---
license: cc-by-nc-4.0
task_categories:
- audio-to-audio
language:
- en
tags:
- empathy
- audio-generation
size_categories:
- 1K<n<10K

---

For **multi-context** data: Each entry has two contexts, one emotion. 


For **multi-emotion** data: Each entry has two emotions, one context. 


The better responses should be those named with *goodPara*.


If this dataset is helpful, please cite the paper: [AEQ-Bench: Measuring Empathy of Omni-Modal Large Models](https://arxiv.org/abs/2601.10513v1)