own large-scale, strong supervision framework. We introduce a new data-centric pipeline that leverages a hi

Full text archived locally

✦ AI Summary · Claude Sonnet

Computer Science > Sound [Submitted on 26 Mar 2026] Unlocking Strong Supervision: A Data-Centric Study of General-Purpose Audio Pre-Training Methods Xuanru Zhou, Yiwen Shao, Wei-Cheng Tseng, Dong Yu Current audio pre-training seeks to learn unified representations for broad audio understanding tasks, but it remains fragmented and is fundamentally bottlenecked by its reliance on weak, noisy, and scale-limited labels. Drawing lessons from vision's foundational pre-training blueprint, we argue that the audio field must first establish its own large-scale, strong supervision framework. We introduce a new data-centric pipeline that leverages a high-fidelity captioner to create SOTA-quality captions and the first Unified Tag System (UTS) that bridges speech, music, and environmental sounds. We then conduct a systematic comparative study of different pre-training objectives on these strong source data. Our experiments suggest that data quality and coverage are the primary drivers of performance, while the choice of objective dictates downstream task specialization. Comments: Accepted to CVPR 2026 Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS) Cite as: arXiv:2603.25767 [cs.SD] (or arXiv:2603.25767v1 [cs.SD] for this version) https://doi.org/10.48550/arXiv.2603.25767 Focus to learn more Submission history From: Xuanru Zhou [view email] [v1] Thu, 26 Mar 2026 07:18:04 UTC (1,668 KB) Access Paper: HTML (experimental) view license Current browse context: cs.SD < prev | next > new | recent | 2026-03 Change to browse by: cs cs.AI eess eess.AS References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

💬 Team Notes