This release contains 1,000 multiple-choice spatial-audio benchmark items with first-order ambisonics audio. It corresponds to the benchmark for The World is Not Mono: Enabling Spatial Understanding in Large Audio-Language Models.
本发布包包含 1,000 条一阶 Ambisonics 空间音频多项选择 benchmark 题目，对应论文 The World is Not Mono: Enabling Spatial Understanding in Large Audio-Language Models 的评测集。
Reference:
Yuhuan You, Lai Wei, Xihong Wu, and Tianshu Qu. The World… See the full description on the dataset page:
https://huggingface.co/datasets/KonoyoBC/TWNM-FOA-Benchmark.
