RSA-Bench: Benchmarking Audio Large Models in Real-World Acoustic Scenarios

Yibo Zhang, Liang Lin, Kaiwen Luo, Shilinlu Yan, Jin Wang, Yaoqi Guo, Yitian Chen, Yalan Qin, Zhenhong Zhou, Kun Wang, Li Sun · 2026

Read RSA-Bench: Benchmarking Audio Large Models in Real-World Acoustic Scenarios by Yibo Zhang, Liang Lin, Kaiwen Luo, Shilinlu Yan, Jin Wang, Yaoqi Guo, Yitian Chen, Yalan Qin, Zhenhong Zhou, Kun Wang, Li Sun (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-10384.txt.

Description

Yibo Zhang and co authors test audio models in true sound scenes. Large audio models have grown strong at speech and sound tasks. But lab tests use clean or fake noise. Real scenes layer many sounds at once. A room, a street, and a crowd each have their own mix.

The team call that mix the Acoustic Ecology of a scene. They built RSA-Bench from real scenes. It grades models on layered, true noise, not one fake source. A model that scores well in the lab may fail here. The set shows which models hold up in the wild.

Catalog record

CREATORYibo Zhang, Liang Lin, Kaiwen Luo, Shilinlu Yan, Jin Wang, Yaoqi Guo, Yitian Chen, Yalan Qin, Zhenhong Zhou, Kun Wang, Li Sun
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-10384

Browse the full library · About the Tower of Babel