LAION Releases Web-Scale Audio-XL Dataset with 1.4B Samples

LAION has released LAION-Audio-XL, a web-scale audio corpus of about 1.4 billion samples drawn from academic datasets, web collections and Common Crawl. Its CLAP2 contrastive language-audio models top the MAEB-30 benchmark, with CLAP2-L scoring 53.6%.

2026-11-12 ~ 2026-11-12 · 2 related posts

• LAION releases 1.48B-sample audio corpus; CLAP2-L hits 53.6% on MAEB-30 — serrjoa · 2026-11-12
• LAION Releases Audio-XL: Web-Scale Corpus for Contrastive Language-Audio Pretraining — kastnerkyle · 2026-11-12